Compare commits
41 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
| bcb9dc730a | |||
| 958f31cd7e | |||
| 0415e9d5ea | |||
| 41e5976897 | |||
| 5b51151f59 | |||
| 211117e06d | |||
| 79e9f15f89 | |||
| 5ca65a45fb | |||
| 97f132d3cb | |||
| c8c36c27ee | |||
| e132601b9f | |||
| ba755b0d42 | |||
| 52accafe6b | |||
| b8eca3b5a9 | |||
| 7152d654e3 | |||
| 0054ea2821 | |||
| 30ad156abf | |||
| 2f2b459c18 | |||
| 1e3097004e | |||
| 35a6d6d9b2 | |||
| e306f40cae | |||
| f7705d9634 | |||
| 44b6a09b06 | |||
| 3d649b3b69 | |||
| 70d1d8c7a3 | |||
| 1a9c51b15a | |||
| 4a03ec528a | |||
| 87ca075175 | |||
| d5207dc6cc | |||
| 50d799c6b4 | |||
| 880c88c7fc | |||
| 49dbc2f74c | |||
| 36c6be4ff8 | |||
| 0b4f1de439 | |||
| af42b949ef | |||
| a43262c112 | |||
| 470d694831 | |||
| 1a8975769b | |||
| ee97aaae73 | |||
| 9e1feeebb3 | |||
| 71f1805561 |
@@ -1,4 +1,6 @@
|
||||
* text=auto eol=lf
|
||||
tests/fixtures/ci-sources/sources/** -text -whitespace
|
||||
tests/fixtures/ci-replay/original-purge/upstream/** -text -whitespace
|
||||
# Official fixture captures keep their original bytes and vendor formatting.
|
||||
tests/fixtures/official-model-registry/sources/** -text -whitespace
|
||||
docs/embedded-browser-development-evidence/** -text whitespace=blank-at-eol,blank-at-eof,space-before-tab,cr-at-eol
|
||||
|
||||
@@ -0,0 +1,47 @@
|
||||
name: Official contracts and Windows source CI
|
||||
|
||||
on:
|
||||
push:
|
||||
branches: [feat/task-agreement-1.25, main]
|
||||
pull_request:
|
||||
workflow_dispatch:
|
||||
|
||||
jobs:
|
||||
windows-source:
|
||||
runs-on: awb-windows-x64
|
||||
permissions:
|
||||
contents: read
|
||||
actions: write
|
||||
steps:
|
||||
- name: Check out committed source
|
||||
uses: https://gitea.com/actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
|
||||
with:
|
||||
fetch-depth: 1
|
||||
persist-credentials: false
|
||||
- name: Pin Node and bundled npm
|
||||
uses: https://github.com/actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020
|
||||
with:
|
||||
node-version: 24.22.1
|
||||
check-latest: false
|
||||
- name: Pin the base QA Python
|
||||
uses: https://github.com/actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065
|
||||
with:
|
||||
python-version: 3.11.15
|
||||
check-latest: false
|
||||
- name: Fresh dependencies, contracts, build, tests and owned Windows smoke
|
||||
shell: pwsh
|
||||
run: |
|
||||
$ErrorActionPreference = 'Stop'
|
||||
$qaPython = (Get-Command python.exe -ErrorAction Stop).Source
|
||||
$qaNode = (Get-Command node.exe -ErrorAction Stop).Source
|
||||
& $qaNode scripts/ci.mjs --profile windows --python $qaPython --output-root .qa/ci
|
||||
if ($LASTEXITCODE -ne 0) { exit $LASTEXITCODE }
|
||||
- name: Preserve original reports and failures
|
||||
if: always()
|
||||
uses: https://gitea.com/actions/upload-artifact@ff15f0306b3f739f7b6fd43fb5d26cd321bd4de5
|
||||
with:
|
||||
name: windows-source-ci
|
||||
path: .qa/ci/**/evidence/**
|
||||
if-no-files-found: error
|
||||
include-hidden-files: true
|
||||
retention-days: 14
|
||||
@@ -1,5 +1,43 @@
|
||||
# Agent Workbench项目执行规范
|
||||
|
||||
2026-11-12 1.25.14修复与验收958f31cd及前四条已普通增量推送,远端HEAD/重新fetch和五条逐SHA一致,见[回执](docs/gitatlas-credential-persistence-1.25.14-push.json)。回执与完成接续最后一起提交/推送并核最终HEAD。正式1.25.14/PID29884,七条Key实测重启恢复,原数据/选择/密文/420请求保持;715成员封存不追加,推送证据另根。下一M8另冻,历史异常与未验不改签;下方收尾中为原时点。
|
||||
|
||||
2026-11-12 重启丢Key已正式修复为1.25.14:源码5b51151f/41e59768,程序/源码固定41e59768,验收驱动修正0415e9d5单独保留;54相关、1000输入/30编译独立重建、程序/源码全成员、最终Edge/实际Electron10步5图与实际隔离Inno四组通过。唯一正式Setup234945ms先保留PID14388/Provider,再实际退出并从原物理数据目录启动新PID29884/instance81cade0e-6306-4035-a128-15e2cd243175;七条Key自动恢复、原加密副本/登记193/选择/全部原行和139导出及5583工件两阶段exact,109项目/9290记录/24211事件/420请求保持、空闲、9229关闭。入口4318;正常新填/目录/确认导入默认Windows用户加密保存,“保存当前Key”仅在进程内明确保存且不改变原对象/修订。见[计划](docs/CREDENTIAL-PERSISTENCE-1.25.14-PLAN.md)和[验收](docs/QA-CREDENTIAL-PERSISTENCE-1.25.14.md)。715成员178252120字节封存逐SHA通过,首失败/基线/原生启动参数和依赖链接封存拒绝保持。当前限定收尾提交/既有GitAtlas普通增量推送;真实Key未读取/复制/重加密,真实供应商请求0,不委派或续旧样本/Purge。M4/Claude403/unknown和全部旧未验边界保持,下一M8另冻。下方旧只读/不重存说明不覆盖本轮明确修复授权。
|
||||
|
||||
2026-11-12 新协调来源只读登记在[同Goal工件连接记录](docs/integration/MNEMO-GOAL-WORKER-READONLY-2026-11-12.md):Mnemo指定旧AG42aacbd1/b330b966有限连接/显式complete已过,113=19+94、独立69及31隔离HTTP仅按来源回执范围,原9/build/57未重跑,本增量模型0;4321原观察404、未载入,生产者/完整指纹/派发接收/父交付/语义未验。AD与M4三工程分别登记,旧AG异常/Purge未验和Claude现有两403/unknown及541/41封存不改签。本消息不请求新执行或源线程回复;本次不读Key、不查正式服务、不重跑/重派或新增Claude/Purge,仅本地登记来源,后续开发阶段再按既有授权增量推送。
|
||||
|
||||
2026-11-12 本轮提交推送完成:修后Claude驱动97f132d3/验收5ca65a45已普通增量推送并与远端/fetch逐SHA一致,见[回执](docs/gitatlas-claude-proxy-correction-push.json);完成回执与接续最后另提交核最终HEAD。原M4收尾c8c36c27也已推送,两根541/41封存保持;当前Claude上游403客户端分组权限仍未通过,不追加本轮两请求。下一M8新原版Purge独立冻结,历史异常/unknown/真实预留/Mnemo语义/手机/长期边界保持。
|
||||
|
||||
2026-11-12 修后Claude独立两预检结束:目录完整14型号/选定在内;两新请求均上游403 permission_error“分组仅允许Claude Code官方客户端”,无响应型号/SSE/usage,用量/费用null、工具执行0。见[修后验收](docs/QA-CLAUDE-PROXY-CORRECTION-2026-11-12.md)。原58298行/137导出/418请求/5576工件及当前七连接/池/选择/配置/凭据行SHA一次exact,41成员36284086字节新封存SHA9b8a1cfc通过;原M4的40请求/541清单保持。不追加或重派两请求,不回退人类更正或冒充Claude Code,供应商须另开放标准Messages客户端。当前验收提交/普通推送收尾,产品未改不重发行/重启;M8原版Purge另冻,全部历史unknown保留。
|
||||
|
||||
2026-11-12 M4收尾提交e132601b和此前六条已普通增量推送,远端HEAD/重新fetch七条逐SHA一致,见[推送回执](docs/gitatlas-m4-live-push.json)。回执与完成接续最后另提交并核远端;原40请求/541封存保持,1.25.13未改。下一仅按人类Claude代理Key更正另冻至多两新预检,旧失败和Purge未验保留;下方“推送收尾中”为原时点。
|
||||
|
||||
2026-11-12 M4原批次完成最终审计和封存:40真实请求/629106报告tokens/0unknown,Node、React、Python原检查及限定Git全部通过;541成员/134884464字节清单SHA558c9b54逐成员复核通过,见[最终验收](docs/QA-M4-LIVE-BASELINE.md)。NRO七份公开来源另根只读登记,不代签Mnemo语义或原AD/AG。当前只剩限定收尾提交及既有GitAtlas普通增量推送;1.25.13产品未变,不重发行/安装/重启。封存后人类反馈Claude代理Key已更正,按新公开修订另冻文本/工具至多两请求,不回退人类修改、不改旧两mismatch和目录缺项,不续旧任务或重跑三工程/Purge;凭据不读/复制/重存,工具只观察。Purge另冻、真实预留边界、生产语义/物理手机/长期和普遍收益unknown保持。下方阶段“下一封存”均为原时点。
|
||||
|
||||
2026-11-11 M4三工程均完成:新Python十请求176083tokens/0unknown、原10pytest/compileall/三源码Gitfa2ce16a/终端原证据与最终verified_committed通过,原owner全部index保持,见[Python验收](docs/QA-M4-PYTHON.md)。本批三线六预检加三工程40请求629106tokens/0unknown;Claude两身份mismatch及目录缺项仍失败,外部你好一请求4301tokens另列。python-fc99c576一次原53527行/127导出/5244工件/377旧请求及本批九捕获导出exact,config124→130/registry176→182等明确增量核验;六连接/池/原选择/PID63892与instance保持、空闲和9229关闭。下一最终封存/回顾/普通增量GitAtlas推送,程序未变不重发行1.25.13;Purge另冻可比实验,历史异常/Mnemo生产语义/物理手机/长期/真实预留/普遍收益unknown保持。不读Key、不委派、不续旧样本,下方下一Python为React原时点。
|
||||
|
||||
2026-11-11 M4新React阶段完成:同一正式1.25.13/63892与DeepSeek官方连接,十四请求271333tokens/0unknown、原9/9/typecheck/build、实际Edge桌面和390px/两图实看、四源码Gitc55644a3及owner全部index通过。终端Git原证据verified、最终报告verified_committed,未补造delivery_job或进程认证。原53527行/127导出/5244工件/377旧请求保持,累计本批30请求/453023tokens,外部你好一请求仍单列;config124→128/registry176→180与前插/诊断/序列精确复算。见[React验收](docs/QA-M4-REACT.md)、[机器记录](docs/m4-react-2026-11-11.json)。下一独立Python原10pytest/编译,之后最终封存、普通增量推送;Claude身份/历史异常/Mnemo语义/物理手机/长期/真实预留/Purge收益未验。未读Key、重启、改产品或委派,下方Node阶段“下一React”为原时点。
|
||||
|
||||
2026-11-11 M4真实兼容与新Node阶段完成:正式1.25.13/63892未重启。三线六次真实预检/1227报告tokens/0unknown,DeepSeek官方Chat和GPT代理Responses完整matched;Claude Messages代理两次自报gpt-6.1-sol且目录未列claude-opus-5-5,必填max_tokens=128000已发送但真实Claude身份未验。DeepSeek新Node自主十请求180463tokens/0unknown、原9/9/build/三源码限定Git3cd994ff、owner全部index保持;内置delivery和最终报告已completed,七不同文件读取/零同字节范围重复、122112cacheRead仅算观察。原53527行/127导出/5244工件和377旧请求exact,config124→126/registry176→178及前插/诊断/序列明确重算;外部“你好”一新会话/一请求单列,两个审计失败保持,见[阶段验收](docs/QA-M4-COMPATIBILITY-NODE.md)和[机器记录](docs/m4-compatibility-node-2026-11-11.json)。下一独立React,完整成功后Python,再另冻Purge;旧异常/Mnemo生产与语义/物理手机/长期/真实预留边界/Purge收益unknown保持。不读真实Key、不委派、不续旧样本;阶段限定提交与既有普通增量GitAtlas推送授权有效。下方M4的0真实请求为阶段2原时点。
|
||||
|
||||
2026-11-11 当前进入M4:正式1.25.13/63892已重新公开核对为空闲、unrestricted/passthrough/full。用户在模型中心恢复DeepSeek官方、Claude Messages代理及GPT Responses代理,原无凭据观察保留;新M4计划和驱动在docs/M4-LIVE-BASELINE-PLAN.md。最后隔离驱动11边界及固定Node5请求、原9/9/build/限定Git与owner全部index、原字节下载和恢复通过,前两断言失败/第三原通过保持,见docs/m4-driver-validation-2026-11-11.json。当前0真实请求,先限定提交后另冻真实兼容/工程,不改1.25.13或续旧样本。AD现成任务只读来源在docs/integration/WORKBENCH-AD-NODE-READONLY-SOURCE-2026-11-11.json;46允许文件/原报告in_progress、终端Git、两个原日志描述符缺项及非原生v2保持,不向源线程回信。不读真实Key、不委派,既有普通增量GitAtlas推送授权保持。
|
||||
|
||||
2026-11-11 1.25.13 已正式交付:产品 e306f40c、发行来源35a6d6d9;1399输入/30编译独立clean npm ci重建exact,186相关/4打包、实际Inno11/11、F跨版/兼容恢复7/7、Mnemo合成后端24/24与安装副本网页/Electron各4组/10图、基础文件9/9通过。唯一正式Setup351789ms成功,保留原PID63892/instance/Provider与Key存在状态;102项目/127会话/8741记录/22085事件及outbox/5244工件、374旧请求精确迁移,原字段/全部原字节保持;仅新增公开规格/契约/账本字段与两项MCP导出字段逐项核验。正式两端1280/390四图实看、final复核和9229关闭通过,新版桌面已正常恢复。见[验收](docs/QA-MNEMO-RELEASE-1.25.13.md)、[机器身份](docs/mnemo-1.25.13-release-identity.json)。782份811582693字节选择性证据清单611c4c40封存,首失败/旧封存保持。源码/打包修复/验收1e309700已普通增量推送并与远端及fetch逐SHA一致,见[回执](docs/gitatlas-mnemo-1.25.13-push.json);完成接续与回执另提交并再次核对;下方“尚未发行”按旧时点理解。Mnemo入口工具箱→项目记忆→Mnemo项目记忆桥,默认关闭;本次凭据只存宿主内存、重启需重新授权,真实生产绑定/语义收益/物理iPhone/Purge收益未验。下一M4另冻新真实兼容/工程基线,不续旧任务、不读真实Key,不因本次合成通过改写历史异常或收益unknown。
|
||||
|
||||
2026-11-11 用户要求立即开发、提交、验收,当前按 [1.25.13 交付计划](docs/M3-B4B-RELEASE-1.25.13.md) 完成 M3+B4b 正式源码/程序/安装器和实际升级。该授权覆盖下方历史阶段性“不提交/不发行/不查 health”安排,保留旧封存、真实模型/网络/unknown/重复读取/Purge 未验证结论。沿用既有 GitAtlas 普通增量推送授权;不委派、不续跑旧任务、不读取或重存真实 Key。每阶段实际验证后限定提交,源码包不含项目数据,安装另核原数据保持。
|
||||
|
||||
B4b默认关闭的Mnemo单项目记忆桥源码及隔离工程验收完成:最后相关按仓库固定并发2为104/104,916输入/30编译独立重建exact,后端24/24及N25网页/隐藏Electron/伴随HTTPS窄屏通过;两端各4组/10图、最终8图实看。37账本尝试/31物理原文及336原工件另有Mnemo独立核对,原回执all25CasesVerified:false保持其后端范围。全部原失败(含两轮默认并发103/104)保留,未改超时阈值。见[验收](docs/QA-B4B-MNEMO-NATIVE.md)、[机器矩阵](docs/b4b-mnemo-native-2026-11-08.json)、[使用说明](docs/MNEMO-PROJECT-MEMORY-BRIDGE.md)。
|
||||
|
||||
当前仍f7705d96基础上的M3+B4b未提交工作树,本轮不提交/推送/发布/安装或正式重启,不查正式health、不读真实Key/请求真实模型、不委派/续旧工程/向源线程回信,不改Mnemo源码。自有合成宿主均正常关闭,最后59752/private/所属Node归零,四候选PID/端口监听及所属Electron0;M3旧封存清单78ce1df7保持。下一M3最终产品commit/版本/源码程序包/InnoSetup及独立安装,再另冻正式升级/新真实样本。历史W/AD/AE/AG成功/网络/unknown、真实语义收益/重复读/R-W边界/Purge/长期/物理iPhone未验保持,开放透传和持久ask/auto/full不变。下方M3时点的“B4b全pending/不强行切换”已由后续明确授权及本次原生证据更新,其他历史按原时点理解。
|
||||
|
||||
M3迁移/兼容恢复源码与隔离验收已完成,统一发行未做:当前仍是f7705d96上的未提交工作树,1117输入/30编译经独立clean npm ci及重建逐字节一致,最后相关82/82、实际F跨版7/7、网页/隐藏Electron各5组通过;旧原生390帧不采纳,0新请求补图两视口实看通过。首索引回退失败、HOST_PROCESS_UNVERIFIED/迟到原因unknown及所有构建/QA失败原件保持,自有失败宿主已证明归属后正常停止。见[验收](docs/QA-M3-MIGRATION-RECOVERY.md)和[机器记录](docs/m3-migration-recovery-2026-11-08.json)。本轮不自动提交/推送/发布,未查询正式health、升级/重启、读取真实Key或请求真实模型,不委派/续旧工程。M3最终commit/编号/源码与程序包/InnoSetup/正式升级仍待;不强行切B4b,Mnemo v2八输入已登记但25原生案例全pending,旧账本SHA不补签当前代码。开放默认/持久权限及历史真实交付/网络/unknown、重复读/R-W真机边界/Purge收益unknown保持。下方“下一M3”按旧时点理解。
|
||||
|
||||
I27/I28推送完成:源码与验收44b6a09b已普通增量推送至既有GitAtlas分支,远端HEAD和重新fetch的14个提交逐SHA一致,见[回执](docs/gitatlas-m2-ci-push.json)。回执/完成接续最后另提交并再次核对;最终4a03ec52干净1222/1222及7595封存原件保持,正式未部署/重启,runner/job/真实供应商及历史收益unknown不代签。下一按docs/M3-RELEASE-PLAN.md推进实际schema2迁移与新结果后的兼容回退;下方进行中说明按原时点理解。
|
||||
|
||||
M2第六批I27/I28本地CI与官方契约完成:来源a43262c1、最终程序/测试4a03ec52(含87ca0751工具广告对齐);真正干净npm/Python QA/Electron安装、146文件1222/1222,实际Edge/隐藏Electron7组4图全部实看,1376导出输入/29编译前后SHA一致。首1194/1211、官方下载主动结束与第二轮原生画面滞后原件保持。新7595份150586505字节封存,见[验收](docs/QA-M2-CI-OFFICIAL-CONTRACTS.md)与[机器记录](docs/m2-ci-2026-11-07.json);旧五根21297份7184046930字节只读exact。本地实现/验收已完成,文档与普通GitAtlas增量推送收尾;runner/job/上传仍unknown,不将本机通过称远端通过。下一[M3](docs/M3-RELEASE-PLAN.md)精确schema2迁移及产生新结果后的兼容回退,再统一发行,不能删账本/降schema或去掉原行工件保持核对;2001操作回归872353ms只证明记录保留,M7规模性能待办保持。0真实模型/Key/正式写入,不委派/发行/正式重启或续旧工程,不作新正式health声明;默认开放/透传/持久权限、W/AD/AE/AG网络/unknown、重复读/R-W真机边界/Purge收益unknown保持。Mnemo/国内入口仅固定原件只读登记,未接通服务/改默认或回复源线程。下方旧接续按各自时点理解。
|
||||
|
||||
M2第五批I09/I11源码与隔离验收完成:来源71f18055、核心9e1feeeb、界面ee97aaae。原生缓存块/工具元数据与状态身份绑定、会话/Task/Run/关闭保持;纯Chat混池误带首成员reasoning实际缺陷已修。最后专项16/16,首候选127/129与两改文件26/26按178后端/213测试输入SHA复用覆盖,不能称最后单轮全套全绿。最终880输入/29编译/74runtime、网页/实际Electron各8组/10图/三配色/1280及390、原metadata/cipher/存储快照/历史exact,20图SHA/8图实看、自有候选进程和记录端口监听0。5126份2008555583字节已封存,见[验收](docs/QA-M2-NATIVE-CACHE-STATE.md)与[机器记录](docs/m2-native-cache-state-2026-11-07.json);源码与验收1a897576已普通增量推送、远端逐SHA一致,见[回执](docs/gitatlas-m2-native-cache-state-push.json);完成接续最后另提交核对。下一I27/I28 CI,再M3实际schema2迁移增量及新处理器结果后回退,不能删除新账本或去掉逐表全等守卫。本批0真实模型/Key/正式写入,不发行/正式重启/委派或续旧工程,不作新正式health声明;默认开放/透传/持久权限及W/AD/AE/AG网络/unknown、重复读/R-W边界/Purge收益unknown保持。下方旧接续按原时点理解。
|
||||
|
||||
M2第四批I14/I12源码完成:来源337d1bc9、核心f95507ab、界面26472d28;新/旧版本与认证分开、Beta/头/同源路径、冻结契约及非提示指纹,实际fetch设置与原Run推导不混同。最后相关38/38、首候选143/143后端输入未变,最终871输入/29编译/74runtime、网页和实际Electron各8组/8图/三配色/1280及390,原metadata/cipher/存储快照exact,原I13/I15封存只读一致、自有进程/已知监听0。见[验收](docs/QA-M2-REQUEST-CONTRACT.md)与[计划](docs/M2-REQUEST-CONTRACT-PLAN.md)。3207份1498866416字节已封存,源码与验收04c0556c已普通增量推送至既有GitAtlas分支、远端逐SHA一致,见[回执](docs/gitatlas-m2-request-contract-push.json);完成接续最后提交再核对。下一I09/I11缓存块/亲和/状态及CI,再M3真实schema2迁移增量和新处理器结果后的回退。0真实模型/Key/正式写入,不发行/重启/委派或续旧工程,不作新正式health声明,开放默认/持久权限与W/AD/AE/AG、重复读、R/W边界、Purge收益unknown保持。下方旧进行中说明按原时点理解。
|
||||
|
||||
M2 I15推送完成:0cbf14a2/ba4b4491/b5aaf1f7与验收bf07e66d已普通增量推送至既有GitAtlas feat/task-agreement-1.25,远端逐SHA一致;[回执](docs/gitatlas-m2-official-registry-push.json)与完成接续最后提交再核对。5688原件/858输入/29编译保持,正式未部署/重启。下一I14实际版本/头/指纹,随后缓存块/亲和/状态及CI,M3需真实迁移增量/激活后回退;历史unknown保持。下方进行中说明按原时点理解。
|
||||
|
||||
@@ -1,5 +1,18 @@
|
||||
# Agent Workbench 更新记录
|
||||
|
||||
## 1.25.14
|
||||
|
||||
- 修复正常配置的模型 Key 在重启后丢失:手动填写、上游目录选择和确认导入默认使用 Windows 本机加密保存,重启自动恢复;可明确选择仅本次使用。
|
||||
- 普通编辑连接保留已保存的凭据,不要求重复填 Key;保存失败保留输入,清除或替换为临时 Key 后不恢复旧副本,自动同步保持来源已选的保存方式。
|
||||
- 模型中心直接显示本机保存、重启恢复与加载失败状态;目录新增型号共用源连接的保存凭据,旧配置接口的连接与 Key 共同提交。
|
||||
|
||||
## 1.25.13
|
||||
|
||||
- 新增可选 Mnemo 单项目记忆桥:连接项目绑定、核对完整来源、确认共享、召回及逐请求来源回执;来源变化或桥服务异常时省略远端记忆,继续原任务。
|
||||
- 请求账本支持完整分页、原始请求和缓存/推理用量核对;模型能力、原生参数、版本和缓存状态在实际请求中保持一致,未知值如实显示。
|
||||
- 支持精确 schema2 迁移和保留新结果的兼容恢复,安装不删除原账本、不降低数据版本;程序包与用户项目数据分开。
|
||||
- 保持默认开放透传、完整历史和已保存的执行权限。记忆桥默认关闭,项目凭据仅存当前宿主内存;重启后需在界面重新授权。
|
||||
|
||||
## 1.25.12
|
||||
|
||||
- 协作子任务分别选择职责、交付要求和模型连接/池;规划时绑定原生参数与修订,切换默认模型不再误阻止原任务续接。
|
||||
|
||||
@@ -7605,6 +7605,28 @@ ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR
|
||||
IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
|
||||
|
||||
|
||||
========================================================================
|
||||
yaml 2.8.1
|
||||
License: ISC
|
||||
Source package: node_modules/yaml
|
||||
|
||||
LICENSE
|
||||
|
||||
Copyright Eemeli Aro <eemeli@gmail.com>
|
||||
|
||||
Permission to use, copy, modify, and/or distribute this software for any purpose
|
||||
with or without fee is hereby granted, provided that the above copyright notice
|
||||
and this permission notice appear in all copies.
|
||||
|
||||
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH
|
||||
REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
|
||||
FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT,
|
||||
INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS
|
||||
OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
|
||||
TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF
|
||||
THIS SOFTWARE.
|
||||
|
||||
|
||||
========================================================================
|
||||
yargs 15.4.1
|
||||
License: MIT
|
||||
|
||||
@@ -0,0 +1,15 @@
|
||||
# CI 官方来源与执行范围
|
||||
|
||||
2026-11-07,承接 [M2 CI 计划](M2-CI-PLAN.md)。GitAtlas 公开只读 API 实返 Gitea **1.26.2**、当前仓库 `has_actions=true`;运行/runner 接口匿名均 HTTP 401,没有据此确认可用 runner。仓库 API 当前 `private=false` 作为观察保留,未修改其配置。
|
||||
|
||||
采用同一 1.26 文档系列(页面当前标注 1.26.4)的官方原文,逐份响应/原字节/SHA 保存于 `tests/fixtures/ci-sources/SOURCE.json`。它们解释工作流契约,不是本仓库远端执行回执:
|
||||
|
||||
- [Quick Start](https://docs.gitea.com/1.26/usage/actions/quickstart/):工作流位于 `.gitea/workflows/`,实例/仓库 Actions 与已注册 runner 均为运行前置条件。
|
||||
- [Act Runner](https://docs.gitea.com/1.26/usage/actions/act-runner/):`runs-on` 匹配标签;Windows 使用明确的 `:host` 标签。默认容器的 Node 16 不满足本产品 Node 24 的构建身份。
|
||||
- [与 GitHub Actions 的差异](https://docs.gitea.com/1.26/usage/actions/comparison/):支持完整 action URL;不能依赖尚不支持的 workflow `timeout-minutes`/复杂 `runs-on` 等语法来判定执行结果。
|
||||
|
||||
原无尾斜杠入口的 308、旧 runner 路径的 301、所有匿名 401 及本次公开 API 原响应保留在 `.qa/m2-ci-2026-11-07/`,未跟随到带凭据的地址。无账户 token、真实模型 Key 或模型请求。
|
||||
|
||||
官方模型回归复用 I15 注册表、I14 请求版本/头和 I09/I11 原生缓存/状态的固定原件;新 CI 检查将其来源、规格版本、字段和弃用变化关联到实际回归输入。夹具及受控本机 HTTP 结果不代表官方服务、代理或真实模型自主交付已通过。
|
||||
|
||||
可移植契约检查与 Windows 全源码/实际桌面/隔离安装检查分别报告,环境缺失必须失败,不以跳过改绿。M3 真实跨版/schema2 迁移及新处理器结果后的回退仍是独立验收,不在 CI 文档中冒称正式安装完成。
|
||||
+35
@@ -0,0 +1,35 @@
|
||||
# 可重复的源码与官方契约 CI
|
||||
|
||||
当前入口从已提交 Git HEAD 导出新的源码 ZIP,逐源文件核 SHA,展开到唯一目录后实际 `npm ci`。不复用工作区 `node_modules` 或 junction,不读取用户项目、运行服务、模型 Key/配置或旧实验。可审查源码改动应先提交;文档/其他非编译改动会在报告记录,不参与程序构建。
|
||||
|
||||
环境固定于 `scripts/ci-environment.json`:Node 24.22.1、随包 npm 11.19.1;Windows 全套另需 Python 3.11.15。pytest 8.4.2、pip 与全部框架依赖通过 `scripts/ci-python-requirements.txt` 精确 wheel SHA 安装到本次新 venv。Playwright 1.64.1、Electron 45.2.0 和其余 npm 输入由 lock 固定。缺环境、版本不同、依赖安装或实际测试失败均非零退出。
|
||||
|
||||
```powershell
|
||||
& 'D:/Ai-project/REA逆向/tools/node-v24.22.1-win-x64/node.exe' scripts/ci.mjs --profile contract
|
||||
& 'D:/Ai-project/REA逆向/tools/node-v24.22.1-win-x64/node.exe' scripts/ci.mjs --profile windows --python '绝对路径/python.exe'
|
||||
```
|
||||
|
||||
其他开发机使用自身已核对的相同版本 Node 路径。npm 入口为 `npm run ci:contract` / `npm run ci:windows -- --python ...`。默认输出 `.qa/ci/run-.../evidence/report.json`,可用 `--output-root` 指定;每次新建目录,旧失败/日志不覆盖。数据/依赖/缓存位于同 run 的独立目录,不作为发布源码或 CI 证据上传。
|
||||
|
||||
默认 Electron 从官方 GitHub 发行地址下载。环境访问异常时可显式加 `--electron-mirror https://npmmirror.com/mirrors/electron/`,报告记录该公共镜像和官方锁定 `electron/checksums.json` 的 SHA;不继承电脑的镜像/代理/凭据配置,也不关闭校验。原官方失败仍单独保留。Python 命令明确 UTF-8,中文路径的原日志可直接核对。
|
||||
|
||||
| 范围 | 实际执行与判定 |
|
||||
| --- | --- |
|
||||
| contract | 四来源清单原字节、注册表及请求默认版本/字段/弃用快照、14指定行为回归和 CI/源码包专项;完整应用构建。可在支持 Node 的平台执行,当前其他平台实测状态另列。 |
|
||||
| windows | 上述来源/构建与全部 `.test.mjs`,实际产品 Edge 必需;固定 Playwright 使用实际 Edge 及隐藏 Electron 执行 1280/390 冒烟。安装/回退相关回归使用自有隔离 fixture,不接触正式服务。 |
|
||||
|
||||
Node TAP 的失败、取消、跳过、todo、缺完整统计或零测试均不能判定通过,不能通过删 Python/浏览器测试掩盖环境缺项。每条命令保存所属 PID、完整 argv、真实 exit/signal、原 stdout/stderr 字节及 SHA;源码和全部编译工件在结束后再次核对。网页/Electron 原截图、DOM、错误与所属关闭状态另存。
|
||||
|
||||
三个历史回归套件使用 [最小回放夹具](../tests/fixtures/ci-replay/README.md),不在独立源码包中寻找包外历史档案。CI 导出后先逐 SHA 核其 65 份输入,原来源和解析字段均在 INDEX 中记录;JSON子树与60份公开锁定Purge源码分别标明,原完整档案仍单独保留。缺失或变化直接失败,原失败/请求/回执的断言不删除,也不将回放当真实模型或收益验收。
|
||||
|
||||
GitAtlas 工作流为 `.gitea/workflows/ci.yml`,push / pull_request / 手工触发。Gitea 1.26.2 与仓库 Actions 已公开核实;运行/runner API 匿名 401,**可用执行器及远端运行尚未确认**。工作流要求独立 Windows x64 runner 标签 `awb-windows-x64:host`、PowerShell、Git、Edge及 action 所需 Node 16/20 执行能力,随后 action 安装固定 Node/Python。不会在正式电脑自动安装持久 runner。Action 均固定完整 commit,Checkout 同 commit 使用 Gitea 官方镜像;原 GitHub 网络失败完整保留。证据上传使用 Gitea actions v3 固定契约,明确包含 `.qa` 的隐藏证据路径,仅上传 evidence 白名单;实际上传与远端执行需要真实 job 回执。
|
||||
|
||||
官方更新先保留原响应/失败/SHA,修改注册表和对应行为回归,再生成新快照、审查差异并追加 `tests/fixtures/official-contracts/INDEX.json` 的 history;旧快照保留。`--freeze` 只冻结,不能当作通过。命令:
|
||||
|
||||
```text
|
||||
node scripts/verify-official-contracts.mjs --freeze NEW_SNAPSHOT.json
|
||||
node scripts/verify-official-contracts.mjs --diff OLD_SNAPSHOT.json NEW_SNAPSHOT.json --output NEW_DIFF.json
|
||||
node scripts/verify-official-contracts.mjs --output NEW_VERIFICATION.json
|
||||
```
|
||||
|
||||
同型号的协议/API版本分别成键,字段来源、缓存/状态、弃用/别名、认证及新旧默认的差异会出现在报告;实际事实或请求默认变更不增加相应版本会失败。所有检查离线使用已固定官方原件,本机 HTTP/源码结果不等同真实官方/代理兼容、模型自主工程、缓存或 Purge 收益。M3 实际 schema2 迁移增量和新处理器产出后的回退仍另验,正式发行尚未发生。
|
||||
@@ -0,0 +1,24 @@
|
||||
# 1.25.14 凭据重启恢复修复
|
||||
|
||||
本轮开发与正式验收已完成:最终源码/程序`41e59768`、验收修正`0415e9d5`,54相关、独立构建/档案、实际网页/Electron与隔离Inno四组通过。正式Setup先保留PID14388,再实际重启为PID29884/1.25.14;原七条Key自动恢复、加密副本/选择/登记与全部原行/导出/工件保持,请求420未重放。715成员证据封存SHA`031b0da2…`,原失败和历史未验证事项保持;当前限定收尾提交与普通GitAtlas增量推送。下文的初始无Key及阶段安排按原时点保留。
|
||||
|
||||
2026-11-12,来源 `211117e0` / 正式安装版 1.25.13。用户反馈正常填写的 API Key 在重启后丢失。当前七条连接公开 `hasKey=false`;109 项目、139 全部会话、9290 记录、24211 事件、420 请求,原数据目录的 SQLite `quick_check=ok`。
|
||||
|
||||
根因:已有 Windows CurrentUser DPAPI 和自动恢复,但连接表单、目录选择及确认导入默认只保存宿主内存,持久保存必须额外勾选。仅改变提示不能修复此使用路径。
|
||||
|
||||
本轮范围:
|
||||
|
||||
- Windows 上新填和确认导入的 Key 默认加密保存,显式 `persistKey=false` / `persistCredentials=false` 保留临时方式;自动同步保持来源已选的保存方式。
|
||||
- 密文和连接配置共同提交;加密或数据库失败不显示成功、不留下部分连接或新内存凭据。普通编辑不重复要求 Key,不重新解密保存已有凭据。
|
||||
- 显式清除或替换为临时 Key 不允许下次启动恢复旧副本。源连接及目录新增型号共用同一保存来源。
|
||||
- 修复三个入口的默认选择、字段名称、无新 Key 的编辑及准确成功状态。旧配置接口的新 Key 也按相同策略原子保存。
|
||||
- 先保留当前源码的复现失败,再执行针对性回归、构建、网页/桌面与 390px 核对,使用合成 Key 运行安装包的真实 Windows DPAPI、退出/新进程恢复和本机兼容服务鉴权。
|
||||
- 按阶段审查并限定提交;发行 1.25.14 源码、程序及安装器,隔离验证后更新正式服务,按既有 GitAtlas 授权增量推送。
|
||||
|
||||
工具不读取、复制或回传真实 Key,不向聊天或日志输出凭据。已经退出且没有加密副本的旧内存 Key 无法恢复。使用原正式数据目录,历史原文、失败、unknown、重复读取及 Purge 未验证结论保持;本轮不请求真实供应商、续旧任务或推进新的 Purge 实验。
|
||||
|
||||
首源码提交/首候选包后公开核对发现用户已重新补填七条 Key,原 PID14388 保持。按本轮修复授权增加“保存当前 Key”:工作台在自身进程内加密当前凭据,调用方只传连接 ID/修订并收到状态,原 Key 对象和凭据修订不改变。先用合成凭据验证,再由新处理器在保留原宿主的升级后保存当前七条连接,全部成功才正式重启,不要求用户再次填写。首候选包保留,新增能力需另冻最终候选并复验。
|
||||
|
||||
后续 `18-formal-status.json` 公开复核:七条连接都已有有效 DPAPI 副本,连接登记193、请求仍420、原PID/instance保持且空闲。因此正式升级和重启保留原加密副本,不再次保存或加密;“保存当前 Key”通过合成旧版本保留进程升级及新进程恢复验证。正式操作前再次冻结最新公开状态,任何未保存/来源失配不得先停止宿主。
|
||||
|
||||
关联升级回归发现已有虚拟数据目录案例在读取宿主索引前未采用物理根目录。`211117e0` 独立导出基线复现原两项失败。此次复用同一无符号链接校验/物理根目录解析,先按物理根读取索引,原 Store 的采用仍在可回退阶段中完成,不重开已有 SQLite、不改变原行或 Provider;保留首失败和基线记录。首次 UI 验收的配置格式选择器过于严格,四个已过步骤及失败原件保留,修正验收选择器后另根复验。
|
||||
@@ -1,5 +1,23 @@
|
||||
# 官方优先开发路线(2026-11-06)
|
||||
|
||||
2026-11-12 插入完成重启丢Key缺陷修复1.25.14:默认Windows用户加密/原子保存与当前Key保存,正式新PID29884恢复七条Key,原109项目/139会话/5583工件与全部原行/导出/选择/420请求保持。54相关、独立源码/程序、实际两端与隔离Inno四组通过,715成员封存;见[验收](QA-CREDENTIAL-PERSISTENCE-1.25.14.md)。当前收尾提交/既有普通增量推送,随后M8另冻;Claude两403/unknown、Mnemo生产语义、手机/长期/预留/重复读与Purge收益仍按原范围未验。
|
||||
|
||||
2026-11-12 本轮M4收尾c8c36c27与修后Claude5ca65a45均已普通增量推送且远端/fetch一致;[修后回执](gitatlas-claude-proxy-correction-push.json)和完成接续另提交核最终HEAD。541/41封存保持,Claude403兼容仍失败,下一M8原版Purge另冻,不重派本轮或代签M5–M7/真实收益。
|
||||
|
||||
2026-11-12 Claude修后范围已执行并单独封存:完整目录14/选定在内,两新请求被上游403分组客户端权限拒绝,usage/身份/费用未知,工具执行0;见[修后验收](QA-CLAUDE-PROXY-CORRECTION-2026-11-12.md)。41原件/当前七连接与全部原状态通过,原M4的40请求/541封存保持;提交推送收尾,供应商须另开放普通Messages客户端,不追加两请求或重跑工程/Purge。M8后续另冻,真实收益与M5–M7边界保持。
|
||||
|
||||
2026-11-12 M4验收e132601b及前六条已普通增量推送,远端/fetch逐SHA一致,见[回执](gitatlas-m4-live-push.json)。完成回执另提交核最终HEAD;原40请求/541封存和1.25.13程序保持。当前先另冻人类修后Claude两预检,随后M8原版Purge另做可比实验;下方推送收尾为原时点。
|
||||
|
||||
2026-11-12 当前:M4新三线六预检与Node/React/Python完整结束,共40真实请求629106报告tokens/0unknown,原检查/Git及状态审计通过,541原件134884464字节封存逐SHA复核;见[最终验收](QA-M4-LIVE-BASELINE.md)。普通GitAtlas增量推送收尾中,1.25.13程序不变。封存后人类更正Claude代理Key,另冻当前修订及新Task/Run最多两文本/工具预检,旧身份失败/目录缺项保留,不重跑工程/Purge或覆盖人类配置。原版Purge的M8对照另冻,M5–M7、Mnemo生产语义/真实手机/长期与各因果收益仍须自身证据。下方“下一M4”按历史时点理解。
|
||||
|
||||
2026-11-11 当前:M3+B4b 已随1.25.13正式激活,来源35a6d6d9;独立1399输入/30编译、186相关/4打包、实际Inno11/11、F跨版7/7、Mnemo24后端及网页/Electron/伴随工程路径通过,正式原PID63892/102项目/127会话/5244工件保持。见[验收](QA-MNEMO-RELEASE-1.25.13.md)。源码/验收1e309700已普通增量推送且远端/fetch逐SHA一致,见[回执](gitatlas-mnemo-1.25.13-push.json);下一M4另冻当前启用线路的新真实兼容与自有工程基线,Mnemo生产绑定/语义收益另验;历史网络/unknown/重复读/R-W边界/Purge收益不改写。下方“未发行/下一M3”均保留原时点含义。
|
||||
|
||||
I27/I28推送完成:源码与验收44b6a09b已普通增量推送至既有GitAtlas分支,远端HEAD和重新fetch的14个提交逐SHA一致,见[回执](gitatlas-m2-ci-push.json)。回执/完成接续最后另提交并再次核对;最终4a03ec52干净1222/1222及7595封存原件保持,正式未部署/重启,runner/job/真实供应商及历史收益unknown不代签。下一按M3-RELEASE-PLAN.md推进实际schema2迁移与新结果后的兼容回退;下方进行中说明按原时点理解。
|
||||
|
||||
2026-11-08 M2 I27/I28本地CI与官方契约已完成,最终4a03ec52干净146文件1222/1222、1376输入/29编译、网页/实际隐藏Electron7组4图实看通过;原失败保持,7595新原件封存及五旧根21297原件只读exact,见[验收](QA-M2-CI-OFFICIAL-CONTRACTS.md)。普通GitAtlas增量推送收尾,远端runner/job/上传仍unknown,不能称远端自动化已验证。M1/M2所列源码切片均已有本地证据,正式未部署;下一[M3计划](M3-RELEASE-PLAN.md)实现精确schema2迁移和新结果后的兼容回退,再统一发行与M4新真实基线。2001操作14.5分钟回归只证明记录保留,M7性能待办仍在;M5–M8和历史真实异常/重复读取/缓存/R-W边界/Purge收益unknown不因CI全绿代签。
|
||||
|
||||
2026-11-07 M2第五批I09/I11已完成源码及隔离验收:来源71f18055/核心9e1feeeb/界面ee97aaae,原生缓存块/工具元数据保持、跨Run状态身份与四范围连接保持,纯Chat混池实际误带状态缺陷已修。最后16/16、首127/129加两文件26/26分开按SHA复用,最终880输入/29编译/74runtime、网页和实际Electron各8组/10图/三配色/1280及390、原存储/密文/历史exact,5126原件封存,见[验收](QA-M2-NATIVE-CACHE-STATE.md)。源码与验收1a897576已普通增量推送、远端逐SHA一致,见[回执](gitatlas-m2-native-cache-state-push.json);完成接续最后另提交核对。下一I27/I28 CI,再M3实际schema2迁移增量和新处理器结果后的回退,不能删除新账本或直接移除逐表全等守卫。正式未部署/重启,无真实Key/模型,历史网络/unknown/重复读/R-W边界/Purge收益unknown不动;下方历史接续按原时点理解。
|
||||
|
||||
2026-11-07 M2第四批I14与I12本批指纹已完成源码337d1bc9/f95507ab/26472d28:请求版本/Beta/认证/头/路径与目录身份绑定、实际fetch参数与原Run冻结设置分别记录;原body/ordered-prefix和密文保持。最后相关38/38、首候选143/143后端输入未变,最终871输入/29编译/74runtime、两端各8组/8图/三配色通过,旧I13/I15封存只读一致,见[请求契约验收](QA-M2-REQUEST-CONTRACT.md)。3207原件已封存,源码/验收04c0556c已普通增量推送、远端逐SHA一致,见[回执](gitatlas-m2-request-contract-push.json)。完成接续最后提交再核对;下一I09/I11缓存块/跨Run亲和/原生状态和I27/I28 CI,再M3声明迁移增量与新处理器结果后回退。正式/真实Key/模型/旧实验与历史unknown不动,下方历史接续按原时点理解。
|
||||
|
||||
2026-11-07 M2第三批I15已完成ba4b4491/b5aaf1f7及来源0cbf14a2:26记录/13精确型号/17来源、完整目录/覆盖/原规格快照/凭据绑定保持。最终858输入/29编译、两端各6组/7图与5688原件核对,见[注册表验收](QA-M2-OFFICIAL-REGISTRY.md)。首候选62/62后端输入未变,最后UI补强另有新构建和两端证据。普通增量推送收尾中;下一先I14实际版本/供应商头/指纹,再I09/I11/I12缓存块/亲和/状态及I27/I28 CI。M2/M3与真实收益未因此完成,正式/Key/模型/历史unknown不动。
|
||||
@@ -101,6 +119,12 @@ M1第三批实际2001次手机reserve/2000次finish及4002审计保持成功,
|
||||
|
||||
Mnemo/其他产品从双方实际原件契约实现导出、入站委托、归属/撤销/失败;多Run Task和单Run包不能混同。手机以当前物理设备跑完整配对/选择/批准/停止/断线/撤销,不能以本机模拟替代。
|
||||
|
||||
2026-11-11来源依赖登记:智枢独立RS为7/7交付、续办质量0/4、零工具执行,四后续检查点未逐字保留完整原任务;不能据此断言语义全部消失或旧RD04根因。原任务/配对成功工具回执跨摘要保留方案仍在来源方隔离实施,未发布或证明收益,见[只读协调登记](integration/MNEMO-RESUME-SUCCESSOR-READONLY-2026-11-11.md)。未来相关集成先核来源与完成权威、更新/撤销和真实收益;当前仍按M4推进,不因来信立即适配、恢复本地额度或续跑旧实验。
|
||||
|
||||
2026-11-11流式依赖登记:智枢NFD/ISS-041已部署明确failed分支的已有事件尾页排空,抛原错、不伪造成功、不重派;105回归/14部署等按来源结果,unknown/旧CT03及质量根因不改签。EQD已结束且不采用当前35B32K执行候选,严格7/8及原时延保留。见[只读登记](integration/MNEMO-NODE-FAILURE-TAIL-READONLY-2026-11-11.md);本轮无运行时复验或Workbench程序/模型变更,持续公网/手机/P8/真实收益边界与M4顺序保持。
|
||||
|
||||
2026-11-12 M7来源补充:Mnemo指定旧AG42aacbd1/b330b966同Goal原件连接和显式完成检查点有限范围完成,113相关/69独立/31隔离HTTP与原9/build/57复用单列,0模型。常驻4321原只读404/未载入,生产者认证/完整指纹/派发接收/父交付/语义仍未验;见[只读登记](integration/MNEMO-GOAL-WORKER-READONLY-2026-11-12.md)。原AD、M4三工程与该旧AG分开,不据此重派、重验Claude403或启动Purge,也不代签生产语义和收益。
|
||||
|
||||
### M8:真实收益单独证明
|
||||
|
||||
缓存指标取上游实际read/write/uncached;区分新增文件读取、内容变化后的必要回读与同SHA/同范围重复读取。提醒、经验、角色分别固定来源和开关,比较工程成功率/质量、tokens、时延及原失败,不把工具进展观察当业务交付。
|
||||
|
||||
@@ -0,0 +1,63 @@
|
||||
# M2 第六批:CI 与官方契约回归闭环
|
||||
|
||||
2026-11-07,基线 `470d694831c315c034e3f36891e4f3faf94cbe0f`,分支 `feat/task-agreement-1.25`,干净工作树。承接 I09/I11 已封存交付,不重做上一批。完成 I27 可重复的源码/契约/构建/安装检查入口与 I28 官方来源、版本差异及弃用的回归关联;随后 M3 单独冻结统一发行。
|
||||
|
||||
1. [complete] 固定470d6948基线/tree/锁文件/Node身份,planning catchup退出0。GitAtlas实际1.26.2且Actions启用,runner/runs匿名401,未确认执行器;1.26官方三篇HTTP200原件与重定向/401保存,见[来源](CI-SOURCES.md)。当前外部Python3.11.15缺pytest,缺环境失败原件保留。
|
||||
2. [complete] 唯一run/原日志/exit/PID/环境、源码ZIP前后SHA、固定Node/npm/Python及带哈希venv依赖、官方四来源清单/26记录/版本字段弃用差异与14回归输入已实现。新专项和源码包最终11/11,首10/11保留;来源SHA/省略凭据getter/unknown命令/跳过不能改绿的边界通过。尚不替代独立干净环境结果。
|
||||
3. [complete] `.gitea/workflows/ci.yml` 与本地contract/windows入口、真正YAML解析与固定action refs已实现;精确该workflow路径进入源码包白名单,实际自有Git ZIP和其它.gitea配置排除通过。实际Windows网页/隐藏Electron入口1280/390与关闭核对准备,独立运行尚待。
|
||||
4. [complete] 最终已提交4a03ec52独立干净源码实际新npm/Python QA/Electron安装,146文件1222/1222、0失败/取消/跳过/todo;实际Edge/隐藏Electron7组/4图并逐张实看,1376导出输入/29编译前后一致。来源/环境/原日志与TCP清理复核通过,不将本地结果称远端执行。
|
||||
5. [complete] 分阶段源码与验收44b6a09b已普通增量推送,远端HEAD/重新fetch的每项提交SHA一致;7595新原件/清单和五旧根21297原件保持。见[推送回执](gitatlas-m2-ci-push.json),回执与完成状态最后另提交并再次核对。外部runner/job/上传仍未知,不安装持久runner或以本地通过代签远端。
|
||||
|
||||
新证据只写 `.qa/m2-ci-2026-11-07/`,原 I10/I13/I15/I14/I09-I11 封存根只读。本批不委派、不读真实模型 Key、不请求真实模型、不操作正式配置/数据/安装/重启,不续旧实验或用户工程,不安装持久 runner。公开 GitAtlas API/官方资料仅作只读调查,失败与无法确认状态均保留。
|
||||
|
||||
默认 unrestricted/passthrough/preserve 与持久 ask/auto/full 保持;CI 不恢复预算/上下文/输出/历史限制或逐步审批。W/AD/AE/AG 网络/unknown、重复读取普遍收益、R/W v2 真实剔除边界及 Purge 收益 unknown 不变。官方夹具只证明确定性契约,本批无真实供应商验证。M3 必须验证声明的 schema2 迁移增量及新处理器实际产生结果后的回退,不能删新账本或直接移除逐表全等守卫。
|
||||
|
||||
## 调查与原失败
|
||||
|
||||
- 接续读取组合输出过大导致工具截断,随后限定相关段落;未改产品或旧证据。
|
||||
- PowerShell 未引用的 `HEAD^{tree}` 被解释为脚本块,Git 调查失败;改用引号固定 revision 参数,未产生提交或数据改动。
|
||||
- 只读调查误猜 `visible-browser.test` / `request-contract.test` / `browser-launch` 和 PowerShell 不展开的 rg 路径通配符;改用实际文件列表,不记作产品失败。
|
||||
- GitAtlas 公开 API 实测 Gitea 1.26.2、仓库 Actions 启用;runner 和运行 API 匿名均 401,执行环境仍未知。当前仓库 API 的 `private=false` 作为观察保存,不更改用户仓库配置。
|
||||
- Gitea 无尾斜杠/旧 runner 入口返回 308/301,原件保持;另取对应 1.26 官方三篇 HTTP 200 正文,不能将重定向当原正文通过。
|
||||
- 外部Python3.11.15当前缺pytest,`import pytest`实收ModuleNotFoundError;计划新建独立QA环境并按带哈希的锁文件安装,不污染原解释器环境。
|
||||
- 来源准备首轮把Colorama实际`py2.py3`通用wheel误判为缺wheel;未处理的并行失败使其余部分响应未落盘。第二轮误假设所有响应都存在,实收ENOENT。第三轮顺序复用已存原件并仅补缺失PyPI元数据,7个精确wheel及SHA已固定;两原脚本/失败保持,不改成通过。
|
||||
- Checkout原metadata HTTPS首次fetch失败、第二次45秒超时、第三次公开Git clone连接被重置(128);第四次Gitea官方actions镜像按同一v4.2.2 commit取得HTTP200原metadata。四记录分开,不能称GitHub网络已恢复;工作流采用同commit镜像URL。
|
||||
- YAML依赖准备器首相对import多退一级,实收ERR_MODULE_NOT_FOUND;原脚本保持,第二轮正确import成功,只新增yaml和根依赖声明两个lock条目。
|
||||
- 官方快照首准备器把已有受控同级fixture引用误判为逃逸;原helper保存,改为整个固定`tests/fixtures`范围,仍逐原字节SHA核对,未改原SOURCE.json或来源正文。
|
||||
- 首专项11项为10过1失败:新源码ZIP fixture未声明workflow的LF,实际Git archive按全局属性导出CRLF。保留原11项日志/测试,只为自有fixture明确声明与产品一致的workflow LF,再单独补验该文件。来源索引检查首.gitattributes工作文本CRLF与Git LF不一致,原字节保存后只规范该文本,14暂存原字节再核对通过。
|
||||
- 源码提交首CI入口末尾空行被diff检查拒绝,随后误继续启动CI被未提交输入前置检查拒绝。原preflight失败完整保留,修空行后才af42b949提交并另起干净run,不把失败算通过。
|
||||
- af42b949独立npm/Python/来源/工作流已通过;Electron包读取路径会spawnSync隐式下载,缺独立日志,随后实际成功并继续受记录的Chromium安装。保留原运行/进程观察,根源码改为在过滤环境内显式electron-install命令,后续另run;当前原run继续验证其自身冻结源码,不复写/冒签新入口。
|
||||
- af42b949的Chromium下载正文持续0字节;只读HEAD为307→storage200但不证明正文成功。按pid62420/父64192/命令行核对后结束自有下载子树,原run非零失败保留。实际产品浏览器契约只要求Edge;改CI按实际Edge与Electron通过固定Playwright运行,不省略浏览器检查,缺Edge仍失败。
|
||||
- 上传action默认不包含隐藏文件,复核发现需显式`include-hidden-files:true`才能保留白名单`.qa`证据;新增反例断言,不将原工作流静态通过当实际上传通过。
|
||||
- 自有Chromium结束首条件误要求相对CLI命令带绝对工作目录、第二次JSON日期经Parse丢UTC类型导致误差8小时,均安全拒绝未结束进程。第三次从原JSON字符串精确解析UTC,pid/父/程序/参数/启动时点差0.003436秒一致后才结束;原run exit1、625374ms原下载日志保留(记录interrupted=false仅是外部taskkill未经过runner信号处理,不能解读为自然网络失败)。
|
||||
- 0b4f1de4第二干净run在过滤环境下官方Electron下载暂存0字节,552652ms后经原PID/父/UTC身份核对主动结束,自有命令exit1/原日志保持。公共HEAD/首1024字节与完整响应首块探测成功只证明这些范围,未当成完整下载。新增显式公共二进制镜像选择(版本目录/原官方checksums不变)、Python UTF-8和URL不得携带凭据的边界,后续新run;不继承真实API或电脑下载配置。
|
||||
- Mnemo记忆桥第一次源SHA实读一致,后来准备封存时已被并发更新为763d796b;校验即拒绝,当前新原文与旧观察分开保存,未采用新版本/接入服务/发模型/回复来源线程。
|
||||
- 36c6be4f第三干净run确实完成npm ci、独立venv/wheel安装、带官方checksums的显式镜像Electron及build;全部146测试文件持续运行,不提前中断。首四项为近期原生块/品牌/输入容量与请求绑定的旧断言:限定三个测试已核对语义并补原生往返、独立上下文/输入容量和请求绑定,专项41/41通过,原失败原日志保持;测试文本随后仅按已声明LF规范化,最终干净CI另验最终字节。
|
||||
- 全套另暴露engine-context测试在模块加载时依赖源码包故意排除的历史档案;静态核对还有model-reservation和original-purge同类包外输入。下一只将必要的解析子树/公开锁定源码整理为带原SHA/Git blob的测试夹具,保留历史全集,不扩大源码包到项目/会话/凭据或证据档案,不删套件或加skip改绿。
|
||||
- 最小回放准备实读112份已提交来源并核当前字节与Git blob一致;生成65份2452465字节(四engine子树、一R原证明/六回执/请求子树、60锁定公开Purge源码)。INDEX逐SHA/字段来源及公开源码原license/NOTICE保持,原来源前后不变;三个套件保留所有断言,仅改为本地声明输入。新CI前置核全部夹具,漂移/缺失/逃逸反例待专项验证,不将JSON投影冒称完整原会话字节。
|
||||
- 四engine新/旧原生输入、模型配置、工具回执子树和实际engineContextView输出逐项一致;R原证明/六回执/请求一致,112原来源再次核SHA通过。原三个回归套件加CI负向边界专项35/35(39303ms),未知文件/同长度SHA变化/缺文件/路径逃逸均不能通过;未新增真实模型或实际Harness派发。旧全套仍持续收集,尚不记CI全绿。
|
||||
- 冒烟清理复核发现原`portClosed:true`仅由HTTP health不可取和PID退出推导,未直接核监听。增加仅针对本次记录127.0.0.1端口的TCP接收/拒绝观察;缺health但仍监听不能通过,异常/超时保持失败,不结束外来进程。同页watch去重避免把一次浏览器事件记两次。首次直接工具专项8/8(没有独立磁盘原日志),规范最终文本后另用记录器核最终输入;实际Edge/Electron仍待最终Windows run。
|
||||
- 36c6be4f首完整Windows run自然结束:146文件/1211项,1194通过17失败,0取消/跳过/todo,3534259ms;未运行后置桌面冒烟。原报告/原日志/最终MANIFEST不追加,另在`remaining-before-2/first-complete-ci-review.json`只读核1308源码输入与编译文件前后一致。此前12项由旧断言/包外回放输入修正覆盖,余项仍单独追因,不冒称首全套全绿。
|
||||
- 交付报告增长后,Python HTTP测试仍把64KB预览当完整JSON;改用原下载路由并与工件元数据逐字节长度/SHA核对,再保留原Git/工程/测试/文件来源断言,不裁短报告或删账本。虚构`deepseek-owned`前缀不再推出官方厂商;补未知兼容、显式provider和精确官方型号三种来源断言,原原生发送与负向输入核验保留。
|
||||
- Python深路径独立复现原ensurepip及开启UTF-8后的失败;内层原错明确为setuptools包路径WinError206,并非编码根因。该用例只运行stdlib unittest,改为真实`venv --without-pip`,149字符项目环境仍静态deferred、实际解释器3.11.15和1项unittest通过。CI自身带哈希pip/pytest安装不省略,产品解释器识别/身份守卫未改。
|
||||
- 局部首复验32项27通过5失败;复验器误额外设置`AGENT_WORKBENCH_PYTHON_PATH`,覆盖项目`.venv`选择导致其中3项失败,与首完整CI环境不同。保留原命令/输入/日志,移除复验器该覆盖后另轮验证。其它推理旧断言与研究读取失败再现,提醒原20秒等待在局部原用例通过;不把这些替代首并发失败。
|
||||
- 研究原失败JSON逐层复核发现首因是旧16K有限夹具的Evidence字段读取返回“预算无法容纳元数据”,随后本机模型夹具解析失败并触发其重试,才出现第4/5次工具范围断言。完整原状态在`research-first-failure-state-1.json`,这是受控夹具错误,不是外部供应商异常。下一让该研究功能用例采用当前不限/完整历史默认,同时核对开放诊断工具广告与既有执行契约一致;不恢复额度或新增执行拦截。
|
||||
- 原提醒五请求/原20秒等待另做逐请求和状态观察,11.6秒独立用例通过;原并发超时的最后状态未封存,不能补造唯一性能根因。后续验收按五个实际请求的进展观察等待,仍核一次发送/当前证据/resolved/跨会话拒读/重启持久,不把总20秒误当产品运行额度。
|
||||
- 三处测试修正最终规范文本专项23/23,0失败/取消/跳过/todo,102798ms;正确QA解释器环境下,先前误覆盖的三项识别/身份用例均通过。全部产品/测试/脚本与编译输入前后一致,见`remaining-python-profile-fixed/run-windows-mw6jyjbh-5f925ebf/evidence/`。这仅是复用已建依赖/编译的专项,最后干净安装全套仍待;原首轮与局部失败未改写。
|
||||
- 核对现有执行端后,开放/full的诊断仍按`taskToolAllowed`保留只读意图;原工具广告却在unrestricted下绕过它。只修发送工具列表和排除原因,使广告与已有执行契约一致,未新增执行守卫;研究HTTP用例采用当前不限/完整历史默认,原5请求/1分析/原Task/Evidence/诊断只读断言全部保留。新增full诊断/无需测试广告、强行写入仍按原契约记错、无审批/文件变化/主历史替换的回归。
|
||||
- 最后新编译下三个实际文件22/22(47111ms),另实际`tool-prefix-observation.test.mjs`4/4(12270ms)通过;分别保留来源/编译SHA。首专项命令误带不存在的`tool-surface.test.mjs`,Node未拒绝该参数;只承认实际三个文件的22项,不宣称四文件完成,纠正在`task-surface-actual-input-review-1.json`。局部复验器随后加实文件前置;最终CI原本已逐个读取测试输入SHA,不采用该错误参数。
|
||||
- 提醒验收按5个实际请求及最终结束逐阶段观察,每阶段仍20秒停滞失败并保存最后运行状态;新编译下5请求/resolved/原设置与跨会话/重启全部通过。它只是验收等待方式,未修改产品不限时长、模型等待或预算策略;原31秒并发失败及两次原20秒局部通过分别保留。
|
||||
- 第一次实际Edge冒烟在1280模型中心通过,390切换后导航仍点击刚变inert的侧栏而30秒失败;真实失败DOM28131字节/PNG42376字节已保存并实看,0页面脚本错/外部请求,所属宿主65732及记录TCP58255已实际退出/拒绝连接。未运行Electron,不能冒称桌面通过。改模型中心验收导航使用两宽度均可见的工具箱;不force点击或改产品布局。失败捕获已补DOM及隐藏Electron原生截图,原错不被捕获失败替换。
|
||||
- 第二次实际冒烟`run-windows-mw6ko1ui-9d5564cc`原报告7组/4图通过,但实看原生1280仍为加载画面、390仍为工具箱,DOM与PNG不同步;不接受为完整视觉通过,原报告的`passed:true`与原图照存。第三次`run-windows-mw6kwdlc-8ead8ec3`在模型卡片/可用默认选择、字体、有限动画及两个实际RAF后原生截帧,7组/4图通过(命令32314ms),四图逐张实看均为加载完成的模型中心;宽窄网页/原生无文档横溢,窄原生内部滚动条保持。隐藏QA窗口临时关闭background throttling,尺寸逐项核对,不更改产品窗口行为;录入字段改名`requireLoadedModelCard`明确它是要求而非预填成功证据,观察在DOM的`painted`字段及原图。
|
||||
- 第三轮来源/构建前后未变,0模型尝试/页面脚本错误/外部页面请求;宿主62812/Electron13276退出,TCP62009实收ECONNREFUSED。第二轮宿主67464/Electron39816与TCP65462亦结束,不能由截图不足推导产品服务失败。两轮原报告/PNG/DOM不改写,另存只读视觉及SHA复核;最终已提交源码的干净Windows全套仍待。
|
||||
- 本次接续planning catchup退出0无补遗;组合读取历史内容再次遇工具输出截断,随后按当前规则/计划/实际diff限定输出。新日期为2026-11-08(本批来源与证据目录仍沿2026-11-07),不重做已完成专项,不改原证据或正式服务。
|
||||
- 本次提交前辅助索引核对误猜专项快照名`inputs-before.json`,实收ENOENT并以exit1停止,未继续提交。按实际证据文件名核对后再提交;这是辅助读取错误,不改原专项统计或产品。
|
||||
- 必要CI专项辅助记录器首命令id误含`.mjs`被既有仅字母/数字/横线守卫拒绝,尚未启动任何测试。原helper/已创建输入保留,另存check-2用稳定数字id后再执行;不更改产品记录器或放宽守卫。
|
||||
- 运行时/研究/任务广告修复已限定提交`87ca0751`,暂存字节与22项专项冻结输入逐SHA一致。提交前八图/八DOM SHA和PNG实际尺寸复核通过,最新两张原生图本次再次实看;两轮记录PID均不存在、TCP65462/62009重新实收拒绝,只观察自有QA端口。提醒/导航/截帧最终代码语法与CI pipeline8/8通过,来源SHA前后一致,`requireLoadedModelCard`仅记录语义更名;下一限定提交后以最终已提交源码另起真正干净Windows全套。
|
||||
- 提醒/导航/截帧与本次进展已提交`4a03ec52`;最终干净`run-windows-mw6lllq7-33d6c439`从此commit导出1376输入(源码ZIP5105901字节),真正npm ci40965ms、新venv/带哈希依赖、显式镜像Electron/官方checksum30666ms与build7013ms均通过,全部146文件正在运行。随后Mnemo只读文档单独提交`1a9c51b1`,没有改变最终已冻结的程序/测试输入,不以当前文档HEAD冒称其冻结来源。
|
||||
- 独立预读M3时误猜旧core-database、model-attempt-store/request-ledger-store路径且使用PowerShell不展开的storage通配路径,rg实际读取报错;改为`rg --files`列出的storage/ledger输入及`install-manager.mjs`,仅规划观察,未写安装/存储实现或正式数据。
|
||||
- 最终全套大记录阶段的辅助只读SQL观察器移到更深目录后仍用原相对import,首ERR_MODULE_NOT_FOUND未打开数据库;原helper保留,另存正确深度再观察聚合计数。测试自身继续运行,未改产品/测试或终止任务。
|
||||
- 最终Windows自然结束:4a03ec52、146文件1222/1222,3392886ms,全部命令exit0/interrupted=false;新依赖/完整构建之后真实桌面39077ms通过7组/4图。最后只读核1376源码输入/29编译、锁定Node/npm/Python/Playwright/Electron/Edge、原命令/报告/清单 SHA一致;四PNG逐张实看加载完成的模型中心,原生宽窄没有第二轮画面滞后,所属宿主/Electron和记录TCP再次退出/拒绝,0模型尝试/页面错误/外部页面请求。最终2千操作回归872353.5061ms,保持完整4002审计/unknown/重启;这证明记录保留,不代表规模性能已经优化,M7性能待办保持。
|
||||
- 协调方国内入口四固定原件SHA和反馈9efcc256一致,离线60原键/一次POST/源与本机任务/closeout输入SHA核对通过,仅保存字段/计数而未重跑HTTPS/34API/12Edge/900秒调用。来源报告7.828秒与原手机/SSE/unknown分别保持,文档3d649b3b单独提交;主CI/M3默认域/模型Key/route/预算/权限未改,也未向来源线程回复。
|
||||
- 最终记录patch首误匹配机器生成器不存在的独立行,工具拒绝且整组未应用;按实际finalUI锚点再保存规模时长字段,未更改原CI报告或程序输入。
|
||||
- 7595份150586505字节已封存,MANIFEST SHA aaeaf694a1cb2444ee8bdfd5ec74ccc768c5659a38fdd6b3be36d6d69abe4951;60个执行依赖/缓存/临时目录明确排除且磁盘保持,无链接遍历,原失败未删除。当前33指南链接/11文档及统计/输入复核通过,验收44b6a09b普通增量推送成功;远端HEAD与fetch提交链逐SHA一致,原封存根不追加,推送记录只写新close根。下一M3实际迁移/保留新结果回退,正式未操作。
|
||||
@@ -0,0 +1,17 @@
|
||||
# M2 第五批:原生缓存块与跨 Run 状态
|
||||
|
||||
2026-11-07,基线 `bdfeb29c85b3231bfd7bc24b8b5f49d40a3beebf`,干净工作树。承接 I14/I12 的冻结契约/当次 fetch 数据区分,完成 I09 原生块与工具缓存注解往返、I11 可配置会话/Task亲和与供应商状态绑定;I27/I28 CI 与 M3 单独续接。
|
||||
|
||||
1. [complete] 固定当前官方缓存/状态来源、复用 I15 原件并核SHA;8份来源清单见 tests/fixtures/native-cache-state/SOURCE.json 与 docs/NATIVE-CACHE-STATE-SOURCES.md。保留原丢注解、跨 Run 选线和状态身份缺项观察。Mnemo新结果只读登记,不作为本地验收。
|
||||
2. [complete] 保留 Messages 系统/用户/工具块及工具元数据;新增原生状态身份绑定和会话/Task/Run亲和,修复纯Chat混池仅换model造成旧状态误用,optional亲和存储失败不改变原响应。原用户设置、完整历史和未知原件保留,默认不添加缓存参数。source-5最终16/16,源码类型检查通过。
|
||||
3. [complete] 自有 HTTP JSON/SSE、混池、跨 Run、手选换模、修订/停用、停止/恢复/重启与所有工具结果不重放的确定性核对;上游原错误保留。source-5为16/16;首候选127/129,两项旧亲和假设准确修正后两文件26/26。最终178后端输入与首候选一致、213测试输入中211未改,按SHA复用103原通过项,不冒称最终单轮全套全绿。
|
||||
4. [complete] 复用 Orven 模型中心/原生参数/上下文,四范围保存/重载、跨Run沿用/自动重选/手选与原错通过。build-5/candidate-5为880输入/29编译/74runtime,逐SHA核对;web-ui-5与electron-ui-1各实际8组/10图/三配色/1280及390、原metadata/cipher/存储快照/原历史exact。实际Electron保持隐藏窗口原生capturePage,正常关闭后独立宿主恢复。
|
||||
5. [complete] 来源71f18055、核心9e1feeeb、UI ee97aaae与验收1a897576分阶段提交并普通增量推送,远端逐SHA一致。5126原件封存/22Git输入/四旧根16171原件/两端8组10图/自有进程与监听0核对完成;原失败/unknown保持,回执及完成接续最后另提交复核。
|
||||
|
||||
本批不委派、不读真实Key、不请求真实模型、不操作正式数据/配置/安装/重启,不续旧实验或用户工程。默认 unrestricted/passthrough/preserve 与持久 ask/auto/full 保持,不增加预算、上下文、输出、头或历史额度。工具权限以当前模式为准,不能把状态核对做成新增审批。缓存使用与供应商命中由原 usage 表示,保留 unknown;稳定前缀/亲和或相同指纹不是缓存收益证明。
|
||||
|
||||
W/AD/AE/AG 网络/unknown、重复读取普遍收益未证明、R/W v2真实排除边界与 Purge收益unknown保持。M3须真实 schema2迁移增量及新处理器产生结果后的回退,不能删新账本或直接移除逐表全等守卫。新证据只写 `.qa/m2-native-cache-state-2026-11-07/`。
|
||||
|
||||
接续保留原 source-1/source-2/source-3:首夹具错误、亲和来源断言失败和跨线状态整正文搜索失败仍待准确核对,不能把接续摘要当全绿。首次接续 mkdir 遇既有目录即停止;改为读取已有原件并使用新文件名,未覆盖证据。规划辅助只读路径/输出错误见 progress.md。
|
||||
|
||||
source-4纯Chat混池问题已修后source-5通过,原source-1至4保留。网页1精确标签、2宽屏缩到手机后抽屉隐藏、3误期待失效亲和仍选A、4手机抽屉挡后方工具箱的原失败/图/驱动分别保留;已按真实UI返回路径和自动选线结果修正,新增手选A一次原400验收。最后文本规范化检查0项改动,来源原尾空白保持。
|
||||
@@ -0,0 +1,30 @@
|
||||
# 1.25.13:Mnemo 记忆桥与迁移恢复交付
|
||||
|
||||
2026-11-11。用户要求立即开发、提交和验收;本次授权接续并覆盖旧接续说明中的阶段性“不提交/不发行”安排。沿用现有分支和 GitAtlas 普通增量推送授权,不启动其他线程或旧工程。
|
||||
|
||||
## 交付范围
|
||||
|
||||
- 将已实现的 M3 schema2 精确迁移、结果保留和兼容恢复,与 B4b 默认关闭的单项目 Mnemo 记忆桥正式发行。
|
||||
- 保留默认开放透传、完整历史及持久 ask/auto/full。源码包不含用户项目、运行数据或凭据。
|
||||
- 记忆桥当前凭据仅存宿主内存,重启需重新授权;真实生产 Mnemo 绑定、真实模型语义收益、长期效果、物理 iPhone 和 Purge 收益不以合成验收代签。
|
||||
- 保留历史失败、网络和 unknown、重复读取、R/W 真机边界及旧封存原件,不自动续跑旧实验。
|
||||
|
||||
## 阶段与证据
|
||||
|
||||
1. **完成**:复核 68 项原待提交文件和现有验收;加本轮计划/版本/根记录后冻结 76 项 SHA。正式只读观察为 1.25.12/PID 63892,未套用历史 PID。已更新 package/lock/诊断常量和发行文案。
|
||||
2. **完成**:固定 Node 24 构建通过;13 文件联合相关回归 **186/186,316246.0644ms,0 跳过/取消**,覆盖 M3 六文件和 B4b 七文件。diff 与敏感文件审查后限定提交产品、测试和原验收文档;本轮不是全仓单轮测试,原 M2 全量证据保持其原输入身份。
|
||||
3. **完成**:最终来源 35a6d6d9;1399 输入/30 编译独立 clean npm ci 与重建 exact,程序/Setup 生成及归档登记通过,打包不改工作树。
|
||||
4. **完成**:最终候选实际 F 跨版迁移/兼容恢复 7/7、Mnemo 后端 24/24、网页/隐藏 Electron 各 4 组/10 图、基础文件 9/9 通过,八图实看;实际 Inno 生命周期 11/11,包含初装、原窗口/在途请求/原审批保持、故障回退、保留 PID/凭据/全部数据的更新、原生入口、重启恢复合成凭据及卸载保留数据。
|
||||
5. **正式激活、验收与推送完成**:唯一正式 Setup 351789ms 成功,正式 1.25.13/原 PID 63892/原 instance 保持,schema1→2 精确迁移 374 旧请求;102 项目、127 会话、8741 记录、22085 事件及 outbox、5244 工件的全部原行/原字节/公开配置保持。公开规格/契约/账本和两项 MCP 导出新增字段逐项核验后,其余原字段全等;网页/实际 Electron 的 1280/390 四图实看后再次 final 核对通过,9229 关闭,新版用户桌面正常恢复。验收和[机器身份](mnemo-1.25.13-release-identity.json)已生成,782 份/811582693 字节选择性证据清单封存;验收1e309700和打包修复已普通增量推送,远端/fetch逐SHA一致,见[回执](gitatlas-mnemo-1.25.13-push.json);完成接续与回执另提交后再核最终HEAD。
|
||||
|
||||
本轮新证据根为 `.qa/m3-b4b-release-2026-11-11/`;旧 `.qa/m3-release-2026-11-08/` 和 `.qa/b4b-native-2026-11-08/` 只读。每阶段记录实际来源、首失败与结果,测试变化才重验受影响范围。
|
||||
|
||||
首构建使用固定 Node 启动 npm 后,其子命令仍命中系统 Node 16,失败原日志 `build-first.log` 保留;驱动给完整子进程 PATH 前插 Node 24,第二轮自然通过。初始快照脚本中文 file URL 路径错误未写出快照,改用 `fileURLToPath` 后冻结;工具错误和记录保留。构建现有大块提示保持,不以提示冒称失败。
|
||||
|
||||
源码实现 `e306f40c` 已提交并普通推送,远端 SHA 相同。首独立源码 1398 输入与 30 编译 exact,真正 clean npm ci 138 包通过。首程序包成功生成后,登记检查发现打包器回写源码许可证而使工作树变脏,登记失败原件保留、首包未安装。现修为只在发行输出生成许可证,并补存 YAML ISC 文本和 notices;最终另提交来源、重新导出和打包,新证据使用 `.qa/m3-b4b-release-2026-11-11-r2/`,不覆盖首轮。
|
||||
|
||||
## 当前限制与验收边界
|
||||
|
||||
恢复载荷必须冻结具体来源和 schema2 兼容范围。未经适配的旧 schema1 程序不能在新结果产生后作为恢复目标;不能删除账本或降 schema。相同发行的独立恢复副本通过,也不能宣称恢复到不同旧业务实现。
|
||||
|
||||
正式状态仅采用本轮实际观察。真实模型 Key 不读取、复制或重存;本次工程验收使用合成凭据和本机受控模型服务,不自动执行用户项目。
|
||||
@@ -0,0 +1,35 @@
|
||||
# M3:统一发行、schema2 迁移与保留结果的回退
|
||||
|
||||
2026-11-11 更新:M3+B4b 已作为 1.25.13 正式交付,来源35a6d6d9,独立1399输入/30编译、实际Inno11/11、F跨版7/7及Mnemo两端工程验收通过;唯一正式Setup成功,原PID63892及全部原数据保持,公开新增字段单独核验,正式四图和final通过。见[发行验收](QA-MNEMO-RELEASE-1.25.13.md)与[机器身份](mnemo-1.25.13-release-identity.json)。源码/验收1e309700已普通增量推送且远端/fetch逐SHA一致,见[回执](gitatlas-mnemo-1.25.13-push.json);下方未发行段落为历史。兼容恢复仍为同源程序副本,真实Mnemo生产绑定/语义及Purge收益未验证。
|
||||
|
||||
后续接续更新:在M3源码/隔离验收后,B4b已按明确授权实现默认关闭的Mnemo单项目记忆桥并完成25项工程证据及最后相关104/104,见[B4b验收](QA-B4B-MNEMO-NATIVE.md)。M3+B4b仍未提交或统一发行,不改变本计划步骤5的最终产品身份、源码/程序/安装器及独立安装要求;下方“B4b只登记”是M3原收尾时点。
|
||||
|
||||
2026-11-08收尾:步骤1–4的源码及隔离验收已完成,见[验收与回顾](QA-M3-MIGRATION-RECOVERY.md)及[机器记录](m3-migration-recovery-2026-11-08.json)。最后1117输入/30编译独立clean依赖重建exact,相关82/82、实际F载荷7/7、两端各5组通过;原失败与未知保持。兼容recovery为同源编译的另一副本,未证明回到不同旧业务版本。当前未提交,按本轮约束不自动Git交付/发行/正式升级;步骤5的最终commit、版本、源码/程序包/InnoSetup及正式安装仍待,不将本轮称为M3统一发行完成。B4b只登记接续、不自动切换。
|
||||
|
||||
2026-11-08,承接 [官方优先路线](DEVELOPMENT-ROADMAP-OFFICIAL-FIRST.md) 和 [I27/I28 CI](M2-CI-PLAN.md)。最终 Windows CI 从 `4a03ec52` 自然完成146文件1222/1222及实际两端7组4图,[验收](QA-M2-CI-OFFICIAL-CONTRACTS.md)已封存;远端runner/job尚未验证。下文保留立项时的基线、实施顺序与结束条件;当前完成范围以上方收尾和验收为准。正式状态仍仅引用 F 的历史记录,不进行新的 health 查询。
|
||||
|
||||
本批源码发行不包含用户项目或运行数据。安装和回退验证另用新建合成项目、独立安装/数据目录与本次记录的端口;正式激活在独立载荷和跨版验收完成后才进入发行步骤,不自动续跑旧实验或用户任务,不读取/复制真实模型 Key。
|
||||
|
||||
## 立项时确认的问题
|
||||
|
||||
原[安装器](../src/core/desktop/install-manager.mjs) 的 `snapshotPersistentData` 核对原表全部行与工件,激活后按完整快照相等判断。这会拒绝合法新增的请求账本表和迁移标记;本轮已按精确增量修复,原行核对继续保留。
|
||||
|
||||
[请求账本](../src/core/models/request-ledger.mjs) 在事务中增加 `model_attempts`、索引、`modelAttemptLedger` 并提升 user_version 为 2,原 `modelAttempts` metadata 字节保持。[存储](../src/core/storage/state-database.mjs) 的新读法投影完整账本;旧 schema1 程序不具备这些读写语义,fresh open 也会拒绝 schema2。
|
||||
|
||||
[热升级](../src/server/hot-upgrade.mjs) 和现有 `rollbackInitialization` 只证明“初始化后没有新增或修改结果”的失败可撤销。新处理器产生记录后,它明确拒绝删账本,不能把这条守卫去掉来制造旧版回退成功。
|
||||
|
||||
## 实施与验证顺序
|
||||
|
||||
1. 固定最后通过 CI 的产品/测试/构建输入、原 F 载荷及新合成 schema1 基线,保存安装指针、宿主身份、全部表行和工件、原 metadata 与合成凭据字节。先独立复现现有安装器对合法迁移的拒绝,保留原操作记录。
|
||||
2. 为候选声明精确的迁移身份和增量:来源/目标 schema、对应源码、原账本 metadata SHA、逐条 legacy_position 和 current/initial payload SHA、新表/索引及标记。安装器只接受这一已核验迁移;原有表行、原工件、原配置字节仍逐项核对。覆盖少行/重复归属/同长度篡改/未知标记/额外表/错误 schema/并发变化;失败不自动重发请求。
|
||||
3. 区分两个回退场景。尚无新结果的初始化失败继续验证原宿主、指针和数据 exact 撤销;已产生结果的回退必须保持 schema2 和完整账本,选择具备 schema2 读写能力的恢复载荷。未经适配的原 schema1 载荷不作为后者的可用回退目标,不能通过降 schema、删新行或覆写旧 metadata 适配它。
|
||||
4. 在实际安装载荷中完成 schema1→schema2 激活,用本机受控供应商产生新的成功、原上游错误和 unknown 请求记录,再实际切换恢复载荷。核对新旧 initial/current 原字节、全部当前行和工件、宿主/Provider 保持及重复工具执行为零;随后真正关闭并重启,验证完整读取/分页/导出以及下一次请求继续写入同一账本。热状态能读不能代替 fresh open 通过。
|
||||
5. 冻结最终版本和产品 commit,独立生成源码包、程序包、安装器及清单;新干净构建、实际跨版安装与回退、关键 wire、网页/隐藏 Electron 核心路径分别通过后,封存并提交。正式安装再核原数据/配置/工件和公开凭据存在状态,保留原错误;源代码发布与这一步的数据保持核对分开记录。
|
||||
|
||||
## 结束条件与保持事项
|
||||
|
||||
每阶段通过相关验证后限定提交;输入变化重新验证受影响范围,文档变化不重复程序全套。恢复载荷的来源、版本、兼容范围必须先冻结并实际验证,目前尚未实现或选择发行编号。若回退事实不完整,保留原错误和当前数据并如实记录未验证,不把删除数据写成恢复。
|
||||
|
||||
默认 unrestricted/passthrough/preserve 和持久 ask/auto/full 不变。这里的安装数据保持与用户要求的无本地模型额度分别处理,不新增任务逐步审批。GitAtlas 本地 CI 与远端 runner/job 分别记录;本批不安装持久 runner。
|
||||
|
||||
M4 新真实官方/代理基线在统一发行之后另冻;W/AD/AE/AG 网络与 unknown、重复读取普遍收益、R/W v2 真机剔除边界以及 Purge 收益 unknown 继续保持。Mnemo 仍只有 [固定原件只读登记](integration/MNEMO-PROJECT-MEMORY-BRIDGE-V1-READONLY-2026-11-07.md),不以本计划代替原生接入或联合验收。
|
||||
@@ -0,0 +1,19 @@
|
||||
# M4 封存后:人类修正 Claude 代理凭据的独立复核
|
||||
|
||||
原 M4 四十请求和 541 成员封存、最终验收及 GitAtlas 推送已完成,收尾 HEAD `c8c36c27`。本轮收到人类反馈“现在应该可以了,我改了Claude 代理,之前是写错key了”;原因属于人类报告,未独立证明。来源公开通知和 SHA 已另存 `.qa/m4-claude-correction-2026-11-12/`,不向来源线程回信,不输出、读取、复制或重存 Key。
|
||||
|
||||
首次公开观察仍为 1.25.13/PID63892、空闲及 unrestricted/passthrough/full、累计/步数均0。原选定池和其他偏好未动;Claude `9ae348a5-d0a6-42ed-8664-0da5f9effe34` 保持 Messages/claude-opus-5-5,连接 revision8→9、credentialRevision1→2,registry182→186、config130保持。另新增 Fable 连接只登记,不推断其操作者,不选择或验收它。公开目录元数据已有人类更正后的新抓取;本轮仍独立按当前修订读取目录。
|
||||
|
||||
先验证独立驱动和原预检判断逻辑,限定提交后一次冻结当前公开连接/配置/宿主与所有现存行、导出、工件 SHA;不能恢复旧凭据或旧注册表。驱动 `scripts/m4-claude-correction.mjs` 复用原已验证 Provider 单次预检与完整原字节下载,不改产品或旧 M4 驱动。
|
||||
|
||||
执行范围仅为这条 Claude 连接:一次目录读取、原文本和工具两用例,各一个全新 session/Task/Run/请求键,合计至多两物理模型请求。每例沿用180秒观察窗口和一个物理尝试,不保存额度/身份策略,不自动重放失败、unknown 或已有 claim。原工具调用仅观察,执行数必须0;不重跑三已通过工程或 Purge,不修改其他连接/偏好,不新增推理或媒体样本。
|
||||
|
||||
逐项保存实际请求型号和官方必填 max_tokens(当前精确型号后备128000)、message_start 型号、完整 SSE/message_stop、原 usage/nullable费用、原错误、请求/响应工件 SHA 与 Task/Run/物理尝试ID。成功只表示本连接文本/工具/流式和公开身份字段的兼容范围,不认证权重或证明工程/长期稳定。
|
||||
|
||||
结束后严格核对冻结前的全部原行/字段/schema、导出和工件。仅允许本批一个新项目/两新会话及其请求/事件/工件、明确项目前插位置和两序列增量;当前七连接/池/选择、配置和凭据行SHA保持。如有外部变化,保留失败和真实增量,不通过回退配置使审计过关。新原件单独封存、回顾、提交和普通GitAtlas推送;原 M4 两 mismatch、目录缺项和四十请求/541清单不改签。
|
||||
|
||||
历史网络、unknown、输出截断、重复读取、R/W真机预留、Purge/Mnemo语义/物理手机/长期与普遍收益边界仍保留。M8新原版Purge对照留到后续独立冻结。
|
||||
|
||||
执行结束:驱动97f132d3在真实前10项离线通过,当前公开状态一次冻结。目录14型号完整/选定在内,两新文本/工具请求各HTTP403 permission_error,分组仅允许Claude Code客户端;两次usage/身份未知、工具执行0,未追加请求。结束审计原58298行/137导出/418请求/5576工件及当前七连接/池/选择/配置一次exact;41成员36284086字节单独封存逐SHA通过,原M4的541清单保持。见[修后验收](QA-CLAUDE-PROXY-CORRECTION-2026-11-12.md)。兼容未通过,执行与证据收尾完成;供应商权限修正后也须另有新范围,不重派本次原键。
|
||||
|
||||
驱动97f132d3/验收5ca65a45已普通增量推送且与远端/fetch逐SHA一致,见[回执](gitatlas-claude-proxy-correction-push.json)。回执及完成接续另提交并核最终HEAD,两封存不再追加。
|
||||
@@ -0,0 +1,48 @@
|
||||
# M4:1.25.13 新真实兼容与自主工程基线
|
||||
|
||||
2026-11-11 开始。承接已正式验收的 1.25.13,不续跑 AD/AE/AG、X、R/W 或用户工程。当前源码基线 `0054ea282180b20061797b00cf5122ee8baa1df4`;本次驱动提交、正式宿主、连接、参数和工程输入另行冻结。
|
||||
|
||||
首次公开观察保存在 `.qa/m4-live-baseline-2026-11-11/observation.json`:正式 1.25.13/PID 63892,空闲、unrestricted/passthrough/full,三个累计/步数设置为 0,六条连接均 `hasKey=false`。已请求用户只在模型中心恢复所需连接的凭据,聊天不接收 Key。此观察不能作为后来凭据可用或真实兼容通过的证明。
|
||||
|
||||
## 阶段与结束标准
|
||||
|
||||
1. 核对源码、现有正式身份、公开连接与选择,恢复规划记录。无凭据时先完成可运行的新驱动与冻结工程准备,不发真实请求。
|
||||
2. 使用新证据目录、新 claim 和自有工程验证驱动。合成服务只验证请求、原件归属、取消、错误、统计和验收机制,固定回答不算模型能力。保留驱动首次失败及后续修正的来源。
|
||||
3. 凭据恢复后,在每个请求之前冻结实际可用、用户启用的连接。官方直连和兼容代理单列;保留用户原选择及参数。不读、解密、复制或重存真实 Key。
|
||||
4. 每连接分别验目录、文本、工具、流式、身份声明、实际 wire 参数、原错误与 raw usage。每项一个全新会话、一个物理请求,不自动重放失败/unknown。官方 Messages 必填 `max_tokens` 按精确型号及实际参数来源发送;其他协议不主动追加本地输出限额。单次响应或工具返回不代表长期稳定或真实模型权重身份。
|
||||
5. 从新的自有 Node 基线开始,模型独立读取、修改、执行原检查并完成限定范围的本地 Git 交付。原测试/配置/owner 文件及原 index 保持,禁止提供实现答案或为过关改变断言。Node 完整成功后依次另冻 React、Python;不拿旧成功代签新版。
|
||||
6. 实际缺陷按小阶段修复、相关验证、回顾与限定提交;实验原件封存、复核和普通增量 GitAtlas 推送。只有程序实际变化需要新的独立发行;文档或驱动不触发重打已验收的 1.25.13。
|
||||
|
||||
## 固定实验规则
|
||||
|
||||
默认保持开放透传及持久 full 权限,不设置本地 context/output/累计 tokens/步数门槛,不裁剪原任务和历史来制造通过。预检只允许一个物理请求;工程观察窗口十分钟,连续三分钟没有新增读取事实、源码变化或实际检查事实时,只停止本次自有 Task 并记实验观察结束。这是观察设计,不改变产品默认能力;在途未报告用量保留 unknown。
|
||||
|
||||
工程选线在结果前固定:用户本轮明确恢复为验收对象的可用官方 DeepSeek 连接优先,其后是这些连接中的 Responses、Chat、Messages 代理。当前用户原选择为 Claude 单成员自动池,工程临时单线选择完成后恢复该原选择。必须本连接目录完整并列出型号、文本/工具预检完整且身份 matched,才运行该工程。失败不换线续跑同一个工程。临时选线只在确有必要时使用公开 API,恢复时以修订为前提,保护用户期间的新设置,不覆盖外部变化;不为结果修改连接定义或保存身份策略。
|
||||
|
||||
用户已回复“恢复了三个模型”。随后公开观察确认为 DeepSeek/deepseek-flash(Chat,revision 20)、apikey-Claude/claude-opus-5-5(Messages,revision 8)、apikey/gpt-6.1-sol(Responses,revision 15),三者 credentialRevision=1、streaming=true、原生参数为空,原选择不变。原无凭据观察保持,新冻结在驱动验证和提交后完成。
|
||||
|
||||
选择和恢复使 configRevision 单调递增,legacy-default 的公开 revision 跟随该精确增量;逐字段核对后单独列明这一投影变化,不把它称为真实连接被改写,也不忽略其他字段。三条真实连接的定义、凭据存在状态和归属、原池及选择须保持。前两轮隔离驱动的状态/公开投影断言失败保持原件,未改变产品行为或原工程测试。
|
||||
|
||||
成功必须有实际源码差异、原检查数量和退出状态、实际提交范围及提交字节等于最终文件、owner 字节/index 保持,且所属请求完整、无 unknown。模型声称完成、返回提案、合成通过、终端文字均不能代替实际工具/检查/Git 证据。内置交付报告和终端 Git 事实分别核对。
|
||||
|
||||
记录新增读取与同 SHA/同范围重复读取、cache read/write/uncached、reported/unknown tokens、时延、错误来源及人工介入。缓存收益、提醒/经验/角色收益均需另冻可比实验;不承诺上游必然命中缓存。
|
||||
|
||||
## 保留事项
|
||||
|
||||
历史网络异常、unknown/失败、输出截断、重复读取和 R/W 预留 v2 真机边界保持原结论。默认透传时不恢复旧限额验证旧边界。真实 Mnemo 生产绑定、语义收益、长期与物理 iPhone 验收继续单列。
|
||||
|
||||
原版 Purge AB-off/AB-on/BA-on/BA-off 必须在当前目标线路有完整可比工程成功基线后另冻插件/宿主/输入/身份。旧 AG 三次网络失败和唯一 ba-off 成功不改写。前置不足则不执行,收益继续 unknown。
|
||||
|
||||
阶段2原时点:最终隔离 m4-driver-40fb93c8 在 1.25.13 正式处理器的隔离宿主上通过 11 边界及固定模型 Node 的五请求实际执行、原九测试/build/限定 Git 与 owner 全部 index,见 [机器记录](m4-driver-validation-2026-11-11.json)。前三根原件及两次失败保持;原件捕获现为 download=1 逐原字节核对。当时 0 真实请求,随后以 `7152d654` 限定提交并另冻正式原状态。
|
||||
|
||||
当前阶段:3、4及5的Node已完成,见[阶段验收](QA-M4-COMPATIBILITY-NODE.md)。三线六次新预检/1227报告tokens/0unknown;Claude两项身份mismatch且目录缺项,不进入工程。DeepSeek新Node十请求180463tokens/0unknown,原9/9/build/限定Git及内置交付completed通过。原127导出/5244工件/53527行保持;外部一新“你好”会话/一请求单列,每行和原件明确SHA核对,原两次审计失败保留。阶段限定提交后继续独立React、完整成功后Python;Purge收益仍须另冻,不改1.25.13或续旧实验。
|
||||
|
||||
更新:5的React完整通过,见[React验收](QA-M4-REACT.md)。14请求271333tokens/0unknown、原9/9/typecheck/build、实际桌面/390px交互与两图实看、仅四源码Git及owner/index保持;终端Git来源和最终verified_committed准确登记,没有补造内置作业。原状态审计一次通过,累计本批30请求/453023tokens,外部一请求仍单列。下一独立Python原十pytest/compileall,完成后最终封存、回顾和普通增量推送。
|
||||
|
||||
更新:5的Python完整通过,见[Python验收](QA-M4-PYTHON.md)。10请求176083tokens/0unknown、原10pytest/compileall/三源码Git与终端来源最终报告,原字节/index保持。原状态及本批九份完成捕获导出一次exact。累计本批40请求629106tokens/0unknown,三工程成功;Claude两身份失败、外部一你好请求和全部历史异常单列。下一6的最终封存、回顾、限定提交和普通增量GitAtlas推送;收益和Purge另冻,不重发1.25.13。
|
||||
|
||||
2026-11-12 更新:6的最终审计与封存完成,541成员/134884464字节,MANIFEST SHA256 `558c9b548559798279d7f8e61deb3b7344920753a53dbf4e84eec374c6f3a3df`;十三最终驱动、所有原失败、三工程及两PNG保持,逐成员核对通过。[最终验收](QA-M4-LIVE-BASELINE.md)与机器记录已生成,限定收尾提交及普通增量GitAtlas推送进行中。原根不再追加或修改。
|
||||
|
||||
封存后收到人类明确Claude代理Key更正;当前公开修订需另冻结,不能恢复旧凭据或用封存前配置覆盖用户变更。修后只做原文本/工具两项、新Task/Run/请求键及至多两新请求,工具观察不执行;不得改签本批旧两mismatch/目录缺项,不续旧unknown,不重跑三工程或Purge。结果独立登记,原40请求统计和541清单保持。M8原版Purge随后另冻,不以修后兼容预检代签工程或收益。
|
||||
|
||||
2026-11-12 阶段6推送完成:验收e132601b及前六条已普通增量推送,远端与重新fetch七条逐SHA一致,见[回执](gitatlas-m4-live-push.json)。回执/接续另提交并核最终HEAD。原批次执行、审计、封存和推送已完成;修后Claude为独立后续,不追加原M4根。
|
||||
@@ -0,0 +1,26 @@
|
||||
# Mnemo 单项目记忆桥
|
||||
|
||||
这是默认关闭的可选项目记忆来源,沿用工作台的项目记忆入口。它共享用户明确确认的来源修订;不改变模型输出、上下文透传、累计额度、完整历史或持久执行权限,也不新增工程执行审批。
|
||||
|
||||
## 连接和共享
|
||||
|
||||
1. 打开工具箱 → 项目记忆 → Mnemo 项目记忆桥。核对界面显示的生产者和项目 ID,在 Mnemo 为这一组合创建明确授权的项目绑定。
|
||||
2. 填写 Mnemo 根地址、项目绑定 ID 和本次项目凭据,再启用并保存。公网使用 HTTPS,本机可用 loopback HTTP;地址不能带路径、查询、跳转或内嵌凭据。凭据保存后输入框清空,仅在当前宿主内存保留。重启或恢复后需重新授权,不会从备份自动恢复授权。
|
||||
3. 选择一条来源,核对完整正文、来源、工程指纹和到期时间;展开“查看完整来源修订与摘要”可看所有纳入确认摘要的字段。勾选核对后,可以提交候选或确认并共享。候选不会进入后续模型请求;确认针对所显示的确切修订,旧页面不能确认服务端的新修订。
|
||||
4. 核对该来源的同步回执。只有已确认、来源仍匹配、未到期且收到提交回执的条目,才可能成为后续请求的远端来源。本地记忆和远端来源分开记录。
|
||||
|
||||
“每次选入的桥记忆条数”只控制这一可选来源的整条选择,不限制用户消息、本地记忆原文或模型能力。桥协议不能接收的材料明确不入桥,本地原文保留;工作台不会截短材料后假称完整同步。
|
||||
|
||||
## 回执与异常
|
||||
|
||||
任务记录 → 请求上下文 → Mnemo 记忆来源,可查看本次 Run、请求计划、来源序号和远端修订、召回摘要、实际选入摘要,以及每个物理请求的核对和发送状态。桌面可打开完整召回、所选来源和发送前核对工件;伴随手机页显示同次项目授权范围内的摘要和回执。
|
||||
|
||||
每次物理尝试重新核对来源并校验快照;召回两条、选入一条时两份摘要和 refs 分别保留。来源变化、到期、撤权、服务错误或校验失败时省略远端记忆,原任务继续。已经发送的请求不会因桥错误重发;发送前发现选择失效时保留旧计划,再准备一次不含远端记忆的原任务。
|
||||
|
||||
同步结果未知时使用“查询原键”,核对已持久保存的原同步键。未明确核对前不会换键重送。模型请求已接收但响应断线仍按原账本记为 unknown;桥接收器的记录不能把模型账本补签为成功,也不补造费用。
|
||||
|
||||
可撤回共享或永久删除桥来源。永久删除保留墓碑,清除桥中可恢复的同步包正文;本地原记忆与过去已发送的请求工件保留。续期、同目录旧备份恢复和更高序号不能复活已知墓碑。该操作不宣称已删除任何外部系统的历史副本。
|
||||
|
||||
## 已知边界
|
||||
|
||||
核对证明来源声明和当时的内容一致,不认证记忆结论正确,也不授予工程执行权限。跨系统 validate 和模型发送无法保证原子性。当前工程验收采用新本机合成项目及受控 HTTP 接收器;真实生产 Mnemo 绑定、真实模型质量、长期 7/30/90 天效果、物理手机和 Purge 收益分别待验。记忆桥已随 1.25.13 正式发行并激活,默认关闭;见[发行验收](QA-MNEMO-RELEASE-1.25.13.md)。连接实际 Mnemo 仍需按上方入口填写该项目在 Mnemo 已授权的地址、绑定和凭据。
|
||||
@@ -0,0 +1,18 @@
|
||||
# M2 原生缓存与状态来源
|
||||
|
||||
2026-11-07,沿 `bdfeb29c` 接续 I09/I11。来源清单与正文 SHA 在 [SOURCE.json](../tests/fixtures/native-cache-state/SOURCE.json)。OpenAI 两篇完整页面复用本批已取得的官方 Docs 原件,Claude/DeepSeek 六篇复用 I15 的原正文,并逐 SHA 只读核对。接续时再次检索并打开 OpenAI 当前页面;新查询不覆盖先前原件或失败记录。
|
||||
|
||||
| 官方来源 | 本批采用范围 |
|
||||
| --- | --- |
|
||||
| [OpenAI conversation state](https://developers.openai.com/api/docs/guides/conversation-state) | 手动历史保留完整 output 数组、encrypted reasoning 与 assistant phase;不把仅有 response ID 当完整本地状态 |
|
||||
| [OpenAI prompt caching](https://developers.openai.com/api/docs/guides/prompt-caching) | 显式 content breakpoint、缓存 key/TTL/retention 独立透传;会话持续存在或前缀相同不保证命中 |
|
||||
| [Claude prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching) | 系统、消息、工具和工具结果的 cache_control、各块 TTL 及原生元数据保持原位置和顺序 |
|
||||
| [DeepSeek thinking](https://api-docs.deepseek.com/guides/thinking_mode) / [chat](https://api-docs.deepseek.com/api/create-chat-completion) | 同供应商范围内历史 assistant reasoning_content 完整保留;工具身份与原结果不重放 |
|
||||
| [DeepSeek cache](https://api-docs.deepseek.com/guides/kv_cache) | 缓存是上游行为,用原 usage 区分读写、未报告和 unknown |
|
||||
| [DeepSeek Responses](https://api-docs.deepseek.com/api/create-response.md) / [Messages](https://api-docs.deepseek.com/api/create-message.md) | 兼容接口原字段与原生状态分开,不按协议名称推断账户或供应商状态可互换 |
|
||||
|
||||
本地状态绑定记录实际使用的协议、请求 URL、型号、连接、凭据修订和有效请求头模板。它用于防止不透明历史在不同供应商范围误用,不认证缓存命中、模型身份或供应商兼容。普通生成参数和连接修订不等于凭据变更。旧无完整身份的状态保留原件和适用原因;常规文本与原工具调用可继续进入新请求。
|
||||
|
||||
新模型池默认会话保持;已有池沿原 Run 范围或关闭设置。用户可选择关闭、Run、Task、会话;手动选模和当前可用成员优先。跨 Run 亲和仅从已完成响应保存,诊断独立、未知请求不补成成功,不新增执行审批或 token 额度。
|
||||
|
||||
该来源核对及合成验证不能证明全球实际兼容、真实缓存收益、重复读取普遍改善或 Purge 收益。W/AD/AE/AG 的网络与 unknown、R/W v2 真实排除边界和原工程结果保持。
|
||||
@@ -1,5 +1,31 @@
|
||||
# 1.25 计划:任务要求与交付复核
|
||||
|
||||
2026-11-12 1.25.14重启Key修复已正式交付:普通配置默认Windows用户加密/原子保存,正式保留PID升级后实际新PID29884恢复七条Key;全部原行/139导出/5583工件、选择/修订/密文与420请求保持。54相关、独立源码程序、实际两端/隔离Inno及715成员封存通过,见[验收](QA-CREDENTIAL-PERSISTENCE-1.25.14.md)。当前限定提交/普通GitAtlas增量推送,后续M8另冻;旧真实模型、Claude403、网络/unknown/重复读/预留和Purge未验不改签。
|
||||
|
||||
2026-11-12 本轮验收推送完成:原M4收尾c8c36c27和修后Claude5ca65a45远端/fetch一致,见[回执](gitatlas-claude-proxy-correction-push.json)。回执/完成接续另提交核最终HEAD,541/41封存保持。当前上游403仍须供应商普通Messages权限,原两请求不追加;下一M8另冻,历史unknown不改签。
|
||||
|
||||
2026-11-12 Claude修后独立两预检结束,目录完整14/选定在内,两请求上游403 permission_error、分组仅允许Claude Code客户端;usage/费用/响应身份未知,工具执行0。[验收](QA-CLAUDE-PROXY-CORRECTION-2026-11-12.md)与41成员新封存/原状态核对通过;当前七连接/人类更正与原M4的40请求/541清单保持。普通提交推送收尾,不追加请求/重跑工程/Purge或回退配置;供应商权限须另解决,历史unknown不改签。
|
||||
|
||||
2026-11-12 M4验收e132601b与前六条已普通增量推送且远端/fetch逐SHA一致,见[回执](gitatlas-m4-live-push.json)。原40请求/541封存保持,1.25.13未变;完成回执另提交核最终HEAD。下一人类修后Claude至多两新预检独立登记,不改旧失败,不重跑工程/Purge。
|
||||
|
||||
2026-11-12 M4最终审计和封存通过:40请求629106报告tokens/0unknown,三新工程原验收与Git完整,541成员134884464字节逐SHA核对;见[最终验收](QA-M4-LIVE-BASELINE.md)。收尾提交和既有GitAtlas普通推送进行中,产品未变不重打/重启1.25.13。封存后人类修正Claude代理Key,另冻最多两次原文本/工具预检,不回退配置、不改旧mismatch/目录缺项或续旧样本;三工程/Purge不重跑。M8收益实验另冻,全部历史unknown和Mnemo生产语义/手机/长期边界保持。下方“下一封存”按原时点理解。
|
||||
|
||||
2026-11-11 M4三类新工程全部过:Python10请求176083tokens/原10pytest/编译/三源码Git与最终报告,本批累计40请求629106tokens/0unknown;全部原数据及九新捕获导出精确核对。见[Python验收](QA-M4-PYTHON.md)。Claude代理身份失败、外部你好和历史unknown保持;最终封存提交推送后才另冻Purge,1.25.13程序未改不重发行。
|
||||
|
||||
2026-11-11 M4新React14请求271333tokens/0unknown,原九测试/typecheck/build和实际桌面/390px、四文件Git及终端来源最终报告已过,原状态审计一次通过;见[React验收](QA-M4-REACT.md)。下一独立Python,最终封存提交推送;本批累积30请求453023tokens,Claude身份失败、外部你好和历史未验项保持,Purge收益仍须另冻。下方Node下一React按原时点理解。
|
||||
|
||||
2026-11-11 M4真实兼容与Node阶段已验收:三线六预检/1227tokens,DeepSeek官方与GPT Responses代理通过;Claude代理请求claude-opus-5-5却自报gpt-6.1-sol,目录也未列选定型号,失败不改签。DeepSeek新Node自主十请求180463tokens/0unknown、原9/9/build/三源码Git与内置completed通过。原127会话/5244工件及每行保持,外部新增一会话/一请求和两审计失败单列,见[验收](QA-M4-COMPATIBILITY-NODE.md)。下一独立React、Python,Purge另冻可比实验;真实语义/缓存因果/重复读普遍改善与历史未知继续保留,未重打或重启1.25.13。
|
||||
|
||||
2026-11-11 当前:M3+B4b已正式交付1.25.13/source35a6d6d9,独立1399输入/30编译、186相关/4打包、实际安装11/11/跨版7/7及Mnemo后端24/24/两端各4组通过。唯一正式Setup351789ms、原PID63892/全部原行/127导出原字段/5244工件保持,正式四图和final核对通过;[验收](QA-MNEMO-RELEASE-1.25.13.md)、[身份](mnemo-1.25.13-release-identity.json)。源码/验收1e309700已普通增量推送且远端/fetch逐SHA一致,见[回执](gitatlas-mnemo-1.25.13-push.json);下一M4新真实基线;真实Mnemo生产绑定/语义及Purge收益未验证,历史异常完整保留。下方未提交/未发行按旧时点理解。
|
||||
|
||||
2026-11-08接续:M3隔离迁移/恢复源码后,默认关闭的B4b Mnemo单项目记忆桥已完成25项原生工程证据、最后相关104/104、916输入/30编译及网页/实际Electron/伴随HTTPS验收,见[报告](QA-B4B-MNEMO-NATIVE.md)。M3+B4b仍为未提交工作树,统一发行和正式升级未做;真实模型、长期、重复读/R-W边界及Purge收益unknown保持,历史异常不改写。下一M3最终产品commit/版本/源码程序包/InnoSetup及独立安装验收,下方当前顺序按历史时点理解。
|
||||
|
||||
I27/I28推送完成:源码与验收44b6a09b已普通增量推送至既有GitAtlas分支,远端HEAD和重新fetch的14个提交逐SHA一致,见[回执](gitatlas-m2-ci-push.json)。回执/完成接续最后另提交并再次核对;最终4a03ec52干净1222/1222及7595封存原件保持,正式未部署/重启,runner/job/真实供应商及历史收益unknown不代签。下一按M3-RELEASE-PLAN.md推进实际schema2迁移与新结果后的兼容回退;下方进行中说明按原时点理解。
|
||||
|
||||
2026-11-08 I27/I28源码和本地干净CI完成:最终4a03ec52、146文件1222/1222、1376输入/29编译及实际两端7组4图实看通过,见[CI验收](QA-M2-CI-OFFICIAL-CONTRACTS.md)。原失败与历史unknown完整保留,本地通过不代签GitAtlas runner/job或真实供应商/模型/Purge收益。7595原件已封存,文档/授权普通增量推送收尾;下一[M3迁移与回退](M3-RELEASE-PLAN.md)和统一发行,正式服务未部署/重启,0真实模型/Key。下方已交付版本与旧计划按其时点理解。
|
||||
|
||||
M2第五批I09/I11源码71f18055/9e1feeeb/ee97aaae已完成:原生缓存块与跨Run状态身份、四范围连接保持、纯Chat混池实际状态误送修复。最后16/16,首127/129与两文件26/26按SHA分别保留,最终880输入/29编译/74runtime、网页/实际Electron各8组/10图、原metadata/cipher/snapshot/history exact,5126原件封存,见[验收](QA-M2-NATIVE-CACHE-STATE.md)。源码与验收1a897576已普通增量推送、远端逐SHA一致,见[回执](gitatlas-m2-native-cache-state-push.json);完成接续最后另提交核对。下一I27/I28 CI,再M3实际schema2增量迁移及新处理器结果后回退,不能删除新账本或去掉逐表全等守卫。默认开放/透传/持久权限与W/AD/AE/AG、重复读/R-W边界/Purge收益unknown保持,正式未写入/重启。下方旧接续按原时点理解。
|
||||
|
||||
M2第三批I15源码ba4b4491/b5aaf1f7与来源0cbf14a2完成:26记录/13精确型号/17来源,最终858输入/29编译、网页与实际Electron各6组/7图、原metadata/密文/存储快照exact,5688原件封存,见[注册表验收](QA-M2-OFFICIAL-REGISTRY.md)。首候选62/62的未变后端范围和原失败分开记录,文档/授权普通增量推送收尾中。下一I14版本/头/指纹与缓存块/亲和/状态、CI,M2全阶段及M3实际迁移/回退未完成。本批0真实模型/Key/正式写入,未部署/重启或新health查询,W/AD/AE/AG异常/unknown/重复读取与Purge收益unknown保持;下方旧当前按历史时点。
|
||||
|
||||
当前M2第二批I13源码da2e6313完成完整请求账本与逐请求增量存储,原metadata/初始记录/unknown不改。最终build-4独立39/39、网页与实际Electron各5组/7图/23页通过;830输入/29编译与candidate-app-5一致,4328原件只读验封通过,见[验收](QA-M2-REQUEST-LEDGER.md)和[机器记录](m2-request-ledger-2026-11-07.json)。相关102/102发生在最后公开数据库版本修正前,不能混称最后一轮;首失败完整保持。文档与授权GitAtlas普通增量推送收尾中。下一I15官方注册表和I09/I11/I12/I14缓存块/状态/指纹/版本、I27/I28 CI,M3统一发行须实际核验schema2安装增量及新处理器回退,不去掉数据保持守卫。本批未正式部署/重启或新health复查,历史真实异常/重复读取/Purge收益unknown保持;下方旧当前按历史时点理解。
|
||||
|
||||
@@ -0,0 +1,88 @@
|
||||
# B4b Mnemo 单项目记忆桥:原生工程验收
|
||||
|
||||
2026-11-08。默认关闭的 Mnemo 单项目记忆桥已实现,25项工程案例具备证据:原生后端 24/24,N25网页和实际隐藏 Electron 的桌面/窄屏及伴随HTTPS回执路径通过。最后相关回归104/104,详见[机器记录](b4b-mnemo-native-2026-11-08.json)。
|
||||
|
||||
基线为 `f7705d96347cb19ea4ceab745bb2408061ef7621`,分支 `feat/task-agreement-1.25`;M3 和本轮实现仍在未提交工作树。当前没有新的产品 commit、发行编号、安装包或正式部署。按本轮接续约束,不自动提交、推送、发布、安装或重启正式服务。
|
||||
|
||||
## 完成的产品行为
|
||||
|
||||
- [协议客户端](../src/core/memory/mnemo-protocol.mjs)按七份 V1 契约独立计算定序摘要,严格核对来源、绑定、主体、私有范围、快照期限、完整正文和 refs;HTTP 拒绝保留原错误码/状态和诊断正文,真实传输失败另记。未使用重定向或旧远端缓存补签结果。
|
||||
- [项目桥](../src/core/memory/mnemo-bridge.mjs)保留持久生产者 UUID、每项目独立默认关闭设置、原同步键和递增来源序号。本次凭据只留宿主内存,重启/恢复不恢复授权。确认基于同一份完整 review 和摘要,候选不自动共享;withdraw/delete、墓碑及同目录旧 metadata 恢复分别处理。
|
||||
- 同一 Store 的所有项目共享串行锁,防止跨项目整份 metadata 覆盖。永久删除擦除桥可恢复的同步包正文和历史响应副本,保留原键/摘要/回执/墓碑;本地原记忆和原已发送工件保持。
|
||||
- [请求规划](../src/core/context/request-plan.mjs)将授权远端记忆与本地 source-checked 记忆分开,保留完整召回和整条实际选入两份工件、摘要及 refs。桥容量只控制可选来源,不限制原模型、原本地全文或用户历史。
|
||||
- [运行时](../src/server/runtime.mjs)在每个真实物理尝试前重新核对 manifest 并 validate。发送前失效时保存失败旧计划,只准备一次省略远端记忆的原任务;已发送请求不因此重放。原供应商用量/unknown 和请求账本仍为计费权威。
|
||||
- [界面](../src/client/MnemoBridgePanel.tsx)沿用项目记忆入口与 Orven 风格,提供连接、完整来源核对、候选/确认/撤回/删除及原键查询。[任务回执](../src/client/MnemoPlanReceipt.tsx)显示 Run/计划、来源序号/修订、召回及选入摘要、逐物理尝试状态和原工件入口;伴随手机页只投影任务范围内的摘要和回执。
|
||||
|
||||
保持 unrestricted/passthrough/preserve、持久 ask/auto/full;没有增加工程逐步审批或恢复本地模型额度。使用路径见[说明](MNEMO-PROJECT-MEMORY-BRIDGE.md)。
|
||||
|
||||
## 输入、构建与测试
|
||||
|
||||
独立证据根为 `.qa/b4b-native-2026-11-08/`。最初 24 个输入包含 after-M3 接缝、七契约及八宿主输入,逐 SHA 与来源 manifest 一致。首冻结脚本未解码中文 file URL,原输入和原回执已整体保留在 `initial-url-path-mistake/`;修正后的第二轮在 planner 已变化时按预期拒绝,不能把其部分输出当完整首冻结。
|
||||
|
||||
Mnemo `PROTOCOL.md` SHA256 为 `7689cf34ab14d4c5608aa87ee95ea4fd4b33bb65eac46fe44be6d916d3d34379`,向量 SHA256 为 `2ed45ff142981131b9442c76ebddc41625ee0270b4fc95a15fea5acb90bad7e0`。后续 v3 仅更新独立核对器/测试/说明,另冻 `supplement-v3/`,不覆盖原 v2 八文件快照,实际宿主实现及七 V1 契约未变。
|
||||
|
||||
最终 `build-4/`:固定 Node v24.22.1,916 个程序/测试/构建输入,30 个编译文件;根工作树与独立 source-build 的输入、编译及 candidate-app 字节一致。使用现有锁定依赖树的真实副本,本轮没有重新执行 clean npm ci;M3 的独立 clean npm ci 证据保持其原身份。
|
||||
|
||||
| 轮次 | 实际结果 | 适用范围 |
|
||||
| --- | --- | --- |
|
||||
| protocol-second / bridge-first | 5/5、6/6 | 最后界面复核修正前的协议和桥专项 |
|
||||
| related-first / related-final | 41/41、103/103 | 首相关及复核修正前的较广相关轮 |
|
||||
| review-fixes | 12/12 | 非本地前缀 HTTP401、同源完整 review/digest 和原桥边界 |
|
||||
| related-after-review | 103/104 | reasoning-only 超时用例的接收端请求数 0,预期 1;失败保留 |
|
||||
| stream-timeout-isolated | 1/1 | 原失败用例单独自然通过,未改阈值;不据此证明首失败原因 |
|
||||
| related-last-isolated | 103/104 | 相同源码、无其他原生驱动运行,但仍用Node默认并发,同一夹具接收前超时 |
|
||||
| related-standard-2 | **104/104,179360.3584ms** | 恢复仓库package/CI原有并发度2后,完整七文件单轮自然结束,0取消/跳过 |
|
||||
|
||||
七文件为 mnemo-protocol、mnemo-bridge、request-plan、model-center、model-stream、companion、reminder-runtime;不是全仓单轮测试。测试、源码和编译身份另有最终核对。
|
||||
|
||||
## 25 项工程案例
|
||||
|
||||
最后后端轮 `native-after-review/result.json`:**N01–N24 单轮 24/24**,30 个新 Run、31 次受控物理 HTTP 接收。每个案例的原 Run、计划、账本与工件见机器矩阵,N25 由下述两端实际界面补齐。
|
||||
|
||||
| 案例 | 实测行为 |
|
||||
| --- | --- |
|
||||
| N01–N04 | 默认关闭零桥请求、确认后新会话取用、本地 current 不代替确认、候选不入请求 |
|
||||
| N05–N09 | 更正、归档/未核对、来源期限、生产者永久删除及 Mnemo 本机删除;旧已发字节保持 |
|
||||
| N10–N15 | recall 后撤权、续期与旧 grant、实际等待31秒、召回后正文/指纹变化、跨项目绑定拒绝 |
|
||||
| N16–N18 | 事务提交后断线为 unknown,仅查询原键;换正文/低序号拒绝;正文摘要/validate 响应篡改拒绝 |
|
||||
| N19–N20 | 实际召回2条、整条选入1条,两摘要/refs分记;超桥协议上限不截短原本地全文,原请求不添加 max_tokens |
|
||||
| N21 | 请求实际接收后断线,原账本和桥回执继续 unknown,无伪造费用/成功 |
|
||||
| N22 | 显式合成有限两成员策略,JSON503后 SSE 成功;两物理尝试各自 validate、ID/SHA和账本,不重复计费 |
|
||||
| N23–N24 | 同目录旧 metadata 恢复不加载授权、不复活已知墓碑/桥包正文;候选经人工确认后递增为序号2 |
|
||||
| N25 | 网页、真实隐藏 Electron、实际伴随 HTTPS 窄屏中的同次来源/修订/失效回执与原页面流程 |
|
||||
|
||||
`native-offline-audit.json`另用只读 SQLite 核对 current/initial 账本摘要,31次原字节与请求工件完全一致;Run/项目/会话/Task、所选全文工件和逐尝试验证对应,336份原工件 SHA保持。该审计新增模型/HTTP请求0,原 SQLite文件 SHA前后相同。N21的实际接收证据不能把原宿主 unknown 补签成功。
|
||||
|
||||
Mnemo侧另有独立复核,原回执复制在`independent-after-review/`:逐项24/24、37个账本尝试(31实际接收、6发送前拒绝)、336原工件及916输入/30编译逐字节核对;最终宿主观测SHA为`12ef67364502cae9a92c3182af83806dd4b5517d85c1eabe3f6905ff3235033a`。其物理核对器明确保留`all25CasesVerified:false`,因为N25不是该后端核对器的范围;工作台两端/伴随界面证据另列,不改写来源方原回执或把它称为真实质量认证。
|
||||
|
||||
## 网页、隐藏桌面和手机
|
||||
|
||||
- 网页最后轮:`ui-web-third/result.json`。
|
||||
- 实际隐藏 Electron:`ui-electron-first/result.json`,使用 candidate-app 的真实 Electron runtime 和原桌面启动器,窗口保持隐藏。
|
||||
|
||||
两端各4组主要流程、各10张图、各2次新本机合成模型协议请求。覆盖默认关闭、真实表单连接、候选及确切完整确认、刷新、1280/390、原全文工件、实际物理ID/SHA、同Run伴随回执、撤权HTTP401和0远端选入。JS错误、页面外部网络请求均0。
|
||||
|
||||
伴随网关为实际 HTTPS,浏览器经只服务本轮的 loopback relay连接;relay用本轮公开CA验证证书链/IP,不关闭TLS验证,不安装系统证书。真实新设备配对及 observe 项目权限有效;页面不包含凭据、宿主路径或完整原模型历史。物理 iPhone/Safari 未测试。
|
||||
|
||||
最终20张图中实看8张:两端完整修订、网页与桌面的390物理回执、桌面390确认、网页伴随同Run、网页HTTP401失效、桌面伴随HTTP401失效。所见内容与保存回执一致,无意外横向溢出;未把截图存在当作逐张人工审查。
|
||||
|
||||
## 原失败与修正保留
|
||||
|
||||
- 首协议 fixture 路径 ENOENT及误读向量键;修正后保留旧日志和冻结向量。
|
||||
- `native-first`少 `/v1`导致全404,其7个负向表面通过也不采纳;`native-second`实际接收但长查询词项覆盖不足、合法空召回,仅10/22,不能补签正向取用。最终改用确切合成来源标题作为实际用户主题,未改 Mnemo检索。
|
||||
- `native-selection-probe`的两成员策略误写 priority,改用产品已有 round-robin后独立通过,旧失败保留。
|
||||
- 首构建误用系统Node16、独立构建漏 mobile.html、build-2目录联接导致 esbuild相对依赖路径真实变化,均保留,未归一化编译产物凑SHA。build-3和最后build-4使用真实依赖副本后exact。
|
||||
- 网页首轮缩为390后任务侧栏收起,驱动没重新打开真实入口;原截图/DOM/请求保持。第二轮发现产品按 PROJECT_命名空间误分类上游HTTP拒绝。最后改专用错误类型,保留原HTTP码/状态;完整review改由同一服务器快照产出,避免父层旧正文与新摘要混用。两项修正后重新构建、24案例和两端复验。
|
||||
- 两轮相关104项中的原超时失败保留。后续发现验证命令漏了仓库固定并发度2,改用原标准后104/104;单项通过不解释首失败的确定原因,未更改产品/测试超时阈值,也未删除用例。
|
||||
|
||||
## 清理、边界与下一步
|
||||
|
||||
本轮四个合成 Mnemo宿主均通过本轮受保护接口正常关闭。最后实例 `60f59342-6965-4451-8c7c-8dd7840559e3`的59752监听已关闭,private凭据文件由宿主删除,所属Node进程自然exit0;最终 `host-observation.json`、canonical.db及69份物理正文保留在`host-final/`。四个界面候选宿主的保存PID和记录端口均不在,所属Electron0,见`candidate-process-cleanup.json`。不把这项清理称正式服务重启或正式health检查。
|
||||
|
||||
M3旧封存清单SHA仍为 `78ce1df774be24d37c8a3e55d943a24f9cab2dd5e955e1743e631723f0bddd00`,旧根未追加。未访问正式4318、读取真实Key、请求真实模型、重放旧工程、修改Mnemo源码或向来源线程发信,也未委派。
|
||||
|
||||
本轮证据按逐文件SHA清单封存,见`.qa/b4b-native-2026-11-08/MANIFEST.json`;清单自身摘要、逐文件离线复核和总量写在根外`.qa/b4b-finalization-2026-11-08.json`,封存后不再向该根追加。旧失败构建中的目录联接只封存链接声明,不跟随它读取外部目录;最终build-4使用真实依赖副本。
|
||||
|
||||
这轮证明工程协议、原生绑定和界面回执路径,不证明全球真实供应商兼容、模型真实自主交付/语义收益、重复读取普遍改善、R/W v2真机剔除边界、Purge收益或长期7/30/90天效果;W/AD/AE/AG原成功与网络/unknown均保持原结论。跨系统校验与模型发送仍无原子性保证。
|
||||
|
||||
下一步是M3统一发行的最终产品commit/版本/源码程序包/InnoSetup及独立安装验收;正式升级和新的真实官方/代理样本仍按各自计划推进,本轮不自动执行。
|
||||
@@ -0,0 +1,36 @@
|
||||
# 人类修正 Claude 代理 Key 后的独立兼容复核
|
||||
|
||||
目录已通过,模型请求仍未通过:新目录完整返回14型号,包含 `claude-opus-5-5`;原文本和工具两项新请求均被上游 HTTP403 拒绝,原文为“该分组仅允许使用 Claude Code 官方客户端。 This group is restricted to Claude Code clients (/v1/messages only)”。两次没有模型响应型号、SSE或usage;用量和费用均未知,未追加请求或执行工具。
|
||||
|
||||
机器记录:[claude-proxy-correction-2026-11-12.json](claude-proxy-correction-2026-11-12.json);清单核对:[claude-proxy-correction-seal-2026-11-12.json](claude-proxy-correction-seal-2026-11-12.json);[冻结计划](M4-CLAUDE-CORRECTION-PLAN-2026-11-12.md)。原M4[最终验收](QA-M4-LIVE-BASELINE.md)和四十请求/541清单保持,不能将本次两请求并入原封存或改签旧失败。
|
||||
|
||||
驱动 `97f132d3` 和验收 `5ca65a45` 已普通增量推送到既有 GitAtlas 分支,远端和重新fetch逐SHA一致,见[推送回执](gitatlas-claude-proxy-correction-push.json)。回执和完成接续另提交后再核最终HEAD;两根原件继续本地封存。
|
||||
|
||||
## 实际请求与结果
|
||||
|
||||
正式仍1.25.13/PID63892、原instance,unrestricted/passthrough/full,产品累计/步数均0。选定连接 `9ae348a5-d0a6-42ed-8664-0da5f9effe34` 为 Messages/claude-opus-5-5,连接revision8→9、credentialRevision1→2;config130保持,registry182→186。另新增Fable连接只记录公开身份,未选择/测试或推断操作者。人类报告此前Key写错的原因没有独立证明。
|
||||
|
||||
两请求均实际发送 `https://api.apikey.fan/v1/messages`、`anthropic-version: 2023-06-01`、原配置的x-api-key鉴权、`stream:true` 和官方精确型号后备 `max_tokens:128000`。字段及原body SHA已核对,未保存新参数/身份策略或主动本地额度。上游返回完整188字节JSON错误,type=`permission_error`,没有重定向或重试;SSE、工具能力与模型身份未取得证据。目录列出型号不能代替模型调用权限。
|
||||
|
||||
| 新用例 | Task / Run | 本地物理尝试 | 上游请求ID |
|
||||
| --- | --- | --- | --- |
|
||||
| 文本 | 22f44c69 / ba7da2e3 | 81f1bf6d-0cae-4a7a-b482-3b1381f4a606 | 02acb49c-9ca1-4fdc-9a4c-c45c71d4b000 |
|
||||
| 工具观察 | 5decb3cb / b0468e4b | 9efaf829-859b-482a-a401-1c02531cd84a | 65157030-cf66-414a-bb14-8db6f4b70acc |
|
||||
|
||||
完整project/session/Task/Run、规划requestId和物理wireAttemptId均在机器记录,两个用例各一新请求。原采集器可选requestKey字段为空,实际请求键从原账本requestId记录,没有补造或重派。原文件后缀按请求流式设置记为 `.sse`,其原字节实际为HTTP403 JSON错误,响应类型/完整性按实际记录,未将后缀当SSE证明。
|
||||
|
||||
两请求均 `usageStatus:unknown`,实际tokens和费用为null;已报告tokens的空集合合计0仅是账本合计,不能声称零消费或零计费。HTTP错误原文完整接收与模型用量未知分别成立。工具执行0,Node/React/Python与Purge均未重跑。
|
||||
|
||||
## 保持核对与封存
|
||||
|
||||
驱动 `97f132d3` 在真实请求前通过10项离线判断(8合成边界、2旧型号不符原回执只读复核),没有把它计为真实兼容。实际结束后 `audit-44ea125a-b46f-4355-bbac-5026b017c2e7` 一次通过:冻结前58298行/全部原字段及schema、137导出、418请求、5576工件及当前七连接/池/选择/配置/凭据行SHA保持。
|
||||
|
||||
仅新增自有一个项目/两会话/两请求,36events及outbox、16records、7新工件;项目前插+1、事件序列+36/请求序列+2明确核对。没有恢复旧Key或人类更正前注册表,PID/instance与空闲保持,9229关闭;未读、复制、解密或重存Key,产品程序未改,不重发行或重启。
|
||||
|
||||
新根 `.qa/m4-claude-correction-2026-11-12/` 封存41成员/36284086字节,MANIFEST SHA256 `9b8a1cfcc4a8842e5c751990fdfc4a274831bf4dcafe73c557514c992490f2e5`,全部成员逐bytes/SHA验证通过。原封存回执bytes表示成员总字节,独立公开回执另列manifest文件字节与原回执引用。原M4的541成员也再次只读核对通过,两根不再追加,原件不进Git。
|
||||
|
||||
## 回顾与下一步
|
||||
|
||||
当前明确阻塞是上游分组对客户端的权限要求。需要供应商开放该Key分组给标准Messages API客户端,或使用允许此客户端的分组;本轮不冒充Claude Code客户端或尝试更多请求。两次兼容复核已按授权范围结束,模型身份仍未验证,型号匹配本来也不认证权重。
|
||||
|
||||
历史两次自报GPT和旧目录缺项不改签,新403单列;网络/输出截断/重复读/R-W真机预留边界保持。Purge收益、Mnemo生产绑定与语义、物理手机/长期、其他模型能力仍依各自新实验,下一M8原版Purge另冻,不能以目录成功或已通过三工程替代对照收益。
|
||||
@@ -0,0 +1,40 @@
|
||||
# 1.25.14 模型凭据重启恢复验收
|
||||
|
||||
2026-11-12。用户反馈正常填写的 Key 在重启后全部丢失。本轮以 `211117e0` / 已安装 1.25.13 为来源,已修复默认保存流程、发行并实际重启到1.25.14;七条现有Key自动恢复,原数据保持。源码与验收已普通增量推送,远端与重新fetch逐SHA一致;本回执与完成接续最后提交推送并核对最终HEAD。
|
||||
|
||||
## 修复结果
|
||||
|
||||
Windows 上手动填写、上游目录选择及确认导入默认用 CurrentUser DPAPI 加密保存,在新宿主中自动加载。显式 `persistKey=false` / `persistCredentials=false` 才保留临时方式。连接和密文共同提交,加密或数据库失败时恢复原配置、原 Key 和原密文,表单输入保留。
|
||||
|
||||
普通名称、并发及原生参数编辑不重复要求 Key、不重加密旧块。显式清除或改用临时 Key 时删除旧保存副本,防止重启恢复旧值。目录新增型号共用源连接;自动同步保持来源原保存选择,旧来源不自动变更。旧配置接口也原子保存新 Key。地址、型号或鉴权变化仍保留原保存块和具体不匹配状态,界面提示更新本机保存。
|
||||
|
||||
已经退出且没有保存副本的旧 Key 无法恢复;首候选包后公开核对到用户已补填七条 Key,原 PID14388 保持。本轮增加明确的“保存当前 Key”操作,工作台在原进程内加密保存,调用方不接收 Key、原对象和凭据修订保持。先验合成凭据,再在保留原宿主的升级后处理当前七条连接,全部成功后才正式重启。工具不读取、复制或回传真实 Key。
|
||||
|
||||
## 源码与进程证据
|
||||
|
||||
- 修改前新增三项复现全部失败,原日志 `01-before-default-tests.log` 保留。
|
||||
- 首关联轮 35/36,凭据从共用转为独立时的版本回归已修,独立复验 1/1。随后 48/52 中两项旧临时默认断言改为显式临时或验证新的默认恢复;另两项虚拟数据目录失败在独立 `211117e0` 来源上同样复现。
|
||||
- 复用原无符号链接校验与物理根解析后,52/52 相关检查通过,覆盖默认加密、真实 DPAPI、独立 Provider 恢复、临时选择、清除、导入同步、事务回退、目录、HTTP 配置和热升级。随后补旧配置的正确显示名,定向 HTTP/旧配置/加密失败 3/3 和最终构建通过。
|
||||
- 初次网页验收四步骤已过,配置格式选择器的严格文本匹配失败原件保留;修正验收选择器后另根完成 8 步。
|
||||
- 实际 Edge:手动保存、留空编辑、临时选择、上游两型号共源、导入、退出及新进程恢复、恢复后鉴权、390px 保存失败保留输入通过,三图已实际查看。截图位置之后调整为直接展示目标凭据字段,最终程序包继续复验。
|
||||
- 四个加密保存来源重启后恢复为五条可用连接,临时连接没有恢复;密文 SHA 不变、公开数据和 SQLite 检查未包含合成 Key 明文。两宿主均正常退出。只向本机合成接收器发送 1 次模型请求,真实供应商请求 0。
|
||||
- 增加“保存当前 Key”后,`14-current-key-tests.log` 2/2 验证原对象/请求身份保持及磁盘失败原数据回退;`16-source-ui-current/` 实际Edge9步通过,五个保存来源恢复六条连接、临时选项不恢复、两个进程正常退出、三图实际查看。本机合成请求1次,真实供应商0。
|
||||
- 最终相关 `17-related-current-final.log` 54/54,覆盖新增操作、默认保存、旧配置HTTP、目录/共源、事务回退及保留原SQLite/Provider的升级。安装复核仅放行1.25.13→1.25.14四类精确公开文案/默认字段变化,全部原行、导出、密文和工件仍逐项相同。
|
||||
- 最新 `18-formal-status.json` 显示正式七条连接均已有有效DPAPI副本、原PID14388/instance保持、请求420且空闲;正式阶段保留这些原副本,不重复加密。最终安装包及旧版保留进程升级/当前Key保存/新进程恢复继续单独验证。
|
||||
- 最终程序/源码固定为`41e59768`,程序11094 ZIP成员和源码1425成员CRC/逐字节SHA通过;源码1000编译输入独立离线重建30文件一致。`21-package-ui/`网页9步通过,原生启动首超时保留:验收驱动错误地给已打包Electron附加了默认运行时才需要的main.cjs位置参数,未采用既有`desktopLaunchCommand`。仅修正验收驱动,复用已核验的打包入口与隔离环境;先证明隐藏启动,再showInactive获取Windows实际帧,另根复验。未修改程序载荷、版本或正式进程。
|
||||
- 修正驱动后`26-package-ui-final/`最终10步通过:真实随包Node退出/新进程恢复、六条连接恢复/临时项不恢复、恢复后本机鉴权、实际Electron两宽度和关闭窗口保留宿主。五图已实际查看、密文不变、公开/SQLite无合成明文、原宿主全部正常退出,本机模型请求1、真实供应商0。
|
||||
- `.qa/pv-VoJAEu/result.json`真实Inno隔离四组通过(独立AppId):1.25.13新装49878ms;升级到最终1.25.14为82384ms,原PID1496/instance、35000原事件/35000 outbox及所有原行/导出/工件保持,操作回执5468413字节。随后旧内存Key通过内部save-current保存,原凭据/连接修订不变,新PID27912自动恢复并对本机第三次目录鉴权成功、密文不变;卸载保留全部数据。无真实供应商请求或正式服务写入。
|
||||
|
||||
证据根:`.qa/credential-persistence-1.25.14-2026-11-12/`。源码网页成功根 `09-source-ui/`;全部首失败、基线失败及后续日志保留。源码、程序/安装器、实际安装副本和正式服务在以下阶段分别登记,源码包不含用户项目数据。
|
||||
|
||||
## 发行、正式升级与重启
|
||||
|
||||
独立源码ZIP、不可变程序与Setup、全部清单/编译字节、实际网页/原生桌面和隔离跨版恢复已完成。程序与源码来源固定`41e59768`,候选`Agent-Workbench-1.25.14-win-x64-20261112012725511`;验收驱动修正`0415e9d5`及记录另提交,原程序载荷未改变。注册逐项核对最终包、源码、10步UI、跨版四组、CRC/SHA和独立重建来源,未将不同代码合称同一源码提交。
|
||||
|
||||
唯一正式Setup成功,`formal/installer-exit.json`为code0/234945ms,明确传入原物理数据目录。先保留PID14388/instance`5e941ae4-45c0-40db-a1f5-79c29f80a890`及原Provider升级;109项目/139全部会话/9290记录/24211事件/24211 outbox/420请求、全部原表/行/导出和5583工件逐项exact。只新增既定公开默认/文案字段,选择、登记193、连接与凭据修订、七份DPAPI副本SHA保持,当前Key保存操作0。
|
||||
|
||||
确认七份副本有效后,通过所属宿主接口正常退出,再从新安装目录、原物理数据目录和4318启动新PID29884/instance`81cade0e-6306-4035-a128-15e2cd243175`。七条Key全部自动恢复,连接/凭据修订与密文SHA不变,原行/导出/工件/公开模型中心exact、请求仍420。`formal/restart.json`和`formal/final.json`通过,空闲、桌面窗口0、9229关闭;没有重放任务或供应商请求。工具未读取、复制或重加密真实Key。
|
||||
|
||||
证据根715个常规文件/178252120字节已封存并逐成员核对,`MANIFEST.json` SHA256 `031b0da22a2aae37ce017e9e248cc6711041377ecbd1ee083001d1277ae2e1d1`。基线的既有node_modules依赖链接首次被封存拒绝,另记录已核对的引用目标和锁SHA,保留链接且不遍历;全部原失败保持。GitAtlas推送证据在独立`.qa/credential-persistence-1.25.14-push/`,不向已封存根追加文件。限定验收`958f31cd`及前四条已普通增量推送,远端HEAD/重新fetch和五条逐SHA一致,见[推送回执](gitatlas-credential-persistence-1.25.14-push.json);本回执与完成接续最后再提交、推送及核对。没有force、发布项目数据或称远端CI通过。
|
||||
|
||||
本轮不续旧任务、不发起真实供应商请求、不重跑历史样本。原模型交付、网络异常、unknown、重复读取、Claude 403 和 Purge 未验证结论保持。
|
||||
@@ -0,0 +1,49 @@
|
||||
# M2 CI 与官方契约:验收及回顾
|
||||
|
||||
2026-11-08,基线 `470d6948`,最终程序与测试冻结于 `4a03ec52`,分支 `feat/task-agreement-1.25`。承接 [本批计划](M2-CI-PLAN.md) 和 [CI 使用说明](CI.md)。I27/I28 的本地实现和干净 Windows 验收完成;GitAtlas 可用 runner、实际 job 与上传回执仍 unknown,不能把本地结果称远端自动化通过。
|
||||
|
||||
入口从已提交源码重新导出、实际安装锁定依赖、编译并运行全部测试。当前最终结果为 **1222/1222 通过,0 失败/取消/跳过/todo**;网页及隐藏 Electron 的 1280/390 核心入口实际通过,四张最终图逐张实看。正式程序尚未发行或重启。
|
||||
|
||||
## 实际改动
|
||||
|
||||
- 新本地 contract/windows 入口和固定 action commit 的 Gitea 工作流,保留唯一 run、原命令/PID/exit/日志和源码/编译 SHA。缺环境、取消、跳过、todo、零测试或缺完整统计不能改成通过;上传显式包含隐藏 evidence 路径,实际远端上传仍需 job 回执。
|
||||
- 官方四来源清单、26 条记录、版本/字段/弃用快照与 14 个行为回归绑定;新旧快照保持,不把冻结当验证。三个历史套件改用 65 份最小带 SHA/Git blob 的回放输入,原 112 份来源保持;原完整档案、用户项目和凭据不进入源码包。
|
||||
- Windows 使用产品实际 Edge 与锁定 Electron;显式选择公共 Electron 镜像并继续验证官方 checksums。原官方/Chromium下载异常仍保留。Python 新 venv 按七个精确 wheel SHA 安装,不污染外部解释器。
|
||||
- 修复开放模式的工具广告与既有任务执行契约不一致。诊断和无需测试约定保留;未新增执行守卫或本地模型额度。研究 HTTP 夹具采用当前不限/完整历史默认,原五请求/一次分析、Task/Evidence 与诊断只读断言不删。
|
||||
- Python 报告测试通过原下载路由核完整字节长度/SHA后解析;stdlib unittest 场景用真实 venv --without-pip,不省略 CI 自身的 pip/pytest 安装。推理回归区分自定义名称、显式 provider 和精确官方型号,不恢复前缀猜厂商。
|
||||
- 提醒回归按五个实际请求和结束逐阶段观察,每阶段仍有停滞失败及最后状态;未更改产品时长。窄屏导航使用真实可见工具箱,原生截帧等待已加载卡片/可用选择、字体、有限动画及两个 RAF,并保存失败 DOM/PNG。
|
||||
|
||||
## 验证依据
|
||||
|
||||
| 范围 | 实际结果与来源 |
|
||||
| --- | --- |
|
||||
| 最终干净 Windows | `final/run-windows-mw6lllq7-33d6c439`,1222/1222 通过,0 失败/取消/跳过/todo,测试 3392886.1639ms;实际新 npm ci、新 Python QA、锁定 Electron、构建、全部 146 文件 |
|
||||
| 最终源码与编译 | 1376 个导出输入、29 个编译文件前后一致;源码 ZIP SHA `00186b18e1c4a8cf47596bd53201f129cec01cfb796f5dc0b036d7a49ecfa526`,当前已提交程序/测试输入与冻结 commit 一致,后续指南提交分开记录 |
|
||||
| 环境 | Node v24.22.1 / npm 11.19.1 / Python 3.11.15 / pytest 8.4.2 / Playwright 1.64.1 / Electron 45.2.0 / Edge 156.0.4314.58;公共镜像只是二进制下载来源,官方 checksum 未关闭 |
|
||||
| 实际两端 | 7 组、4 图/4 DOM;1280×900 与 390×844,字体及两帧记录、PNG实际尺寸与 SHA、原项目文件保持;页面脚本错误/外部页面请求/模型尝试均 0 |
|
||||
| 清理 | 原宿主/Electron退出;记录端口原结果和结束后再次实收 TCP ECONNREFUSED,不由 health 缺失推断端口关闭 |
|
||||
| 旧五批封存 | 21,297 个声明原件、7,184,046,930 字节及五个 MANIFEST 只读逐 SHA 相符;依赖链接不递归 |
|
||||
|
||||
2,001 条 companion 操作的完整保留/重启回归实际耗时872353.5061ms,4,002 条审计与旧 unknown 保持;记录正确不表示规模性能已优化,M7 的大 metadata 写入性能待办仍保留。本批实际测试来源没有因长时间暂时无新TAP而主动结束,聚合观察只读取自有测试数据。
|
||||
|
||||
此前独立 contract 的 187/187 来自 `880c88c7`;Python/推理专项 23/23、三个实际文件 22/22 与 tool-prefix 4/4 来自各自冻结输入,不能冒称当时已经最终干净全绿。首专项误带不存在的 tool-surface 文件,原命令不改;只承认实际三个文件 22 项,纠正记录单独保留。最终 Windows 按 146 个真实文件逐 SHA 登记。
|
||||
|
||||
## 原失败与回顾
|
||||
|
||||
1. 首 Chromium 下载 625374ms、第二次官方 Electron 552652ms 后,经 PID/父/参数/UTC 归属核对主动结束。原 exit1 和 interrupted=false 保持,另注明外部人为结束,不能改成自然网络退出;HEAD/首范围成功也不当完整下载成功。
|
||||
2. 首完整 Windows 来源 `36c6be4f`:1211 项、1194 通过/17 失败,0 取消/跳过/todo,3534259ms,自然结束,后置桌面未执行。原报告/日志/最终清单不追加;1308 个导出输入及编译后来只读复核一致。
|
||||
3. 17 项包括四项旧原生/品牌/容量/目录断言、八项包外历史回放依赖、五项 Python/推理/提醒/研究问题。旧断言按实际原生和来源契约修正,历史套件只替换必要来源,不删断言或加 skip。
|
||||
4. Python 原预览因报告增长截断;深路径原错明确为 ensurepip 的 setuptools WinError206,UTF-8不能解决路径长度。辅助复验器曾误覆盖项目 Python 导致另三项失败,原27/32保留,修复器与最终 CI 不带该覆盖。
|
||||
5. 研究首因是旧16K夹具读取 Evidence 元数据失败,随后本机模型夹具解析错误和重试才触发工具范围断言,是受控夹具问题;另确有广告/执行不一致的产品缺陷并已修复。提醒原并发超时缺最后状态,不能补造唯一性能根因,原20秒局部通过与新逐阶段验证分别记录。
|
||||
6. 第一轮网页缩至390时侧栏变inert,导航点击超时,未跑Electron。第二轮原报告7组/4图通过,但原生图仍为加载/工具箱,完整视觉验收不接受,原 passed:true 不改。第三轮补帧同步后四图实看通过;最终干净轮再次独立验实际输入和画面。隐藏QA窗口关闭background throttling只影响本次捕获,不改变产品窗口行为。
|
||||
7. 辅助路径/glob、输出截断、记录id和文档锚点错误记在本批计划,拒绝即保留;未借这些错误改变产品守卫或旧事实。Mnemo 七固定原件和五组向量只是独立只读登记,未接通 SDK/服务或回复来源线程。
|
||||
|
||||
## 封存与后续
|
||||
|
||||
新根 `.qa/m2-ci-2026-11-07/` 的证据/来源/日志/导出输入共 7595 份、150586505 字节;MANIFEST SHA `aaeaf694a1cb2444ee8bdfd5ec74ccc768c5659a38fdd6b3be36d6d69abe4951`。执行依赖、缓存、临时环境和链接按清单明确排除并保留磁盘原件,安装/锁/依赖树/二进制身份在原报告;不将这些排除隐瞒成整个目录的全文件封存。封存后本根只读,推送收尾另写新 close 根。
|
||||
|
||||
完整机器记录见 [m2-ci-2026-11-07.json](m2-ci-2026-11-07.json)。源码/验收 `44b6a09b` 已普通 GitAtlas 增量推送,远端HEAD和重新fetch的提交链逐SHA一致,见[推送回执](gitatlas-m2-ci-push.json);回执与完成状态最后另提交并再次核对。无需 force,不以推送成功代替远端 runner/job 成功。
|
||||
|
||||
本批 0 真实模型请求/真实 Key 读取/正式写入,不发行、安装或重启正式服务,不委派或续旧工程,不作正式 health 新声明;仅沿 F 的历史 1.25.12/PID71628 记录。默认 unrestricted/passthrough/preserve 与持久 ask/auto/full 保持。
|
||||
|
||||
下一按 [M3 计划](M3-RELEASE-PLAN.md)实现精确 schema2 迁移和已产生结果后的可用回退,再统一发行;不能删新账本或移除原行/工件保持核对。之后另冻 M4 真实基线。W/AD/AE/AG 网络/unknown、重复读取普遍改善、R/W v2 真机剔除边界和 Purge 收益 unknown 继续保留,全球供应商兼容与后续产品化事项不由本次本机契约代签。
|
||||
@@ -0,0 +1,50 @@
|
||||
# M2 原生缓存块与跨 Run 状态:验收及回顾
|
||||
|
||||
2026-11-07,承接 [本批计划](M2-NATIVE-CACHE-STATE-PLAN.md) 与 [官方优先路线](DEVELOPMENT-ROADMAP-OFFICIAL-FIRST.md)。来源 `71f18055`、核心 `9e1feeeb`、界面 `ee97aaae`,分支 `feat/task-agreement-1.25`。本批完成 I09/I11 源码和隔离验收;I27/I28 CI、M3 发行、真实供应商与收益仍分别待验。
|
||||
|
||||
原生缓存注解和供应商助手状态能按原协议保持,用户可选择会话、任务、运行或关闭连接保持。旧历史与未知状态仍完整保存;连接、型号、凭据或请求范围变化后,用普通文本及原工具调用构造当前请求,保留上游原错误。连接保持、同指纹或前缀稳定不证明上游缓存命中。
|
||||
|
||||
## 实际改动
|
||||
|
||||
- Messages 系统、用户、工具和工具结果保持原生块顺序、缓存注解、独立 TTL 与元数据;Responses 保持完整 output、encrypted reasoning、phase、工具元数据和用户显式缓存参数。默认不添加缓存参数,来源见 [官方原件说明](NATIVE-CACHE-STATE-SOURCES.md)。
|
||||
- 新 schema2 状态绑定实际协议、请求 URL、精确型号、连接、凭据修订和请求头模板。不能解释或不适用的原状态保留原件、原因与 SHA;继续发送普通文本、原工具调用和结果,不执行历史工具。
|
||||
- 修复纯 Chat 混池的实际缺陷:原实现只构造首成员正文,换线只改 model,导致首成员 reasoning_content 进入后备连接。存在状态时逐成员投影,同型号但不同凭据也分别核对。
|
||||
- 新模型池默认会话保持;用户可保存 task/run/none,原池的 run/关闭设置不补写新默认。跨 Run 仅记录完成响应的连接,诊断独立、手选优先;可选亲和存储失败不把成功响应变成失败或重放请求。
|
||||
- 模型中心保存范围,账本显示本次连接状态及沿用/重选依据,上下文逐候选连接显示状态。候选连接不重复相加同一原状态;旧无元数据的 unknown 记录不补造身份或用量。复用 Orven 三配色和现有 React 流程。
|
||||
|
||||
## 验证
|
||||
|
||||
| 范围 | 原证据与结果 | 实际证明 |
|
||||
| --- | --- | --- |
|
||||
| 最后核心专项 | `source-5.log`,16/16;类型检查通过 | Messages/Responses JSON/SSE 原生块、混池状态、停止/同任务恢复、重启、unknown 与可选存储故障 |
|
||||
| 首独立候选 | `candidate-backend-1`,127/129 | 11 相关文件;两项旧亲和假设失败原样保留 |
|
||||
| 准确修正后两文件 | `candidate-backend-2`,26/26 | 跨 Run 权重测试显式选择 run;关闭保持不暗存连接,重新启用后先选择再沿用 |
|
||||
| 最终输入/编译 | `build-5` → `candidate-app-5`,880 输入/29 编译/74 runtime | 300 产品源码、178 后端输入、213 测试输入及全部编译/复制逐 SHA 相符 |
|
||||
| 最终网页 | `web-ui-5`,8 组/10 图,1280/390 | 真实保存重载、自动重选、手选、原 400、账本/上下文及宿主恢复 |
|
||||
| 实际桌面 | `electron-ui-1`,8 组/10 图,1280/390 | Electron 原生隐藏 capturePage、正常关闭后独立宿主重启及原存储恢复 |
|
||||
|
||||
最终 178 后端输入与首候选相同,213 测试输入中 211 未改,两改文件与 26/26 候选相同。因此复用 103 个原通过项并加两文件 26 项,覆盖 129 项;**没有把首轮失败改成全绿,也不宣称最终完整单轮全套运行**。源码与编译身份见 `input-identity-1.json`、`source-git-final-proof.json`,22 个变更冻结输入与提交原字节相符;8 份官方原件与 SOURCE 清单的 SHA/Git 原字节核对通过。
|
||||
|
||||
最终两端各 5 次自有 HTTP POST:3 成功、2 原 400,另保留 1 条原 unknown,最终各 6 条账本。所有原 metadata、DPAPI 密文、存储模型快照和原历史 exact;合成 Claude 凭据与原生设置在独立宿主重启后恢复,实际只执行 1 次 list_files,未重放。以上 10 次是最终两端请求数,不代替本阶段全部合成请求总数。
|
||||
|
||||
三配色实际操作与截图通过,页面错误/外部页面请求均 0,没有文档横向溢出。20 张最终图逐 SHA 核对,8 张实际打开复看;中文、长地址/状态 SHA、保存与原错误可读,被测正文与提示对比度至少 5.46。原横向标签滚动与文档溢出分别核对。
|
||||
|
||||
## 原失败与回顾
|
||||
|
||||
1. source-1 的错误夹具 API、source-2 亲和来源断言、source-3 整正文查找及 source-4 14/15 保留。收窄 source-3 断言后 source-4 确认纯 Chat 混池实际缺陷,修复后 source-5 16/16;不能把产品缺陷全归为测试问题。
|
||||
2. 首候选两失败是旧测试假设:新池默认会话保持不应按三 Run 轮询;关闭保持不应暗存连接供以后启用。测试按其真实目的显式选择 run,并核对关闭不暗存/重新启用首次选择/后续沿用,保留原失败。
|
||||
3. web-ui-1 精确标签匹配失败,实看下拉存在后补 aria-label;web-ui-2 宽屏打开检视再缩至手机触发既有隐藏行为,改为先设窄屏再打开入口。
|
||||
4. web-ui-3 已按失效亲和正确选 B,驱动误期待 A 的修改头;按原 requests 核对自动重选,并新增真实 UI 手选 A 的原头/原错检查。web-ui-4 在手机抽屉打开时点击后方工具箱,被可见检视区拦住;先 Escape 返回后继续,不强行点击后方控件。
|
||||
5. 所有轮次的驱动输入、日志、图和原请求保留。两次错误文档锚点使 patch 整体未应用、只读路径/glob 与输出截断均记录为开发操作错误;文本规范化末轮检查为 0 项。官方原文 6 处尾空白保持原字节,没有修剪来源。
|
||||
|
||||
`final-review.json`、`final-ui-proof.json`、`prior-seals-1.json`、`owned-processes-final.json` 保存回顾、结果与清理依据。I10/I13/I15/I14 四个旧根共 16171 原件及原 MANIFEST SHA 只读复核一致,构建依赖 junction 不递归。自有候选 Node/Electron 进程和记录中的非正式端口监听均 0,未查询正式 4318。
|
||||
|
||||
## 封存与后续
|
||||
|
||||
新根 `.qa/m2-native-cache-state-2026-11-07/` 已封存 5126 份、2008555583 字节。MANIFEST SHA256 `c887364e54fe3f68e5f432155db87517fb8413966f5786156bedceddadbf2217`;五个构建依赖 junction 只登记、不递归。封存后本根只读,推送和后续回执写独立新根。
|
||||
|
||||
完整机器记录:[m2-native-cache-state-2026-11-07.json](m2-native-cache-state-2026-11-07.json)。来源/核心/界面与验收 `1a897576` 已沿既有 GitAtlas 分支普通增量推送,远端逐 SHA 一致;见[推送回执](gitatlas-m2-native-cache-state-push.json)。回执和完成接续最后另提交、再推送核对,不用 force。
|
||||
|
||||
本批只有源码和隔离验收,0 真实模型/真实 Key 读取/正式写入,不委派、不续旧实验或用户项目、不发行/安装/重启正式服务。正式仅沿 F 的 1.25.12/PID71628 历史记录,不作新的健康声明。默认 unrestricted/passthrough/preserve 与持久 ask/auto/full 保持。
|
||||
|
||||
W/AD/AE/AG 网络/unknown、重复读取普遍改善未证明、R/W v2 真实排除边界未触达、Purge 收益 unknown 均保留。下一 I27/I28 CI,然后 M3 统一独立源码/程序发行,必须核对实际 schema2 迁移增量及新处理器产生结果后的回退;不能删除新账本或直接移除逐表全等守卫。之后另冻 M4 新真实供应商与工程基线,再按可比结果开展收益验证。
|
||||
@@ -0,0 +1,85 @@
|
||||
# M3 请求账本迁移与兼容恢复验收
|
||||
|
||||
2026-11-08。本轮已完成安装器精确 schema1→2 迁移、无新增结果的初始化撤销、已有结果后的兼容恢复,以及实际独立载荷/网页/隐藏桌面验收。统一发行尚未完成。
|
||||
|
||||
实现保留在未提交工作树。基线 HEAD 是 `f7705d96347cb19ea4ceab745bb2408061ef7621`,不能把它标为本轮最终产品 commit。按本轮接续约束,不自动提交、推送、发布或升级正式服务。最终输入和验收摘要见 [机器记录](m3-migration-recovery-2026-11-08.json),原件在 `.qa/m3-release-2026-11-08/`。
|
||||
|
||||
## 修复与行为
|
||||
|
||||
- [精确迁移校验](../src/core/desktop/data-retention.mjs) 在一致 SQLite 读事务中核对原表、schema 对象、user_version、sequence、原 metadata 摘要和逐条旧账本预期摘要。只允许新增指定账本表、四个索引、合法迁移标记及完整有序旧条目;原表、工件和公开配置继续核对。少行、改归属、同长度篡改、重算初始 SHA、未知对象/标记、错误版本和并发变化均拒绝。
|
||||
- [迁移定义](../src/core/storage/request-ledger-schema.mjs) 与账本共用原 SQL、投影及旧条目编码。定义 `model-attempt-ledger-v1`,schema1→2,定义 SHA256 `f3a7a85b0cf532def430c9f3b608d65d83987acbab952ce821d8498b1c1a7705`。旧 `modelAttempts` metadata 字符串保持,current 与 initial 分别保存。
|
||||
- [构建声明](../scripts/build-data-compatibility.mjs) 绑定实际 server/hot-upgrade 字节和三个存储源文件,声明读写 schema1/2。未声明旧载荷只按 schema1 处理,不能接管已有 schema2 结果。不是依据名称推断兼容。
|
||||
- [安装器](../src/core/desktop/install-manager.mjs) 在发布前后重核数据,提供兼容恢复目标和 CLI rollback。无新结果失败使用当前载荷源绑定撤销;有新增/修改结果则保持完整 schema2,拒绝删账本、降 schema 或覆写旧 metadata。
|
||||
- [热升级](../src/server/hot-upgrade.mjs) 恢复原对象、hub property descriptors、runtime.results、原处理器和 authority,并恢复公共 `service-info.json` 原字节。恢复前核对仍是本次发布字节;外部漂移时保留现场并拒绝覆盖,不重建注册秘密或提取模型凭据。
|
||||
- 九个 UI 验收入口统一使用 [desktopLaunchCommand](../scripts/desktop-launch.mjs):打包 Electron 自行加载 resources/app,默认 Electron 才携带 main 位置参数。四项测试覆盖三种入口和清单篡改拒绝;本轮实际重跑完整账本两端,不把它称为九套界面全量重跑。
|
||||
|
||||
未改变默认 unrestricted/passthrough/preserve 和持久 ask/auto/full;没有恢复本地模型额度或增加逐步审批。
|
||||
|
||||
## 最后输入和构建
|
||||
|
||||
固定 Node `v24.22.1`,二进制 SHA256 `d702daa815eabbff763c7d3808e235cf031cf2ca70211b56847bc72f51e240e5`。
|
||||
|
||||
`final-inputs-before.json` 冻结1117份当前程序/测试/构建输入和30份编译产物。独立 `final-source/` 用公开 npm registry、空的自有 user/global npm 配置及过滤环境执行 `npm ci --ignore-scripts --no-audit --no-fund`,138包安装成功;随后完整 tsc/Vite/mobile/server 构建通过。未执行依赖生命周期脚本,Electron 运行时另沿实际 F 完整载荷核验。
|
||||
|
||||
`final-inputs-verified-v2.json` 证明1117源输入和30编译产物全部逐字节一致。请求账本源码 SHA256 为 `326f360e0ea1537c6561babeed9de268983c51bf009929759f04ad342612ab88`;不能用接续前旧 SHA 为当前实现补签。这里是独立源码重建,尚未生成最终提交的源码 ZIP、程序发行包或 Inno Setup 安装器。
|
||||
|
||||
## 测试证据
|
||||
|
||||
| 轮次 | 结果 | 范围与限制 |
|
||||
| --- | --- | --- |
|
||||
| related-round-1 | 50通过 / 1文件失败 | 新并发夹具未在清理前关闭 writer;核对归属后结束所属子进程,原失败保持 |
|
||||
| retention-round-2 | 26/26 | 修夹具 try/finally 清理顺序后自然结束 |
|
||||
| final-related-round-3 | 76/76,356713.8326ms | 索引恢复最终修复前,不能补签之后代码 |
|
||||
| ledger-index-round-4 | 18/18,36334.3639ms | 含实际 HostAuthority 原索引恢复、外部漂移拒绝 |
|
||||
| final-related-round-5 | **82/82,332880.741ms** | 最后版本六文件单轮自然结束,无取消、跳过或 todo |
|
||||
|
||||
最后六文件为 desktop-data-retention、desktop-install、desktop-host、release-efficiency、request-ledger、desktop-launch。属于相关回归,不称本轮全仓单轮测试;旧 M2 的1222/1222保持其原输入身份。
|
||||
|
||||
## 实际 F 载荷跨版与恢复
|
||||
|
||||
原 F 来源 `6867cd5a3094c133b09a7c50384ae844ef34c038`,程序 ZIP 242383394字节,SHA256 `857993eee3280649505f47801e94006dc4239d4becb04657695173a9697b66d1`。实际原载荷11041文件/653883963字节已完整核验。先前静态原问题复现保持在 `original/reproduction.json`,没有将 mock 升级称作实际 F 热升级。
|
||||
|
||||
最终实际轮:`actual-1794097076868-1cc0cee5/result.json`,**7/7通过**。独立 f/m3/recovery 程序目录、独立数据及 loopback 端口,未使用4318。
|
||||
|
||||
1. 实际 F 宿主 PID68632,生成一次自有合成 HTTP 成功记录和新的合成保存凭据。
|
||||
2. 迁移后明确注入发布故障,无新增结果时原 handler、authority/index、安装指针、schema1、原 metadata、全部持久快照 exact 恢复;原注入错误保留,操作回执为 failed / original-host-and-pointer-restored。
|
||||
3. 正常激活精确 schema1→2,PID/instance、原连接、旧条目及原工件保持。
|
||||
4. 每个新会话各产生一次成功、上游原503和断线 unknown,共4次已记录物理请求;每条 session/task/run、请求 SHA、current/initial SHA对应。
|
||||
5. 原 schema1 恢复目标在改变指针前被拒绝,原状态保持。
|
||||
6. 回退至冻结的 schema2 兼容 recovery,原宿主 PID/instance保持,4条新旧账本的全部 current/initial、全部当前持久行和工件 exact 保持,无重放。
|
||||
7. 真正关闭原宿主,fresh start 为 PID70184。分页和完整 NDJSON导出均4条,随后仅新增一次合成请求,账本变为5条,前4条和原 metadata不变;最终所属宿主正常停止。
|
||||
|
||||
**recovery 与 m3 使用同一源码及同一编译的不同不可变副本。** 本轮证明兼容恢复机制及数据保持,未证明回到不同旧业务实现;正式发行仍需冻结具体发行版本和恢复载荷。
|
||||
|
||||
`actual-final-offline-audit.json` 额外只读核对5个物理请求、9份请求/响应工件原字节及归属。上游503原99字节、request ID、type和原文exact;unknown无响应工件、usage=null/unknown,不补造计费。原合成凭据密文 metadata行 SHA仍在最初快照中,没有解密;工具执行0,审计没有发新请求。
|
||||
|
||||
## 网页与实际隐藏 Electron
|
||||
|
||||
- 网页:`ui/web-506bb8b5-1400-4c4c-8c10-87ef0c4b0a9f/result.json`。
|
||||
- 桌面:`ui/electron-6ae22b83-b73d-4d13-9888-2a06e1b8f5d9/result.json`。
|
||||
|
||||
两端各5组通过、23页、7张存档图,分别只有1次本机合成 SSE请求。覆盖1107旧条目→1108、超过100/1000的完整翻页、未知原初始记录下载、删除连接后的历史筛选、完整/空导出、单独注入读取失败后恢复、新请求用量与原响应下载、真正宿主重启及零重放,1280/390均检查。JS错误和外部网络请求均0;测试正文与当前源验收器完全一致,原生捕获调整见 `ui-driver-identity-v3.json`。
|
||||
|
||||
隐藏桌面不能把 Playwright截图超时或旧合成帧当作产品断言通过。第三轮的截图超时保持;第四轮交互全部通过,但其390来源图出现旧帧,原图保留且不作为来源可见证据。另以第四轮数据的新副本进行 **0新请求** 原生补验:`ui/native-capture-c63151f1-ab68-4166-b90f-6387313c1970/`。capturePage保持隐藏,预捕获、等待两渲染帧后在390/1280捕获来源;数据exact、所属进程退出,两个来源画面均实看正确。
|
||||
|
||||
最后两端及补图共16张存档。抽看7张,其中1张旧帧不采纳,其余6张与状态相符;不称全部图片逐张人工审查。
|
||||
|
||||
## 原错误、纠正与清理
|
||||
|
||||
- 首实际轮 `actual-1794096137228-1e133bdd/` 在迁移验证通过后暴露公共索引未回退:原 handler/schema/pointer已恢复但 service-info仍指新候选,reconnect和cleanup stop拒绝,receipt为rollback-unverified。原件未修改。
|
||||
- 该轮 PID61812 后按保存身份、Windows创建/用户/二进制、安装身份及注册HMAC核对归属,确认idle/无桌面窗口后认证正常停止。错索引和原持久快照exact保持,未修改现场“通过”;见 `actual-first-failure-owned-stop.json`。
|
||||
- 桌面首轮 HOST_PROCESS_UNVERIFIED 与迟到宿主的原因仍unknown。迟到的自有宿主已证明归属后正常关闭。第二轮的main位置参数错误与此不同;第三轮截图超时及第四轮旧帧分别保留,不合并归因。
|
||||
- build-round-2误选系统Node16;build-round-4误用不存在的仓库npm入口;后续使用固定Node24自身npm和命令内PATH。首独立构建4产物因父目录依赖路径不同而SHA不同,原30产物保持。离线npm ci因yaml缓存缺失失败,随后自有空配置的公开registry clean npm ci和重建通过,未用归一化输出冒充原字节一致。
|
||||
- 原 `ui-driver-identity.json` 漏记M3副本的错误启动改写;v2/v3明确纠正,保留首记录。没有改原 M2封存根来隐藏这些失败。
|
||||
|
||||
最终只读观察10份保存身份、9个自有端口,所属宿主/桌面进程及监听均0。76,776份原件、4,258,394,019字节已封存并逐文件离线复核;MANIFEST SHA256为 `78ce1df774be24d37c8a3e55d943a24f9cab2dd5e955e1743e631723f0bddd00`,该根不再追加。清单与验封结果见机器记录及根外 `.qa/m3-finalization-2026-11-08.json`。
|
||||
|
||||
M2原MANIFEST仍为 `aaeaf694a1cb2444ee8bdfd5ec74ccc768c5659a38fdd6b3be36d6d69abe4951`;本轮未重新逐文件验封整个旧根,不能据清单SHA相同冒称旧全部文件再次复核。正式服务未查询新health、未写入/重启;真实Key读取0,真实模型请求0。
|
||||
|
||||
## 接续边界
|
||||
|
||||
当前修复和隔离验收已收尾;[M3计划](M3-RELEASE-PLAN.md)的最终产品commit、发行编号、独立源码包/程序包/Inno Setup、具体恢复业务版本及正式升级仍待,不自动进行。本轮未启动B4b。
|
||||
|
||||
Mnemo八文件输入v2只读登记保持,来源所报13宿主/12物理/40旧回归没有在本轮重跑,25原生案例全部pending。未启动桥/读取其fixture credential/执行checker/跨线程回复。旧inputs.v1账本SHA不适用于当前M3;待冻结最终账本/四接缝七契约,再按默认关闭、独立producer UUID、单项目、持久用户确认、version/sequence/grantId、原30秒期限和完整失效manifest推进。原键unknown只lookup,全文额度与模型额度分开;sameRequestEvidenceConsistent不等于nativeJointAcceptance,不改已发工件迎合checker。
|
||||
|
||||
M4真实官方/代理基线仍在统一发行后另冻。W/AD/AE/AG网络/unknown、真实自主工程交付、重复读取普遍收益、R/W v2真机剔除边界及Purge收益unknown完整保留。
|
||||
@@ -0,0 +1,44 @@
|
||||
# M4 真实兼容与 Node 工程阶段验收
|
||||
|
||||
2026-11-11,正式 1.25.13/PID 63892,冻结驱动提交 `7152d654`。本阶段真实兼容六次请求、新 Node 工程十次请求,共 181690 报告 tokens、0 unknown。Node 自主交付通过;Claude 代理的真实型号一致性未通过。React、Python 和 Purge 尚未由本阶段验证。
|
||||
|
||||
机器记录:[m4-compatibility-node-2026-11-11.json](m4-compatibility-node-2026-11-11.json)。新原件根 `.qa/m4-live-baseline-2026-11-11/`,原无凭据观察、四轮隔离驱动和全部历史样本保持。
|
||||
|
||||
## 真实供应商兼容
|
||||
|
||||
用户在模型中心恢复三个凭据后,公开确认 DeepSeek 官方 Chat、Claude 代理 Messages、GPT 代理 Responses;三条连接均保持原定义、streaming=true、空原生参数及 credentialRevision=1。每连接一次目录预览、文本/工具各一次新会话请求;三个 DNS/TLS 诊断均通过,诊断自身 0 HTTP/模型请求、无凭据访问。目录未应用。
|
||||
|
||||
| 连接与协议 | 目录 | 文本/工具 | 实际请求输出字段 |
|
||||
| --- | --- | --- | --- |
|
||||
| DeepSeek 官方,deepseek-flash,Chat | 完整 2 型号,选定型号在内 | 两项完整、matched,97/430 tokens | 无主动输出限额 |
|
||||
| apikey-Claude,claude-opus-5-5,Messages | 完整 11 型号,选定型号不在内 | 两项完整,但都自报 gpt-6.1-sol,193/266 tokens | 官方必填 max_tokens=128000 |
|
||||
| apikey,gpt-6.1-sol,Responses | 完整 11 型号,选定型号在内 | 两项完整、matched,78/163 tokens | 无主动输出限额 |
|
||||
|
||||
六响应均实际收到完整 SSE 和报告 usage,工具预检只观察原调用、0 工具执行。Claude 原来的 missing max_tokens 未复现,但身份 mismatch 和目录缺项保留,不能据此宣称真实 Claude 兼容通过。没有改型号、改返回身份或自动重试。相同型号标识也不认证上游模型权重;媒体和长期稳定性未在本切片测试。
|
||||
|
||||
## 新 Node 自主交付
|
||||
|
||||
按预声明优先级选择通过全部预检的 DeepSeek 官方连接,独立复制原 `benchmarks/live-delivery-v1/node`,保留原九项测试、构建和配置。模型自行读取七个不同文件、修改三源码、执行检查并完成限定 Git;没有人工批准、监督续接或实现答案。
|
||||
|
||||
- Project `9bc42831-29a8-4a4d-8638-24366d7cc10d`,Session `cc136ea3-8407-4f69-8e94-69157553ebcd`。
|
||||
- Task `8249fea9-3613-4830-b016-53446019c476`,Run `64fbbda1-036b-4e2a-bd04-174504e31e53`。
|
||||
- 十请求全部 completed/matched,input 171023、output 9440、total 180463,0 unknown;实际执行 84451ms。
|
||||
- 独立原九测试 9/9、build 退出 0;只提交 `src/amount.mjs`、`src/summary.mjs`、`src/report.mjs`,提交字节等于最终文件。
|
||||
- Git `3cd994ff22c1658cbe4c26a7e317ce6b7b314745`;owner 全部原字节、原暂存和未跟踪状态保持,全部保护文件 index 未变。
|
||||
- 内置 delivery `1d2a6e2a-46ea-4217-b0a9-d8f717bfe36d` 为 completed、同一 Git;三份原 JSON 报告及对应 Markdown 分别保留,前两份任务仍 running,最后任务 completed,不覆盖早报告或旧 AD 的 in_progress。
|
||||
|
||||
十份原请求均未主动添加输出限额;工具 schema 及原前缀证据按请求保存。上游报告 cacheRead 122112、uncached 48911、reasoning 6113;cacheWrite 十请求均 unknown,不能写成 0。七次 read_file 中同 SHA/同范围重复读取为 0。本次单样本只能证明这些观察,缓存因果收益和普遍重复读取改善继续未验。
|
||||
|
||||
## 原状态与首次审计失败
|
||||
|
||||
首次审计发现批次期间另一条“你好”新会话,未在 M4 声明范围内,严格归属检查停止。只读封存其显式 ID、每行 SHA、七工件、原请求/响应和导出;一请求 4301 报告 tokens、0 unknown、身份 mismatch,单列外部新增,不归入 M4 请求或交付。不推断操作人,不续跑其任务。
|
||||
|
||||
第二次审计误查询不存在的 `model_attempts.project_id`,退出;按真实 schema 改为既有 sessions 关系查询。两次失败结果和各自 after 快照保持,未修改产品或数据库。第三次只读审计 `audits/node-2e02c1a0-cda2-4863-a962-8e6b54771bfc` 通过。
|
||||
|
||||
原 102 项目、127 会话、377 请求、5244 工件及 53527 原行/字段和 schema 核对通过。项目前插位置 +4、configRevision 124→126、registry 176→178、三已知网络诊断追加和两序列按明确增量重算原 SHA;其他原字段、全部原导出和工件字节 exact。新增为 M4 四项目/七会话/十六请求,另上述外部一会话/一请求;合计新增 782 事件及 outbox、196 records、17 model_attempts、118 工件,归属逐项核对。
|
||||
|
||||
原 PID/instance、空闲、9229 关闭通过;原选择恢复 Claude 单成员池,全部连接定义和原池保持。凭据行只比较 SHA,未读取、解密、复制或重存 Key。产品源码、正式发行和服务未重启。
|
||||
|
||||
## 回顾与下一阶段
|
||||
|
||||
新的真实 Node 成功与内置 Git/结束报告一致,说明 1.25.13 的此工程链路已实际闭环。审计新增识别和真实数据库列修正仅属于本批驱动,失败原件继续可核对。提交此阶段后另开 React,再按完整成功前置开 Python。原网络异常、unknown、输出截断、R/W v2 真机边界、Mnemo 生产语义、物理 iPhone、长期与 Purge 收益保持原结论;Purge 必须另冻可比基线及插件/宿主来源。
|
||||
@@ -0,0 +1,44 @@
|
||||
# M4 新真实兼容与自主工程基线:最终验收与回顾
|
||||
|
||||
2026-11-11 发起,2026-11-12 收尾。本批使用已正式交付的 1.25.13/PID 63892,三条用户恢复的连接、新会话和独立原工程;四十次真实模型请求、629106 报告 tokens、0 unknown。新的 Node、React、Python 均实际交付成功;Claude 代理身份未通过,不能宣称全球厂商或所有模型能力已验证。
|
||||
|
||||
机器记录:[m4-final-2026-11-11.json](m4-final-2026-11-11.json);封存回执:[m4-live-seal-2026-11-11.json](m4-live-seal-2026-11-11.json)。程序源码未变,没有重新发行、安装或重启。
|
||||
|
||||
验收与封存提交 `e132601b` 连同此前六条已普通增量推送至既有 GitAtlas 分支,远端和重新 fetch 逐 SHA 一致,见[推送回执](gitatlas-m4-live-push.json)。本回执和完成接续另提交、再核最终远端 HEAD;`.qa` 原件保持本地封存。
|
||||
|
||||
## 真实结果
|
||||
|
||||
| 新样本 | 请求/报告 tokens | 原独立验收 | 实际 Git/报告 |
|
||||
| --- | --- | --- | --- |
|
||||
| 三线文本/工具预检 | 6/1227 | 六完整 SSE、reported usage;DeepSeek 官方 Chat 和 GPT Responses 代理 matched;Claude 两 mismatch | 工具仅观察,0 执行 |
|
||||
| Node | 10/180463 | 原 9/9 测试、build,owner 字节及全部原 index | 3cd994ff,仅三源码;内置 delivery completed、最终 Task completed |
|
||||
| React | 14/271333 | 原 9/9、typecheck、build、实际 Edge 桌面/390px 六交互、两 PNG 实看 | c55644a3,仅四源码;终端来源 verified,最终 verified_committed |
|
||||
| Python | 10/176083 | 原 10/10 pytest、compileall,现成 3.10.11/pytest 8.4.2 | fa2ce16a,仅三源码;终端来源 verified,最终 verified_committed |
|
||||
|
||||
三工程全部保留原测试、配置、owner 字节与暂存/未暂存/未跟踪状态,提交字节等于当前文件。无实现答案、人工批准或监督继续。React/Python 没有内置 delivery_job,实际终端来源与进程归属 unproven 按原证据保留;不补造作业。早晚报告分别封存,不覆盖 in_progress 快照或旧 AD。
|
||||
|
||||
Claude 代理请求 claude-opus-5-5,却两次自报 gpt-6.1-sol,完整目录也未列选定型号。实际 Messages 已发送官方必填 max_tokens=128000,旧 missing-field 错误本轮未复现;身份失败和原响应仍保持。Chat/Responses 所有实际请求没有主动输出上限。没有自动重试、换线续跑失败、修改连接身份策略或补写上游型号。型号字段匹配也不认证模型权重。
|
||||
|
||||
详细阶段记录:[兼容与 Node](QA-M4-COMPATIBILITY-NODE.md)、[React](QA-M4-REACT.md)、[Python](QA-M4-PYTHON.md)。前置隔离驱动的 11 边界、固定 Node 工程及全部首失败单列,不作为真实模型能力。
|
||||
|
||||
## 读取、缓存与边界
|
||||
|
||||
三工程 read_file 分别 7/13/6 次,同 SHA/同范围重复均 0;改后版本回读作为新事实。上游报告 cacheRead 合计 440960、uncached 164142、reasoning 12222;cacheWrite 三十四工程请求全部 unknown,不能写成 0 或增加实际计费。每请求的原 usage、字段来源、完整 wire、工具 schema/前缀和原响应均可核对。
|
||||
|
||||
这些是本批观察,未证明长期稳定、所有推理强度/媒体、缓存因果收益或普遍重复读改善。真实官方 OpenAI/Claude 直连未以本批代理代签。R/W v2 真机剔除边界和旧预算结论保持;默认 unrestricted/passthrough/full、完整历史与产品额度 0 不回退。
|
||||
|
||||
## 最终原状态与封存
|
||||
|
||||
最终审计 `audits/final-bfadc758-619b-4288-b516-95e461da314d` 通过:原 53527 行/全部字段及 schema、102 项目、127 原导出、377 原请求、5244 原工件保持;本批九份完成捕获导出也 exact。configRevision 124→130、registry 176→182、前插 +6、三已知网络诊断和两序列按明确增量复算原 SHA,其他原字段 exact。
|
||||
|
||||
新增为本批六项目/九会话/四十请求,另一个外部“你好”新会话/一请求 4301 报告 tokens,独立逐行/七工件核对,不计入 M4;合计新增 2090 事件及 outbox、533 records、41 model_attempts、332 工件。原 PID/instance、六连接/池/原选择恢复、空闲及 9229 关闭通过,凭据行只比 SHA,未读、复制、解密或重存 Key。
|
||||
|
||||
原件根 `.qa/m4-live-baseline-2026-11-11/` 已封存 541 份、134884464 字节,MANIFEST SHA256 `558c9b548559798279d7f8e61deb3b7344920753a53dbf4e84eec374c6f3a3df`,逐成员 bytes/SHA 核对通过。包括无凭据观察、所有原请求/响应/usage、两次审计失败与全部后续结果、三工程/Git、两原 PNG、机器记录、十三份最终驱动原字节及未采用的附加观察。可按原锁重建的 node_modules 不入清单;原 package-lock 和 npm-ci 日志已封存。原旧根与失败不覆盖,M4 根不再追加;原件不进 Git。
|
||||
|
||||
## 来源同步与下一阶段
|
||||
|
||||
当前阶段边界收到的 NRO 来源另根只读登记,见[协调记录](integration/MNEMO-NRO-READONLY-2026-11-11.md):可选 provider_request_id、HTTP 不完整保留 unknown/原身份原键 GET、不重 POST;四验收/230 清单 SHA/源方 27 独立回执保持范围。AD 与人类另选 AG Task 分列,没有补包、重派、来源运行或回信。它不改变本批四十请求或 Mnemo 生产/语义资格。
|
||||
|
||||
本批已得到可用官方 DeepSeek 的新三工程成功基线,结束本轮执行与证据收尾。普通 GitAtlas 增量推送按既有授权完成后,下一 M8 先另冻原版 Purge 的插件/宿主/输入/身份和 AB-off→AB-on→BA-on→BA-off,先隔离验证驱动、再真实新例;不续旧 AG 或拿这次内置 Harness 成功代签原版收益。Purge、缓存/经验/角色因果、真实 Mnemo 生产绑定/语义、长期、物理 iPhone 和全球原生/媒体/电脑操作仍按各自计划推进。
|
||||
|
||||
封存后人类报告 Claude 代理 Key 此前写错并已更正。本报告和原状态核对属于封存时点,不用于回退该配置修改;原因未独立证明。修后目录与原文本/工具两项另冻公开修订、新 Task/Run/请求键及至多两请求,不执行工具,不重跑已成功的三工程或 Purge。新结果单列,原两次型号不符、目录缺项、四十请求统计和 541 成员清单均不改签。
|
||||
@@ -0,0 +1,27 @@
|
||||
# M4 新 Python 自主工程阶段验收
|
||||
|
||||
2026-11-11,承接 React 阶段 `52accafe`,同一 DeepSeek 官方连接与正式 1.25.13/PID 63892。新的 Python 工程原 10/10 pytest、compileall 和三源码实际 Git 交付通过;十请求 176083 报告 tokens、0 unknown,运行 48767ms。本批兼容加三工程累计四十请求、629106 报告 tokens。
|
||||
|
||||
机器记录:[m4-python-2026-11-11.json](m4-python-2026-11-11.json)。原件 `.qa/m4-live-baseline-2026-11-11/python/`,原 Node/React 与所有旧失败保持。没有修改产品、模型参数或用户配置,没有发行或重启。
|
||||
|
||||
## 真实交付
|
||||
|
||||
独立复制原 `benchmarks/python-delivery-v2/baseline`,按 README 和原测试理解规范化、汇总、渲染要求。使用现成 Python 3.10.11/pytest 8.4.2,不改测试、配置、README 或 owner。模型自行读取、提出三源码、执行实际检查及终端 Git;无人工批准、监督继续或实现答案。
|
||||
|
||||
- Project `7992409f-6e5d-4eb7-a26d-ccf72c96429a`,Session `26b24e0c-db78-4bc0-9fb9-049b2d741966`。
|
||||
- Task `875afb7b-ea0b-41bf-8434-5897f5c800f0`,Run `c430c404-670f-411a-991e-9fdcd73d68f7`。
|
||||
- 十响应全部 completed/matched、usage reported;input 169490、output 6593、total 176083。
|
||||
- 独立原十 pytest 10/10,compileall 退出 0;原 XML、stdout/stderr SHA 保持。
|
||||
- Git `fa2ce16a2e6bd372b0754637df84979b413a09ad`,仅 `normalize.py`、`render.py`、`summary.py`,提交字节等于最终文件;全部保护文件 index、owner 原字节与暂存/未暂存/未跟踪状态保持。
|
||||
|
||||
终端 Git 原证据 verified/currentVerified。早报告 `df41ebb3-0173-4288-aaba-95667eeb8dfa` 为 in_progress/Task running,最后 `7fe2c83a-d65a-4f6e-9d78-ae96e2d97dd7` 为 verified_committed/Task completed/lifecycle final;各 JSON/Markdown 原字节分别保持。此例没有内置 delivery_job,来源 terminal、processAttribution unproven 保留,不把终端观察补签为内置作业。
|
||||
|
||||
六次 read_file 的同 SHA/范围重复为 0;上游 cacheRead 122752、uncached 46738、reasoning 2847,cacheWrite 十请求均 unknown。wire 原字段见机器记录,默认没有主动输出限额,没有裁剪完整历史或恢复旧额度。单样本不证明缓存或重复读普遍收益。
|
||||
|
||||
## 原状态与回顾
|
||||
|
||||
只读审计 `audits/python-fc99c576-f365-4dea-914c-57a07e169dc7` 一次通过;新增核对本批九份完成会话导出与原捕获 exact。原 53527 行/字段和 schema、127 原导出、5244 原工件、377 原请求保持。累计本批六项目/九会话/四十请求,外部“你好”一会话/一请求仍单列;合计新增 2090 事件及 outbox、533 records、41 model_attempts、332 工件,逐项归属核对。
|
||||
|
||||
configRevision 124→130、registry 176→182、前插位置 +6、三已知诊断与两序列按明确增量复算原 SHA,其他原字段 exact;原选择/池/六连接定义和凭据存在状态保持,PID/instance、空闲、9229 关闭通过。没有读、复制、解密或重存 Key,没有续跑旧工程。
|
||||
|
||||
新 Python 的实际工程链路、解释器、终端来源与结束报告已闭环。Node/React/Python 新基线均成功,Claude 代理身份失败仍是独立结论。阶段提交后完成最终原件清单、SHA复核、回顾与普通增量 GitAtlas 推送;Purge 须另冻插件/宿主和可比成功对照,历史异常、真实预留边界、长期、物理手机、Mnemo 生产绑定/语义及收益未验不改签。
|
||||
@@ -0,0 +1,32 @@
|
||||
# M4 新 React 自主工程阶段验收
|
||||
|
||||
2026-11-11,承接已提交的 Node 阶段 `b8eca3b5`,使用同一 DeepSeek 官方连接和正式 1.25.13/PID 63892。新 React 工程原 9/9 测试、typecheck、build、真实页面和四源码 Git 交付全部通过;十四请求 271333 报告 tokens、0 unknown,实际运行 72040ms。
|
||||
|
||||
机器记录:[m4-react-2026-11-11.json](m4-react-2026-11-11.json)。原件 `.qa/m4-live-baseline-2026-11-11/react/`;本阶段没有修改产品源码、连接、模型参数、权限或额度,没有重启服务。
|
||||
|
||||
## 真实工程与页面
|
||||
|
||||
独立复制原 `benchmarks/react-delivery-v1/baseline`,按原锁离线准备依赖;原测试、配置、锁、其它源码、样式和 owner 字节/index 保持。模型自行读取、提出四精确补丁、执行实际检查、两组浏览器检查和终端命令,没有人工批准、继续或实现答案。
|
||||
|
||||
- Project `afc956a2-d08f-4df2-8b39-5ae91338919c`,Session `36f7bbd7-c9d0-490e-b319-ae98f7f4e26a`。
|
||||
- Task `dddcbdf2-588d-4fee-a26c-236e110a925d`,Run `c41c7af7-3e80-42f4-ad66-df997bcec1b2`。
|
||||
- input 264589、output 6744、total 271333;十四响应全部 completed/matched、usage reported。
|
||||
- 独立原九测试 9/9,typecheck 和 build 均退出 0。
|
||||
- Git `c55644a37016c772a9ac8e4d9a98dbc12a6c1223`,仅 `src/App.tsx`、`src/query.ts`、`src/summary.ts`、`src/transition.ts`;提交字节等于当前文件。全部保护文件 index、owner 的原暂存/未暂存/未跟踪状态保持。
|
||||
- 独立实际 Edge 157.0.4341.42 验证 pending/done 筛选、去空白搜索、空态、仅目标任务切换、未被筛选影响的统计和 390px 无横向溢出;pageErrors/requestFailures 均空。
|
||||
|
||||
1280×920 与 390×844 两张真实截图已通过 view_image 检查,原 PNG/SHA 和人工目视记录在 `react/browser/`。桌面显示三个任务、完成两个、待办五小时;窄屏筛选后只显示尚未完成的任务,控件与文字完整。两组模型浏览器检查原报告另行保持。所属会话没有非文本工件描述符,不能声称下载了模型截图;额外只读检查的 0 结果及未采用的采集脚本另存 ancillary,未增加模型请求。
|
||||
|
||||
## Git 与报告来源
|
||||
|
||||
此例采用终端提交,不创建内置 delivery_job。原 `terminal-git-evidence` 已 verified/currentVerified,commit 和四文件范围相同;结束报告 `dcf70953-ad5e-4f58-b974-2477089d70a1` 为 `verified_committed`、Task completed、lifecycle final,gitOutcome 来源 terminal。终端观察的 processAttribution 仍 unproven,不能补造内置作业或进程认证;实际 Git、原工具调用和独立当前字节分别可核对。
|
||||
|
||||
十三次 read_file 的同 SHA/同范围重复为 0,改后读取当前版本作为新事实保留。上游 cacheRead 196096、uncached 68493、reasoning 3262;cacheWrite 十四请求全部 unknown。没有主动输出上限,完整原请求/响应、工具 schema 和前缀来源保留,不能由此宣称普遍缓存或重复读收益。
|
||||
|
||||
## 原状态、回顾与下一步
|
||||
|
||||
只读审计 `audits/react-e49f5eb3-25e9-48e0-b9c8-ad56c8e89fb7` 一次通过。原 53527 行、127 会话导出、5244 工件、377 请求和 schema 保持;本批累计三十请求,外部“你好”一请求仍单列。累计新增五项目/八所属会话,另一个外部会话;共 1531 事件及 outbox、397 records、31 model_attempts、243 工件,全部新增归属核对。
|
||||
|
||||
configRevision 124→128、registry 176→180、项目前插 +5、三已知诊断和序列按明确增量复算,其他字段 exact;原选择和池恢复、六条连接定义及凭据存在状态保持,原 PID/instance、空闲和 9229 关闭通过。没有读、复制、解密或重存 Key,没有续跑旧实验。
|
||||
|
||||
新 React 已取得独立真实交付与页面证据,终端来源得到准确报告,没有把空 delivery_job 列表当成 Git 未完成。提交后按同一冻结线路运行新 Python 原十项 pytest 和 compileall,再做最终封存与普通增量推送。Claude 代理身份、历史网络/unknown/输出截断、Mnemo 生产语义、物理手机、长期、R/W 边界和 Purge 收益继续未验。
|
||||
@@ -0,0 +1,54 @@
|
||||
# 1.25.13 Mnemo 与迁移恢复发行验收
|
||||
|
||||
2026-11-11。承接 [交付计划](M3-B4B-RELEASE-1.25.13.md)。1.25.13 已正式激活并完成本次工程验收;实际来源、程序身份和原失败如下,真实生产记忆绑定及收益仍待验。
|
||||
|
||||
## 实际来源
|
||||
|
||||
- 产品实现:`e306f40caef0592e6f0fb61c8aac007dc8685451`;含原 M3/B4b 源码、界面、测试与完整历史登记,已普通推送 GitAtlas 并核远端 SHA。
|
||||
- 最终发行来源:`35a6d6d9b25f2301bc9b8844b9e6af2a6d3bb94f`,仅追加打包许可证输出修复及许可/记录;产品编译输入不变。
|
||||
- 不可变候选:`Agent-Workbench-1.25.13-win-x64-20261110190745270`,清单 SHA256 `24e6aaccbf0af0ef19d9e020bf4c013438c6c53222b8e4565d224667ce21b660`。
|
||||
- 首证据 `.qa/m3-b4b-release-2026-11-11/` 和最终 `.qa/m3-b4b-release-2026-11-11-r2/` 分开;旧 M3/B4b 封存根只读。
|
||||
|
||||
## 已完成验证
|
||||
|
||||
| 验证 | 实际结果与边界 |
|
||||
| --- | --- |
|
||||
| 相关源码 | 固定 Node 24、并发 2,M3 六文件+B4b 七文件单轮 186/186、0 跳过/取消;不称本轮全仓单轮测试 |
|
||||
| 打包修复 | 语法检查、source-package 与 release-copy 4/4 |
|
||||
| 官方契约 | 固定注册表 2026-11-07.1,26 条/14 测试文件离线核验通过;不是新真实供应商调用 |
|
||||
| 独立源码 | 最终源码包 1399 个导出输入;新目录真实 clean npm ci 138 包,再构建;1399 输入/30 编译逐字节一致 |
|
||||
| 程序包 | 11083 文件的实际程序清单、源码 ZIP/程序 ZIP/Setup SHA 核对及包内 30 编译一致;打包后工作树保持干净 |
|
||||
| 独立安装副本 | 实际 stage/activate/start,公开版本 1.25.13,新数据目录与随机端口,所属宿主正常停止;Inno 生命周期另验 |
|
||||
| Mnemo 后端 | 新合成宿主/新项目/新绑定,N01–N24 单轮 24/24;保留失败/unknown 和原请求 SHA,0 真实模型调用 |
|
||||
| 安装副本网页 | 4 组/10 图,真实连接表单、候选与完整修订确认、刷新、1280/390 回执、伴随证书核验 HTTPS、撤权后原任务继续通过 |
|
||||
| 安装副本 Electron | 实际打包 Electron/隐藏窗口 4 组/10 图全部通过,同样包含窄屏、伴随 HTTPS 和撤权后继续 |
|
||||
| F 实际跨版与兼容恢复 | 7/7,保留原 PID/凭据;精确迁移、初始化失败撤销、三类新结果、拒绝 schema1 目标、兼容恢复及真实停止/fresh restart 后分页导出与继续写入;5 次受控物理 HTTP |
|
||||
| 实际 Inno 生命周期 | 11/11;初装、活动窗口/在途请求/原审批保持、清单与发布指针故障、保留原 PID/instance/DPAPI 密文和全部数据的实际更新、原生入口/窗口登记、重启恢复合成凭据及卸载保留数据 |
|
||||
| 基础文件流程 | 安装副本网页+实际 Electron 共 9/9,选目录/取消/无效路径恢复、刷新保留项目、HTML 及相对资源/交互、源码切换、保存文件刷新、独立页面、切项目和原生选择器取消 |
|
||||
| 正式激活与原数据保持 | 唯一 Setup 返回 0、351789ms;1.25.12→1.25.13,保留原 PID 63892/instanceId 75209495-f90a-4552-9a90-81a680d50c7c/Provider 与 Key 存在状态;schema1→2 精确迁移 374 原请求,102 项目/127 会话/8741 记录/22085 事件及 outbox/5244 工件原行原字节保持 |
|
||||
| 正式公开字段与导出 | 6 连接的新官方规格/契约/能力和完整账本字段按冻结实现独立重算,127 会话两个新增 MCP 计数字段对原记录核验,其余原公开配置/连接/历史和 project/session/execution 原字段全等 |
|
||||
| 正式界面及最后复核 | 网页/实际安装 Electron 各 1280/390,四图全部实看、0 页面异常;final 再核全部原数据和原 PID、9229 关闭;新版用户桌面经正常安装入口恢复,登记 PID 71480,未重跑原任务 |
|
||||
|
||||
目前实看网页发送回执 1280、完整来源 390、撤权后原任务 1280、伴随撤权回执 390、Electron 请求回执 390/完整来源 1280,以及网页/Electron 实际 HTML 预览八图:中文内容、来源修订、原工件入口和错误解释可读,长 SHA 正常换行,关键动作在窄屏内,预览中的相对资源与交互计数实际工作;未把自动断言代替图片观察。原文件名和逐图记录见最终证据根 `visual-review.json`。
|
||||
|
||||
正式四图另见 `formal-ui/visual-review.json`:实际 1.25.13 连接与完整访问模式可见,宽/窄屏控件无横向溢出,原完成会话正常呈现;只是查看历史,未重新执行其内容。正式浏览器的示例项目与 Electron 上次选中项目属于各自原客户端状态,不把截图中的旧交付当本轮真实模型验证。
|
||||
|
||||
机器发行身份见 [mnemo-1.25.13-release-identity.json](mnemo-1.25.13-release-identity.json)。选择性证据清单 `.qa/m3-b4b-release-2026-11-11-evidence-manifest.json` 为 782 份/811582693 字节,SHA256 `611c4c40d9c28b09649d8a041505d1684242f436b7c8b972b7cfbb0c9c65260c`,生成后逐件重读核验。范围含首失败、日志、合成原件、截图、只读正式摘要和驱动;排除重复源码/依赖/安装程序树,源码 ZIP/程序 ZIP/Setup 各用独立 SHA 身份,不称整个 `.qa` 的全量封存。旧 M3/B4b 两份 manifest 原字节重新核对一致。
|
||||
|
||||
## 首失败及回顾
|
||||
|
||||
1. 初始快照脚本把中文 file URL 当路径,ENOENT 且未写出快照;改用 `fileURLToPath` 后冻结。
|
||||
2. 首构建 npm 子命令命中系统 Node 16;固定整条子进程 PATH 的 Node 24 后构建通过,未改产品版本条件制造通过。
|
||||
3. 首包生成后登记发现打包器回写源码许可证,登记失败、未安装首包。修为只在输出目录生成,补存 YAML ISC 文本,另提交和另轮独立导出/打包;最终包登记通过。
|
||||
4. 新合成宿主首次 `--import` 使用 Windows 绝对路径,被 Node ESM loader 拒绝,未启动宿主;改用合法 file URL 后在本轮目录新建宿主和随机端口,旧日志保留。
|
||||
5. 实际旧载荷初始等待期间另做只读清单诊断,完整验证 3665ms 通过;原运行继续,未重放迁移/模型请求。不据此确定原等待原因。
|
||||
6. 基础文件驱动首轮仍假定旧版始终显示文件树,在 Orven 默认对话页等待“页面.html”超时。实际失败截图显示文件入口正常存在;驱动改为先经真实“项目文件”入口打开文件栏,产品代码/编译/安装包不变,首日志和截图保留,另轮复验。这个驱动修正属于后续验收来源,不能冒称已包含在 35a6d6d9 源码包中。
|
||||
7. 首正式快照因一个旧正式桌面窗口仍打开而停止,未运行安装。核对空闲状态、窗口 PID/创建时间/安装路径后只执行正常窗口关闭,原后台 PID 63892 保持;首 PowerShell 窗口身份检查遇到 HTTP JSON 日期自动转成 DateTime,原始观察证明路径/时刻一致,统一 UTC 格式后正常关闭,未强制终止。
|
||||
8. 唯一正式 Setup 返回 0,实际激活 1.25.13,并通过安装器精确 schema1→2 保持核验。之后 QA 首轮仍要求公开模型中心整体 SHA 不变而失败。只读差异清单显示:新官方规格、请求契约、由规格导出的能力状态,以及完整增量账本的总数/分页和存储说明;原连接配置/选择/凭据存在状态/修订与原预览记录无其他差异。新核验逐个用冻结来源计算新增规格、旧连接默认请求契约和能力,断言完整账本总数,再对全部其他原公开字段做严格全等;不删除字段后直接放过、不改产品、不重装或重放请求。首失败与唯一安装回执保留。
|
||||
9. 后续 QA 分别遇到两个尚未按版本处理的断言:会话 execution 导出新增 `mcpJobCount`/`mcpJobsTruncated`,以及 schema2 必须新增的 `sqlite_sequence` 行。现对 127 个导出逐个用原记录只读计数核验两个新增字段,核验后原 execution 摘要和原 project/session 摘要全部一致;序号由原安装器的精确迁移定义核验,不能要求新旧序号表整体相同。公开能力计算的初次只读观察还误用了旧 API 归一化默认声明,已改取逐行保持的非凭据 registry 原声明;不创建 Provider 或读取私有 Key。两项视图核验独立通过为 6 连接/127 会话。以上首失败均保留,期间正式 Setup 仍只有一次。
|
||||
|
||||
恢复载荷采用最终发行程序的另一份不可变副本,验证 schema2 保留新旧结果的恢复机制;不宣称恢复到不同旧业务实现。项目凭据仍只存当前宿主内存,重启须重新授权。
|
||||
|
||||
本轮工程验收使用受控本机 HTTP 接收器和合成凭据。真实 Mnemo 生产绑定、真实模型质量/交付收益、长期效果、物理 iPhone、历史重复读普遍收益、R/W 真机剔除边界及 Purge 收益保持未验证。历史 W/AD/AE/AG 的原网络、失败和 unknown 不改写,不续跑原任务。
|
||||
|
||||
源码e306f40c、打包修复35a6d6d9和验收1e309700均已普通增量推送GitAtlas现有分支,远端与重新fetch逐SHA一致,见[推送回执](gitatlas-mnemo-1.25.13-push.json)。程序ZIP、源码ZIP和Setup为本地发行产物,未放入Git;远端CI runner/job仍未验证。回执与完成指针另作最终提交并核远端,不反复重打程序或重跑已冻结验收。
|
||||
@@ -1,5 +1,31 @@
|
||||
# Tianshu 研究后的 Agent Workbench 采用计划
|
||||
|
||||
2026-11-12 重启丢Key缺陷已随1.25.14正式修复:默认DPAPI及原子配置保存,先保留PID14388升级再实际PID29884启动,七条Key自动恢复、原行/139导出/5583工件/选择/密文与420请求保持。54相关/独立源码程序/实际两端/隔离安装与715成员封存通过,见[验收](QA-CREDENTIAL-PERSISTENCE-1.25.14.md)。本轮真实供应商0、Key未读取或重加密;当前提交/普通GitAtlas推送,后续M8另冻,所有历史真实异常/unknown/重复读/预留/Purge未验保持。
|
||||
|
||||
2026-11-12 M4原收尾c8c36c27及Claude独立验收5ca65a45均已普通推送并与远端/fetch一致,见[修后推送回执](gitatlas-claude-proxy-correction-push.json)。回执/完成接续另提交核最终HEAD,541/41两封存保持;Claude403未通过,下一M8另冻,不追加两请求或改签历史收益。
|
||||
|
||||
2026-11-12 修后Claude两新例另封存:目录14型号/选定在内,两模型请求上游403“仅允许Claude Code客户端”,usage/响应身份未知、费用null、0工具执行;[验收](QA-CLAUDE-PROXY-CORRECTION-2026-11-12.md)。41成员36284086字节/原状态审计通过,当前七连接与人类更正保持,原40请求/541清单不改签。验收普通提交推送收尾,不追加请求或重跑三工程/Purge;M8收益仍须另冻。
|
||||
|
||||
2026-11-12 M4验收e132601b及此前六条已普通增量推送,远端和重新fetch逐SHA一致,见[回执](gitatlas-m4-live-push.json)。原40请求/541封存不变;下一按人类修正另冻Claude两项预检,Purge收益仍须新对照。完成回执和接续另提交核最终HEAD,下方收尾中按原时点保留。
|
||||
|
||||
2026-11-12 M4原批次最终完成:六兼容预检与三独立工程合计40请求629106报告tokens/0unknown,原要求/检查/Git和全部原状态通过;541成员/134884464字节已封存复核,见[最终验收](QA-M4-LIVE-BASELINE.md)。当前收尾提交及普通增量GitAtlas推送,不重发行1.25.13。封存后人类更正Claude代理Key,另用新修订/Task/Run至多两文本与工具预检,旧两mismatch及目录缺项不改签,不重跑三工程或Purge、不覆盖人类配置。M8原版Purge对照另冻;历史异常/真实预留边界/记忆与缓存因果/物理手机/长期未验保持。下方进度按原时点保留。
|
||||
|
||||
2026-11-11 M4新Python也已通过:10请求176083tokens/0unknown、原10pytest/compileall/三源码终端Git和最终报告,原53527行/127导出/5244工件与九新捕获导出exact,见[验收](QA-M4-PYTHON.md)。三类工程完整通过;本批40请求629106tokens,Claude身份失败、外部一你好请求单列。下一最终封存/回顾/普通增量推送,再按目标线路和机制另冻Purge;真实收益及历史异常仍保留,不重发或重启1.25.13。
|
||||
|
||||
2026-11-11 M4新React已通过:14请求271333tokens/0unknown、原9/9/typecheck/build、实际桌面/390px和四源码终端Git/最终报告,原字节/index及53527原行保持,见[验收](QA-M4-REACT.md)。本批累计30请求453023tokens,Claude预检身份失败和外部一你好请求分列;下一独立Python,再最终封存与普通增量推送。Purge/缓存因果/普遍重复读/Mnemo生产语义仍待,不重启1.25.13;下方下一React为Node原时点。
|
||||
|
||||
2026-11-11 M4已完成真实兼容和新Node:六预检1227tokens/0unknown,DeepSeek官方和GPT Responses代理通过,Claude代理身份mismatch/目录缺项原证据保持。DeepSeek自主Node十请求180463tokens/原9/9/build/限定Git及内置最终交付通过,原127导出/5244工件/53527行与全部旧事实核对;外部一“你好”会话/一请求和两审计失败分别保存。见[阶段验收](QA-M4-COMPATIBILITY-NODE.md)。下一独立React、成功后Python,再按插件/宿主和可比成功基线另冻Purge;缓存/重复读普遍收益、Mnemo生产语义及历史unknown仍未验。1.25.13未重启或重新发行,下方“下一M4”为先前时点。
|
||||
|
||||
2026-11-11 当前:M3+B4b已正式交付1.25.13/source35a6d6d9,独立1399输入/30编译、186相关/4打包、实际Inno11/11/F跨版7/7及Mnemo24后端/两端工程路径通过。唯一正式Setup成功,原PID63892/102项目/127会话/5244工件及旧历史保持,正式四图和final核对通过;见[验收](QA-MNEMO-RELEASE-1.25.13.md)。源码/验收1e309700已普通增量推送且远端/fetch逐SHA一致,见[回执](gitatlas-mnemo-1.25.13-push.json);下一M4另冻新真实基线,不续旧实验;Mnemo生产绑定/真实语义/重复读/R-W边界/Purge收益unknown保持。下方未提交/未发行为原时点。
|
||||
|
||||
2026-11-08接续:M3隔离迁移/恢复实现后,B4b默认关闭的Mnemo单项目记忆桥已完成原生工程验收,25项证据、最后相关104/104、916输入/30编译及两端/伴随HTTPS见[验收](QA-B4B-MNEMO-NATIVE.md)。M3+B4b仍未提交或统一发行,本轮未正式操作、真实模型或Purge验证;原失败和历史未知保持。下一M3最终产品身份/源码程序包/InnoSetup及独立安装验收。下方最新/下一均按其历史时点理解。
|
||||
|
||||
I27/I28推送完成:源码与验收44b6a09b已普通增量推送至既有GitAtlas分支,远端HEAD和重新fetch的14个提交逐SHA一致,见[回执](gitatlas-m2-ci-push.json)。回执/完成接续最后另提交并再次核对;最终4a03ec52干净1222/1222及7595封存原件保持,正式未部署/重启,runner/job/真实供应商及历史收益unknown不代签。下一按M3-RELEASE-PLAN.md推进实际schema2迁移与新结果后的兼容回退;下方进行中说明按原时点理解。
|
||||
|
||||
2026-11-08 I27/I28本地CI及官方契约已完成:4a03ec52最终干净146文件1222/1222、实际网页/隐藏Electron7组4图实看、1376输入/29编译一致,见[验收](QA-M2-CI-OFFICIAL-CONTRACTS.md)。首失败/原下载异常/原生画面滞后与历史unknown保持,7595新原件封存、五旧根逐SHA一致;文档/普通GitAtlas增量推送收尾,远端runner/job/上传仍未知。下一[M3](M3-RELEASE-PLAN.md)精确schema2迁移及新结果后的兼容回退,再统一发行;不能删新账本或取消原行/工件保持核对。正式未操作/重启,0真实模型/Key,开放默认/持久权限与W/AD/AE/AG网络、重复读/R-W边界/Purge收益unknown不变;规模性能仍待M7。下方历史计划保持原时点。
|
||||
|
||||
M2 I09/I11原生缓存块/状态与四范围连接保持已完成源码71f18055/9e1feeeb/ee97aaae及两端各8组/10图隔离验收,原存储/密文/历史exact,见[验收](QA-M2-NATIVE-CACHE-STATE.md)。5126原件封存,首失败与16专项/两文件26项按SHA分别保留;源码与验收1a897576已普通增量推送、远端逐SHA一致,见[回执](gitatlas-m2-native-cache-state-push.json);完成接续最后另提交核对。下一I27/I28 CI及M3实际schema2增量/新处理器结果后回退,后续新真实供应商/工程与收益另冻样本。原A–F机制及真实AD/AE成功、W/AG网络/unknown、重复读/R-W边界和Purge收益unknown均保持,正式本批未写入/重启。下方旧当前按原时点理解。
|
||||
|
||||
M2第三批I15来源/纯注册表及现有Orven来源/冲突/推理/兼容覆盖已完成,源码ba4b4491/b5aaf1f7;最终858输入/29编译、两端各6组/7图及原metadata/cipher/存储快照exact,5688原件封存,见[注册表验收](QA-M2-OFFICIAL-REGISTRY.md)。首候选62/62只证明未变后端,原失败保持;文档/授权增量推送收尾中。下一I14版本/头/指纹、I09/I11/I12缓存块/亲和/状态与CI;M3真实迁移增量/新处理器回退待验。本批不部署/重启或触真实Key/模型/正式数据,不续旧实验,历史网络/unknown/重复读取/Purge收益unknown保持。下方按原时点理解。
|
||||
|
||||
当前M2第二批I13源码da2e6313完成完整请求账本与逐请求增量存储,原metadata/初始记录/unknown不改。最终build-4独立39/39、网页与实际Electron各5组/7图/23页通过;830输入/29编译与candidate-app-5一致,4328原件只读验封通过,见[验收](QA-M2-REQUEST-LEDGER.md)和[机器记录](m2-request-ledger-2026-11-07.json)。相关102/102发生在最后公开数据库版本修正前,不能混称最后一轮;首失败完整保持。文档与授权GitAtlas普通增量推送收尾中。下一I15官方注册表和I09/I11/I12/I14缓存块/状态/指纹/版本、I27/I28 CI,M3统一发行须实际核验schema2安装增量及新处理器回退,不去掉数据保持守卫。本批未正式部署/重启或新health复查,历史真实异常/重复读取/Purge收益unknown保持;下方旧当前按历史时点理解。
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,9 @@
|
||||
# B4b 发现
|
||||
|
||||
- M3 当前为 f7705d96 基础上的未提交实现;完整清单封存 SHA 78ce1df774be24d37c8a3e55d943a24f9cab2dd5e955e1743e631723f0bddd00,旧证据根只读。
|
||||
- 桥协议 mnemo.project-memory.v1:持久 producer UUID + 既有 projectId;current 必须有独立持久用户确认,不能从本地 current 推断。
|
||||
- unknown sync 保留原包原键,仅 lookup;每物理尝试 fresh manifest + validate,不复用旧 snapshot 续命。
|
||||
- 本地 source-checked-project-memory 与授权远端镜像独立工件/标签;recallContextDigest 与选入整条的 digest/refs 分记。
|
||||
- 新本机合成宿主提供实际 Mnemo API 及 Chat 接收器、原字节 physical ID,不代表真实模型质量/成本/收益。
|
||||
- 实际授权拒绝的上游错误码不一定以 PROJECT_ 开头;不能据命名空间把已收到的HTTP拒绝误分类为传输异常。错误对象现在保留HTTP状态/原码,原文在独立工件。
|
||||
- 人工确认必须显示与确认摘要同一个快照的完整review;不能把父层稍旧的memory正文和桥view的新digest组合。服务端同一view产出review/digest,客户端显示这份review并可展开所有纳入摘要的元数据。
|
||||
@@ -0,0 +1,31 @@
|
||||
# B4b 进展
|
||||
|
||||
2026-11-08:接续授权与 M3 工作树已核对。已读协议及合成宿主指南,接下来冻结 after-M3/合同输入并开始源码实现。尚未启动合成宿主、尚未请求模型或正式服务。
|
||||
|
||||
已核 24 输入,初次冻结脚本错误地未解码中文 file URL,生成到 percent-encoded 兄弟目录;已验证该目录确为本轮生成物后整体移入本轮证据根 initial-url-path-mistake,原 24 输入及原回执全保留。改为 fileURLToPath 后再执行会按预期发现 planner 已开发变化;第二轮部分输出保留,不补签当前源码。原始基线见 initial-url-path-mistake/input-freeze.json。
|
||||
|
||||
协议首轮因复制路径尚未纠正 ENOENT;第二轮 4/5,通过 transport UNKNOWN 等边界,仅向量 fixture 的 context 键名误读失败(实际 recall.entries/contextDigest),原 protocol-first.tap 保留。只读复核指出 transport 默认409风险,已改为 status0,并验证网络断线转 PROJECT_SYNC_UNKNOWN。
|
||||
|
||||
本轮合成 Mnemo 前台宿主已启动,instanceId 595ac6c4-2eec-4037-8f06-b36a5b194777,端口54961,session14414;只读取明确本轮 private 文件,120分钟自动结束,收尾用受保护shutdown。
|
||||
|
||||
协议修正后5/5。第一原生轮 native-first 的接收器baseUrl少/v1,全部模型请求404,原22案例的7个负向通过也不能采纳为原生物理通过。首轮残余故障不能影响重跑,已受保护shutdown原宿主并保留全观察。修驱动:显式/v1、让query含实际合成来源词项、每次必须看到新物理接收。首构建npm子shell使用系统Node16失败,固定PATH后完整build-second通过;保留首日志。
|
||||
|
||||
只读复核两风险已修:同Store全项目共享锁,防整份metadata交错覆盖;删除同时清桥可恢复运行包正文/responseText,保留原键/事件和正文摘要/墓碑,已发送工件不改。lookup支持原摘要校验已擦正文的历史包。客户端令牌在宿主内存,重启/恢复不加载授权。
|
||||
|
||||
第二原生轮有实际接收但query词项覆盖不足,远端合法返回空召回,10/22通过不代签正向路径;按原词项实现核对后,新驱动用确切来源标题作为实际用户主题,不改Mnemo/检索。新宿主探针5/5(默认关闭/取用/撤权/unknown sync/unknown physical)。N19整条子集探针通过;N22驱动误写不支持priority策略原失败保留,改现有round-robin后两物理尝试通过。
|
||||
|
||||
最终新宿主 instance60f59342-6965-4451-8c7c-8dd7840559e3(59752/session65322)native-final单轮24/24通过,含真实31秒等待、摘要变更、两尝试、恢复墓碑,旧工件SHA保持。N25网页/隐藏Electron/真实伴随网关窄屏待验。
|
||||
|
||||
独立构建首次缺复制mobile.html失败;build-2补齐后发现node_modules目录联接会使esbuild记录 ../../../../node_modules 路径,编译SHA确实不同,不能称字节一致。原候选全保留,build-3改为真实依赖副本(不声称本轮clean npm ci),再重建对比。
|
||||
|
||||
build-3成功:916输入与30编译逐SHA一致;最后相关103/103自然完成,0失败/取消/跳过。网页首轮已实际完成候选确认、刷新和请求工件查看,但窄屏时任务侧栏因布局切换收起,驱动未重新点击真实入口,故N25仍未通过。原失败/DOM/截图/已发送正文保留;仅修驱动重新打开任务记录后另冻新轮,产品源码未变。
|
||||
|
||||
Mnemo侧补充v3只读核对器及native-probe独立6尝试核对,另登记supplement-v3,不覆盖原24输入/v2八文件快照;宿主实现及七V1契约未变。未向来源线程回信,不能用probe替代最后整轮验收。
|
||||
|
||||
网页第二轮实际通过连接/确认、同次正文/工件与390回执、真实伴随HTTPS窄屏,但撤权诊断断言失败:协议catch按PROJECT_前缀误把Mnemo自己的错误命名空间当网络异常,原HTTP状态和正文仍在原召回工件中。已改专用错误类型区分收到的HTTP拒绝和传输失败;新增HTTP401非本地前缀回归。另发现父层记忆列表与桥view不是同一快照,已让桥view返回与确认摘要严格同源的review,界面直接展示这份完整修订及元数据,旧摘要仍拒绝。两模块修正专项12/12通过;build-4重新916输入/30编译exact,相关新轮及完整24原生重验进行中,旧build-3和原生final全部保持不覆盖。
|
||||
|
||||
修正后native-after-review完整24/24,30新Run/31实际物理接收;只读离线审计逐条核对本机SQLite/实际全文及336原工件SHA通过。网页ui-web-third、实际隐藏Electron ui-electron-first各4组/10图/2次新本机合成请求通过,两端最终8图实看,真实伴随HTTPS校验证书路径通过,物理iPhone未验。四个本轮Mnemo宿主均正常关闭,最后59752/所属Node退出、private删除;host-final保留最终观测/库/69份物理正文。四个候选PID及端口监听0、所属Electron0,未访问正式服务。
|
||||
|
||||
related-after-review和related-last-isolated均103/104,同一400ms reasoning-only夹具接收端请求数0;单项孤立1/1通过。复核发现前两条命令漏了仓库package/CI固定的--test-concurrency=2,使用Node默认并发。原失败保持,不改测试或产品阈值;另用仓库标准并发2执行最后完整七文件相关轮related-standard-2,尚待自然结束。
|
||||
|
||||
related-standard-2最后自然完成104/104,179360.3584ms,0取消/跳过/todo;不再重复已通过范围。机器矩阵和QA/使用说明完成,25项工程证据与916输入/30编译同源。Mnemo侧独立24案例、37账本尝试/31物理原文、336原工件及最终宿主观测逐字节通过,原回执复制independent-after-review;其all25CasesVerified:false保持,N25由实际界面独立列证。所有历史失败、真实模型/Purge/长期/物理手机未验项保持。本轮未提交/推送/发行/正式升级,下一为M3最后产品身份与统一发行,不自动执行。
|
||||
@@ -0,0 +1,19 @@
|
||||
# B4b 原生单项目记忆桥
|
||||
|
||||
接续 M3 后的明确授权:实现默认关闭的 Mnemo 桥,在新合成项目、会话和本机 HTTP 宿主验证。不以资料登记代替实现。
|
||||
|
||||
## 阶段
|
||||
|
||||
- [complete] 1. 核对 after-M3 输入、七契约及八宿主输入;冻结独立证据根(原URL路径错误及第二轮漂移拒绝均保留)。
|
||||
- [complete] 2. 持久生产者/项目配置、人工确认、递增来源/墓碑、严格摘要/HTTP及 unknown 原键核对。
|
||||
- [complete] 3. 原生规划器与每 physical attempt 的实时核对/validate;分记 recall 与选入证据。
|
||||
- [complete] 4. 可用配置/完整修订确认/回执界面实现;最终916输入/30编译独立重建exact,网页/隐藏Electron各4组/10图、伴随HTTPS及1280/390通过。
|
||||
- [complete] 5. 最后原生后端24/24和N25两端界面通过;相关按仓库固定并发2为104/104,37账本尝试及336原工件独立核对,合成宿主正常关闭,报告/机器矩阵/回顾与封存完成。
|
||||
|
||||
## 边界
|
||||
|
||||
M3 的未提交源码保持,.qa/m3-release-2026-11-08 已封存,不再追加。本轮不提交、推送、发布、安装或正式升级;不查正式 health、不读真实 Key、不调用真实模型或旧工程、不重放旧请求、不委派、不改 Mnemo 源码。仅允许读本轮合成宿主 connection.private.json 和控制接口。桥限制仅作用于可选桥,保持 unrestricted/passthrough/preserve、持久 ask/auto/full 及原预算/计费/历史权威。历史 W/AD/AE/AG 网络/unknown、真实交付、重复读取、R/W 真机边界、Purge 收益待验保持。
|
||||
|
||||
## 错误与观察
|
||||
|
||||
- 接续初次多文件输出超预算截断:改为小块读取,不将未进入上下文内容当已审读。
|
||||
@@ -0,0 +1,422 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"at": "2026-11-11T16:58:33.956Z",
|
||||
"batch": "m4-claude-correction-2026-11-12",
|
||||
"evidenceRoot": ".qa/m4-claude-correction-2026-11-12",
|
||||
"driverCommit": "97f132d3cb2611dbbb758cde0a1399485690b7da",
|
||||
"driverSha256": "52c8a5ba231f7b2caac8bb2bf9b81dcee5903a735c0a56a135c78abe8015b5d8",
|
||||
"sourceNotice": {
|
||||
"path": "D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-12-claude-proxy-correction-user.json",
|
||||
"bytes": 6801,
|
||||
"sha256": "0a215dcbb4a18e08d25cb7afc5b28822dbc41a7ddc3c0dd2c72872494defe663",
|
||||
"scope": "Public human correction notice only; keys absent; source not written."
|
||||
},
|
||||
"humanReportedCause": "wrong_proxy_key",
|
||||
"historicalMismatchCauseIndependentlyProven": false,
|
||||
"originalM4": {
|
||||
"requests": 40,
|
||||
"reportedTotalTokens": 629106,
|
||||
"unknownRequests": 0,
|
||||
"sealedFiles": 541,
|
||||
"manifestSha256": "558c9b548559798279d7f8e61deb3b7344920753a53dbf4e84eec374c6f3a3df",
|
||||
"mismatchResponses": 2,
|
||||
"configuredModelPreviouslyListed": false,
|
||||
"oldResultsRewritten": false
|
||||
},
|
||||
"formal": {
|
||||
"version": "1.25.13",
|
||||
"pid": 63892,
|
||||
"instanceId": "75209495-f90a-4552-9a90-81a680d50c7c",
|
||||
"configRevision": 130,
|
||||
"registryRevision": 186,
|
||||
"selected": {
|
||||
"type": "pool",
|
||||
"id": "35b99d94-2df1-4f25-afc3-25d063b6ff84"
|
||||
}
|
||||
},
|
||||
"currentConnection": {
|
||||
"id": "9ae348a5-d0a6-42ed-8664-0da5f9effe34",
|
||||
"name": "apikey-Claude",
|
||||
"protocol": "anthropic-messages",
|
||||
"model": "claude-opus-5-5",
|
||||
"revision": 9,
|
||||
"credentialRevision": 2,
|
||||
"hasKey": true,
|
||||
"nativeParameters": {}
|
||||
},
|
||||
"postSealPublicChanges": {
|
||||
"configRevision": [
|
||||
130,
|
||||
130
|
||||
],
|
||||
"registryRevision": [
|
||||
182,
|
||||
186
|
||||
],
|
||||
"selectedConnectionRevision": [
|
||||
8,
|
||||
9
|
||||
],
|
||||
"credentialRevision": [
|
||||
1,
|
||||
2
|
||||
],
|
||||
"definitionChangedFields": [
|
||||
"revision",
|
||||
"updatedAt",
|
||||
"credentialBindingRevision",
|
||||
"upstreamModel",
|
||||
"credentialRevision",
|
||||
"modelSpecification",
|
||||
"capabilities"
|
||||
],
|
||||
"addedConnections": [
|
||||
{
|
||||
"id": "02f67692-e3ce-4add-bde1-977c2162e0af",
|
||||
"name": "apikey-Claude · claude-fable-5-2",
|
||||
"model": "claude-fable-5-2",
|
||||
"revision": 1
|
||||
}
|
||||
],
|
||||
"observationRegistryRevision": 186,
|
||||
"actorOfAdditionalChanges": "unestablished; preserved as observed; no connection restored"
|
||||
},
|
||||
"catalog": {
|
||||
"complete": true,
|
||||
"models": 14,
|
||||
"selectedListed": true,
|
||||
"source": {
|
||||
"path": "catalog.json",
|
||||
"bytes": 24517,
|
||||
"sha256": "78986dce5418965b1fff60c5e3a3b13a17601dd847bc7a3c73ab293c9ca71d50"
|
||||
},
|
||||
"evidenceScope": "Public product catalogue with supplied source metadata; no invented raw HTTP capture."
|
||||
},
|
||||
"preflights": [
|
||||
{
|
||||
"scope": "text",
|
||||
"projectId": "6afd88f9-0e28-41e0-ae2f-48dbf6e9f6c6",
|
||||
"sessionId": "aa93b8ae-31b0-4cfc-8b28-b6db44d2e2b1",
|
||||
"taskId": "22f44c69-5b5e-4d23-9298-9207e7550e34",
|
||||
"runId": "ba7da2e3-a51a-491c-a0fe-c5092843e7f7",
|
||||
"requestId": "b714df32-711a-4307-b7ec-1b68e1000950",
|
||||
"attemptId": "81f1bf6d-0cae-4a7a-b482-3b1381f4a606",
|
||||
"wireAttemptId": "a0c136b9-74f3-4648-9a40-6fa77272dad6",
|
||||
"upstreamRequestId": "02acb49c-9ca1-4fdc-9a4c-c45c71d4b000",
|
||||
"connectionId": "9ae348a5-d0a6-42ed-8664-0da5f9effe34",
|
||||
"connectionRevision": 9,
|
||||
"credentialRevision": 2,
|
||||
"requestedModel": "claude-opus-5-5",
|
||||
"responseModel": null,
|
||||
"modelIdentityStatus": "incomplete",
|
||||
"requestUrl": "https://api.apikey.fan/v1/messages",
|
||||
"responseUrl": "https://api.apikey.fan/v1/messages",
|
||||
"httpStatus": 403,
|
||||
"responseFormat": "json_http_error_body",
|
||||
"responseCaptureComplete": true,
|
||||
"responseBytes": 188,
|
||||
"originalUpstreamError": {
|
||||
"error": {
|
||||
"message": "该分组仅允许使用 Claude Code 官方客户端。 This group is restricted to Claude Code clients (/v1/messages only)",
|
||||
"type": "permission_error"
|
||||
},
|
||||
"type": "error"
|
||||
},
|
||||
"requestOutputFields": {
|
||||
"max_tokens": 128000
|
||||
},
|
||||
"outputFieldSource": "official_model_spec",
|
||||
"streamRequested": true,
|
||||
"sseObserved": false,
|
||||
"sseMessageStopObserved": false,
|
||||
"usageStatus": "unknown",
|
||||
"usage": null,
|
||||
"actualTotalTokens": null,
|
||||
"actualCost": null,
|
||||
"startedAt": "2026-11-11T16:54:45.390Z",
|
||||
"finishedAt": "2026-11-11T16:54:45.684Z",
|
||||
"elapsedMs": 294,
|
||||
"actualToolsExecuted": 0,
|
||||
"requestSource": {
|
||||
"attemptId": "81f1bf6d-0cae-4a7a-b482-3b1381f4a606",
|
||||
"artifactId": "4d9cfdd4-a322-43cf-bac7-7364b5f2f869",
|
||||
"kind": "request",
|
||||
"path": "text/captures/81f1bf6d-0cae-4a7a-b482-3b1381f4a606-request.json",
|
||||
"bytes": 313,
|
||||
"sha256": "b1f387575f4cdb2025ce242cb207f1bf3f160623df84bb2b2285c869aae9b285"
|
||||
},
|
||||
"responseSource": {
|
||||
"attemptId": "81f1bf6d-0cae-4a7a-b482-3b1381f4a606",
|
||||
"artifactId": "a9f7983c-9b5d-4f2d-b1b1-390f54ae95d9",
|
||||
"kind": "response",
|
||||
"path": "text/captures/81f1bf6d-0cae-4a7a-b482-3b1381f4a606-response.sse",
|
||||
"bytes": 188,
|
||||
"sha256": "a9a64c798135516fcbfa0f036deeb21e38954e8d442a0dbc063fbc2905018ef8"
|
||||
},
|
||||
"artifactSuffixNote": "Original collector used configured streaming to select .sse suffix. These original bytes are a complete JSON HTTP 403 error body, not SSE."
|
||||
},
|
||||
{
|
||||
"scope": "tools",
|
||||
"projectId": "6afd88f9-0e28-41e0-ae2f-48dbf6e9f6c6",
|
||||
"sessionId": "d19148aa-c9fe-4b8b-ae8c-34d83944c119",
|
||||
"taskId": "5decb3cb-ea08-4d14-aa28-2f697c37ebb0",
|
||||
"runId": "b0468e4b-2841-4111-8915-fbf6f0688f8a",
|
||||
"requestId": "30510069-50fd-4e88-a35e-5a835799f921",
|
||||
"attemptId": "9efaf829-859b-482a-a401-1c02531cd84a",
|
||||
"wireAttemptId": "cec8475b-59cc-449f-b7d3-c2253eabc0e5",
|
||||
"upstreamRequestId": "65157030-cf66-414a-bb14-8db6f4b70acc",
|
||||
"connectionId": "9ae348a5-d0a6-42ed-8664-0da5f9effe34",
|
||||
"connectionRevision": 9,
|
||||
"credentialRevision": 2,
|
||||
"requestedModel": "claude-opus-5-5",
|
||||
"responseModel": null,
|
||||
"modelIdentityStatus": "incomplete",
|
||||
"requestUrl": "https://api.apikey.fan/v1/messages",
|
||||
"responseUrl": "https://api.apikey.fan/v1/messages",
|
||||
"httpStatus": 403,
|
||||
"responseFormat": "json_http_error_body",
|
||||
"responseCaptureComplete": true,
|
||||
"responseBytes": 188,
|
||||
"originalUpstreamError": {
|
||||
"error": {
|
||||
"message": "该分组仅允许使用 Claude Code 官方客户端。 This group is restricted to Claude Code clients (/v1/messages only)",
|
||||
"type": "permission_error"
|
||||
},
|
||||
"type": "error"
|
||||
},
|
||||
"requestOutputFields": {
|
||||
"max_tokens": 128000
|
||||
},
|
||||
"outputFieldSource": "official_model_spec",
|
||||
"streamRequested": true,
|
||||
"sseObserved": false,
|
||||
"sseMessageStopObserved": false,
|
||||
"usageStatus": "unknown",
|
||||
"usage": null,
|
||||
"actualTotalTokens": null,
|
||||
"actualCost": null,
|
||||
"startedAt": "2026-11-11T16:54:46.069Z",
|
||||
"finishedAt": "2026-11-11T16:54:46.473Z",
|
||||
"elapsedMs": 404,
|
||||
"actualToolsExecuted": 0,
|
||||
"requestSource": {
|
||||
"attemptId": "9efaf829-859b-482a-a401-1c02531cd84a",
|
||||
"artifactId": "fb534e24-b3a7-4f54-8316-55bc61082bce",
|
||||
"kind": "request",
|
||||
"path": "tools/captures/9efaf829-859b-482a-a401-1c02531cd84a-request.json",
|
||||
"bytes": 648,
|
||||
"sha256": "74fa50f9b8d49d9acbe83f7548602644a4a5fb0c5a55cc8d1084e96105e2b439"
|
||||
},
|
||||
"responseSource": {
|
||||
"attemptId": "9efaf829-859b-482a-a401-1c02531cd84a",
|
||||
"artifactId": "d760a4de-d24f-4ef9-a870-619cc15f3871",
|
||||
"kind": "response",
|
||||
"path": "tools/captures/9efaf829-859b-482a-a401-1c02531cd84a-response.sse",
|
||||
"bytes": 188,
|
||||
"sha256": "a9a64c798135516fcbfa0f036deeb21e38954e8d442a0dbc063fbc2905018ef8"
|
||||
},
|
||||
"artifactSuffixNote": "Original collector used configured streaming to select .sse suffix. These original bytes are a complete JSON HTTP 403 error body, not SSE."
|
||||
}
|
||||
],
|
||||
"requests": 2,
|
||||
"maximumPhysicalRequests": 2,
|
||||
"requestsWithKnownUsage": 0,
|
||||
"unknownRequests": 2,
|
||||
"reportedTokenSum": 0,
|
||||
"actualTotalTokens": null,
|
||||
"actualTotalCost": null,
|
||||
"costScope": "Complete HTTP403 errors provide no model usage or cost. Zero sum of available reports is not zero upstream use or billing.",
|
||||
"qualification": {
|
||||
"catalog": true,
|
||||
"text": false,
|
||||
"tools": false,
|
||||
"streaming": false,
|
||||
"responseModelIdentity": false,
|
||||
"requestContractFieldsVerified": true,
|
||||
"modelWeightsAuthenticated": false
|
||||
},
|
||||
"blocker": {
|
||||
"origin": "upstream",
|
||||
"httpStatus": 403,
|
||||
"type": "permission_error",
|
||||
"message": "该分组仅允许使用 Claude Code 官方客户端。 This group is restricted to Claude Code clients (/v1/messages only)",
|
||||
"scope": "Provider group accepts Claude Code clients only; standard Workbench Messages requests were rejected.",
|
||||
"remedy": "Supplier must grant ordinary Messages API client permission or provide a group that accepts such clients. No client impersonation or extra model requests were attempted."
|
||||
},
|
||||
"driverValidation": {
|
||||
"at": "2026-11-11T16:52:49.879Z",
|
||||
"checks": [
|
||||
{
|
||||
"name": "declared complete synthetic Messages receipt",
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"name": "missing mandatory output field",
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"name": "old local 2048 is not the declared maximum",
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"name": "missing message_stop",
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"name": "invalid SSE JSON",
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"name": "physical request count exceeds case scope",
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"name": "observed tool cannot have executed",
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"name": "unknown usage cannot qualify",
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"name": "original sealed text mismatch remains failed",
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"name": "original sealed tools mismatch remains failed",
|
||||
"passed": true
|
||||
}
|
||||
],
|
||||
"syntheticCases": 8,
|
||||
"originalMismatchReceiptsReadOnly": 2,
|
||||
"realModelRequests": 0,
|
||||
"oldRootsWritten": false,
|
||||
"keysReadOrResaved": false
|
||||
},
|
||||
"audit": {
|
||||
"success": true,
|
||||
"at": "2026-11-11T16:56:12.964Z",
|
||||
"pid": 63892,
|
||||
"instanceId": "75209495-f90a-4552-9a90-81a680d50c7c",
|
||||
"version": "1.25.13",
|
||||
"originalSessions": 137,
|
||||
"originalModelRequests": 418,
|
||||
"originalArtifacts": 5576,
|
||||
"checkedOriginalRows": 58298,
|
||||
"addedRows": {
|
||||
"events": 36,
|
||||
"metadata": 0,
|
||||
"model_attempts": 2,
|
||||
"outbox": 36,
|
||||
"projects": 1,
|
||||
"records": 16,
|
||||
"sessions": 2,
|
||||
"sqlite_sequence": 0
|
||||
},
|
||||
"sequenceChanges": [
|
||||
{
|
||||
"name": "events",
|
||||
"before": 24175,
|
||||
"after": 24211,
|
||||
"increment": 36
|
||||
},
|
||||
{
|
||||
"name": "model_attempts",
|
||||
"before": 418,
|
||||
"after": 420,
|
||||
"increment": 2
|
||||
}
|
||||
],
|
||||
"newArtifacts": 7,
|
||||
"currentSevenConnectionDefinitionsPoolsSelection": "exact",
|
||||
"originalExportsAndArtifactBytes": "exact",
|
||||
"originalMetadataAndCredentialRow": "exact SHA; no credential decrypted or copied",
|
||||
"originalProjects": "exact after declared single front insertion",
|
||||
"requests": 2,
|
||||
"inspectorClosed": true,
|
||||
"keysReadOrResaved": false,
|
||||
"noRestorationOfOldHumanSettings": true,
|
||||
"source": {
|
||||
"path": "audit-44ea125a-b46f-4355-bbac-5026b017c2e7/result.json",
|
||||
"bytes": 1097,
|
||||
"sha256": "17ff401656a6353808d2a0163e58d7d0f39fe6ad00751fca8d53804605041950"
|
||||
},
|
||||
"afterSource": {
|
||||
"path": "audit-44ea125a-b46f-4355-bbac-5026b017c2e7/after.json",
|
||||
"bytes": 16447360,
|
||||
"sha256": "9298d9a572e99499dc0254afc0a60bbb238023c07c73d62f1de6abe9473db743"
|
||||
}
|
||||
},
|
||||
"sources": [
|
||||
{
|
||||
"path": "observation.json",
|
||||
"bytes": 1048438,
|
||||
"sha256": "0c0ec165c3dfcc776cc3106f9b1b930cd655104302d5a41330aed274a7f17621"
|
||||
},
|
||||
{
|
||||
"path": "frozen.json",
|
||||
"bytes": 18119119,
|
||||
"sha256": "2710d091eb84aa7f32e1be6872796f51f428082026ca940e392ad131e36ebb9b"
|
||||
},
|
||||
{
|
||||
"path": "claim.json",
|
||||
"bytes": 181,
|
||||
"sha256": "7761a2644549cfd140ca32f9d5d6e3af23339c4912a1855706884a1d95b05ea2"
|
||||
},
|
||||
{
|
||||
"path": "catalog.json",
|
||||
"bytes": 24517,
|
||||
"sha256": "78986dce5418965b1fff60c5e3a3b13a17601dd847bc7a3c73ab293c9ca71d50"
|
||||
},
|
||||
{
|
||||
"path": "catalog-cancel.json",
|
||||
"bytes": 24,
|
||||
"sha256": "5e5626d3e3b68f8f5ba416eb1c04e6f1de60ff931a236cf7bdb48199ff750561"
|
||||
},
|
||||
{
|
||||
"path": "result.json",
|
||||
"bytes": 18687,
|
||||
"sha256": "95caf7e791299218eedbbe24bcdf66cc116f243534bce39d82317084df91987c"
|
||||
},
|
||||
{
|
||||
"path": "canonical-request-identities.json",
|
||||
"bytes": 5799,
|
||||
"sha256": "ece30dca26b2b404860144fe3c76f1caf3d7d9684f4fbf514fd427c1214336f8"
|
||||
},
|
||||
{
|
||||
"path": "driver-validation.json",
|
||||
"bytes": 1069,
|
||||
"sha256": "da2cb65067aafc18218f8573f3a2191a30a72d9a208bdee207041ee08eaccd7e"
|
||||
},
|
||||
{
|
||||
"path": "driver-source.mjs",
|
||||
"bytes": 24449,
|
||||
"sha256": "52c8a5ba231f7b2caac8bb2bf9b81dcee5903a735c0a56a135c78abe8015b5d8"
|
||||
},
|
||||
{
|
||||
"path": "human-correction-source.json",
|
||||
"bytes": 6801,
|
||||
"sha256": "0a215dcbb4a18e08d25cb7afc5b28822dbc41a7ddc3c0dd2c72872494defe663"
|
||||
}
|
||||
],
|
||||
"boundaries": {
|
||||
"keysReadOrResaved": false,
|
||||
"oldM4RootWritten": false,
|
||||
"modelPreferencesRestored": false,
|
||||
"currentSevenConnectionsPreserved": true,
|
||||
"toolsExecuted": 0,
|
||||
"engineeringOrPurgeReplayed": false,
|
||||
"productSourceChanged": false,
|
||||
"releaseOrRestart": false,
|
||||
"messagesSentToSource": 0,
|
||||
"authorizationExpanded": false
|
||||
},
|
||||
"unverified": [
|
||||
"Corrected Claude text/tool/streaming/response identity",
|
||||
"Real upstream model weights",
|
||||
"Purge and cache/experience/role causal benefits",
|
||||
"Mnemo production binding and semantic benefit",
|
||||
"Physical iPhone and long term stability",
|
||||
"Historical R/W live outlier boundary"
|
||||
]
|
||||
}
|
||||
@@ -0,0 +1,28 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"verifiedAt": "2026-11-11T17:00:06.064Z",
|
||||
"evidenceRoot": ".qa/m4-claude-correction-2026-11-12",
|
||||
"files": 41,
|
||||
"bytes": 36284086,
|
||||
"manifest": {
|
||||
"path": "MANIFEST.json",
|
||||
"bytes": 6867,
|
||||
"sha256": "9b8a1cfcc4a8842e5c751990fdfc4a274831bf4dcafe73c557514c992490f2e5"
|
||||
},
|
||||
"originalSealReceipt": {
|
||||
"path": "seal.json",
|
||||
"bytes": 346,
|
||||
"sha256": "aacca201e3b5822ffcda363bc51a55951279f2e55ffb337d4aa8751933f96b73"
|
||||
},
|
||||
"originalReceiptBytesMeaning": "Total member bytes; the manifest file length is explicitly separate above.",
|
||||
"originalM4ManifestSha256": "558c9b548559798279d7f8e61deb3b7344920753a53dbf4e84eec374c6f3a3df",
|
||||
"requests": 2,
|
||||
"requestsWithKnownUsage": 0,
|
||||
"unknownRequests": 2,
|
||||
"actualTotalTokens": null,
|
||||
"actualTotalCost": null,
|
||||
"memberBytesAndSha256Verified": true,
|
||||
"currentSevenConnectionsPreserved": true,
|
||||
"keysReadOrResaved": false,
|
||||
"engineeringOrPurgeReplayed": false
|
||||
}
|
||||
@@ -0,0 +1,39 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"at": "2026-11-11T17:01:40.374Z",
|
||||
"remote": "ssh://git@gitatlas.simosen.cn:2222/mosen/Agent-Workbench.git",
|
||||
"remoteWeb": "https://gitatlas.simosen.cn/mosen/Agent-Workbench",
|
||||
"branch": "feat/task-agreement-1.25",
|
||||
"force": false,
|
||||
"priorRemoteHead": "c8c36c27ee8f026b02324bd0a81d55eb8c443447",
|
||||
"acceptanceCommit": "5ca65a45fb48d654f9235c409d1ab9ee771089c9",
|
||||
"remoteHead": "5ca65a45fb48d654f9235c409d1ab9ee771089c9",
|
||||
"fetchedHead": "5ca65a45fb48d654f9235c409d1ab9ee771089c9",
|
||||
"verifiedCommits": [
|
||||
"97f132d3cb2611dbbb758cde0a1399485690b7da",
|
||||
"5ca65a45fb48d654f9235c409d1ab9ee771089c9"
|
||||
],
|
||||
"evidenceRoot": ".qa/m4-claude-correction-2026-11-12",
|
||||
"evidenceManifestSha256": "9b8a1cfcc4a8842e5c751990fdfc4a274831bf4dcafe73c557514c992490f2e5",
|
||||
"sealedFiles": 41,
|
||||
"sealedBytes": 36284086,
|
||||
"requests": 2,
|
||||
"requestsWithKnownUsage": 0,
|
||||
"unknownRequests": 2,
|
||||
"actualTotalTokens": null,
|
||||
"actualTotalCost": null,
|
||||
"catalogQualified": true,
|
||||
"modelPreflightsQualified": false,
|
||||
"upstreamHttpStatus": 403,
|
||||
"upstreamType": "permission_error",
|
||||
"oldM4Requests": 40,
|
||||
"oldM4SealedFiles": 541,
|
||||
"oldM4EvidencePreserved": true,
|
||||
"currentSevenConnectionsPreserved": true,
|
||||
"keysReadOrResaved": false,
|
||||
"engineeringOrPurgeReplayed": false,
|
||||
"productSourceChanged": false,
|
||||
"releaseOrRestart": false,
|
||||
"remoteRunnerResult": "unknown",
|
||||
"note": "Receipt and completed pointers are committed next; final remote HEAD is verified separately. Original sealed evidence is not edited."
|
||||
}
|
||||
@@ -0,0 +1,27 @@
|
||||
{
|
||||
"success": true,
|
||||
"remote": "ssh://git@gitatlas.simosen.cn:2222/mosen/Agent-Workbench.git",
|
||||
"web": "https://gitatlas.simosen.cn/mosen/Agent-Workbench",
|
||||
"branch": "feat/task-agreement-1.25",
|
||||
"originalRemoteHead": "79e9f15f8969de8d1d080a19868c3b1218cd06eb",
|
||||
"publishedHead": "958f31cd7ec1811469896de341af72197cea04f7",
|
||||
"remoteHead": "958f31cd7ec1811469896de341af72197cea04f7",
|
||||
"fetchedHead": "958f31cd7ec1811469896de341af72197cea04f7",
|
||||
"commits": [
|
||||
"211117e06d1a41f0de4d042b4afe9f94fee65104",
|
||||
"5b51151f59983cb2854cec45f26a8deae22256b1",
|
||||
"41e59768977f9f538f814beef353a50dc620010e",
|
||||
"0415e9d5ea852b28e9361a26ac25b33bb5be586a",
|
||||
"958f31cd7ec1811469896de341af72197cea04f7"
|
||||
],
|
||||
"allCommitsVerified": true,
|
||||
"productSourceCommit": "41e59768977f9f538f814beef353a50dc620010e",
|
||||
"version": "1.25.14",
|
||||
"pid": 29884,
|
||||
"restoredKeys": 7,
|
||||
"requests": 420,
|
||||
"manifestSha256": "031b0da22a2aae37ce017e9e248cc6711041377ecbd1ee083001d1277ae2e1d1",
|
||||
"pushLogSha256": "0969e287e901d56cb2db6e9406d21e673f85a665bb2519cf19174d43c4881f46",
|
||||
"verifiedAt": "2026-11-12T01:58:32.925Z",
|
||||
"method": "Ordinary incremental push, explicit remote ref and re-fetch; each new commit ancestry verified. No force, credential access or model requests."
|
||||
}
|
||||
@@ -0,0 +1,920 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"startedAt": "2026-11-07T17:17:33.246Z",
|
||||
"scope": "Human-authorized ordinary incremental source and acceptance push to the existing GitAtlas branch. No release, model request, key read, formal-service operation, persistent runner installation or source-thread reply.",
|
||||
"branch": "feat/task-agreement-1.25",
|
||||
"remote": "ssh://git@gitatlas.simosen.cn:2222/mosen/Agent-Workbench.git",
|
||||
"remoteWeb": "https://gitatlas.simosen.cn/mosen/Agent-Workbench",
|
||||
"baseline": "470d694831c315c034e3f36891e4f3faf94cbe0f",
|
||||
"force": false,
|
||||
"remoteRunnerResult": "unknown",
|
||||
"remoteArtifactUploadResult": "unknown",
|
||||
"evidenceManifestSha256": "aaeaf694a1cb2444ee8bdfd5ec74ccc768c5659a38fdd6b3be36d6d69abe4951",
|
||||
"localHead": "44b6a09b06570448690e61c8077537f7ccfea299",
|
||||
"committedSourceReview": {
|
||||
"files": [
|
||||
{
|
||||
"file": ".gitattributes",
|
||||
"bytes": 14459
|
||||
},
|
||||
{
|
||||
"file": ".gitea/workflows/ci.yml",
|
||||
"bytes": 1701
|
||||
},
|
||||
{
|
||||
"file": "AGENTS.md",
|
||||
"bytes": 101770
|
||||
},
|
||||
{
|
||||
"file": "docs/CI-SOURCES.md",
|
||||
"bytes": 1967
|
||||
},
|
||||
{
|
||||
"file": "docs/CI.md",
|
||||
"bytes": 4922
|
||||
},
|
||||
{
|
||||
"file": "docs/DEVELOPMENT-ROADMAP-OFFICIAL-FIRST.md",
|
||||
"bytes": 27410
|
||||
},
|
||||
{
|
||||
"file": "docs/M2-CI-PLAN.md",
|
||||
"bytes": 19195
|
||||
},
|
||||
{
|
||||
"file": "docs/M3-RELEASE-PLAN.md",
|
||||
"bytes": 4736
|
||||
},
|
||||
{
|
||||
"file": "docs/PLAN-1.25.md",
|
||||
"bytes": 18252
|
||||
},
|
||||
{
|
||||
"file": "docs/QA-M2-CI-OFFICIAL-CONTRACTS.md",
|
||||
"bytes": 7807
|
||||
},
|
||||
{
|
||||
"file": "docs/TIANSHU-ADOPTION-PLAN.md",
|
||||
"bytes": 26690
|
||||
},
|
||||
{
|
||||
"file": "docs/integration/DOMESTIC-ENTRY-READONLY-2026-11-08.md",
|
||||
"bytes": 3291
|
||||
},
|
||||
{
|
||||
"file": "docs/integration/MNEMO-PROJECT-MEMORY-BRIDGE-V1-READONLY-2026-11-07.md",
|
||||
"bytes": 7904
|
||||
},
|
||||
{
|
||||
"file": "docs/m2-ci-2026-11-07.json",
|
||||
"bytes": 48776
|
||||
},
|
||||
{
|
||||
"file": "findings.md",
|
||||
"bytes": 292969
|
||||
},
|
||||
{
|
||||
"file": "package-lock.json",
|
||||
"bytes": 105888
|
||||
},
|
||||
{
|
||||
"file": "package.json",
|
||||
"bytes": 4361
|
||||
},
|
||||
{
|
||||
"file": "progress.md",
|
||||
"bytes": 452645
|
||||
},
|
||||
{
|
||||
"file": "scripts/ci-desktop-smoke.mjs",
|
||||
"bytes": 11420
|
||||
},
|
||||
{
|
||||
"file": "scripts/ci-environment.json",
|
||||
"bytes": 1022
|
||||
},
|
||||
{
|
||||
"file": "scripts/ci-python-requirements.txt",
|
||||
"bytes": 762
|
||||
},
|
||||
{
|
||||
"file": "scripts/ci-utils.mjs",
|
||||
"bytes": 9949
|
||||
},
|
||||
{
|
||||
"file": "scripts/ci.mjs",
|
||||
"bytes": 13921
|
||||
},
|
||||
{
|
||||
"file": "scripts/official-contract-check.mjs",
|
||||
"bytes": 8356
|
||||
},
|
||||
{
|
||||
"file": "scripts/package-source.mjs",
|
||||
"bytes": 6838
|
||||
},
|
||||
{
|
||||
"file": "scripts/ui-navigation.mjs",
|
||||
"bytes": 1092
|
||||
},
|
||||
{
|
||||
"file": "scripts/verify-ci-workflow.mjs",
|
||||
"bytes": 2856
|
||||
},
|
||||
{
|
||||
"file": "scripts/verify-official-contracts.mjs",
|
||||
"bytes": 2333
|
||||
},
|
||||
{
|
||||
"file": "scripts/verify-replay-fixtures.mjs",
|
||||
"bytes": 3323
|
||||
},
|
||||
{
|
||||
"file": "src/server/runtime.mjs",
|
||||
"bytes": 114262
|
||||
},
|
||||
{
|
||||
"file": "task_plan.md",
|
||||
"bytes": 181953
|
||||
},
|
||||
{
|
||||
"file": "tests/anthropic-messages.test.mjs",
|
||||
"bytes": 26683
|
||||
},
|
||||
{
|
||||
"file": "tests/api.test.mjs",
|
||||
"bytes": 13826
|
||||
},
|
||||
{
|
||||
"file": "tests/catalog-refresh.test.mjs",
|
||||
"bytes": 8662
|
||||
},
|
||||
{
|
||||
"file": "tests/ci-pipeline.test.mjs",
|
||||
"bytes": 9342
|
||||
},
|
||||
{
|
||||
"file": "tests/engine-context.test.mjs",
|
||||
"bytes": 6744
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/INDEX.json",
|
||||
"bytes": 68888
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/README.md",
|
||||
"bytes": 1161
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/engine-context/r1-off.json",
|
||||
"bytes": 96984
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/engine-context/r1-on.json",
|
||||
"bytes": 82839
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/engine-context/r2-off.json",
|
||||
"bytes": 93789
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/engine-context/r2-on.json",
|
||||
"bytes": 94971
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/model-reservation/r.json",
|
||||
"bytes": 66121
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/LICENSE",
|
||||
"bytes": 11341
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/NOTICE",
|
||||
"bytes": 210
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/asset-table.js",
|
||||
"bytes": 16684
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/capability-protocol.js",
|
||||
"bytes": 3268
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/child-process-hide.mjs",
|
||||
"bytes": 3010
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/continue-retry.js",
|
||||
"bytes": 11758
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/core.js",
|
||||
"bytes": 342481
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/default-prompt-inject.en.md",
|
||||
"bytes": 29427
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/default-prompt-inject.md.bak",
|
||||
"bytes": 24455
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/default-prompt-inject.zh.md",
|
||||
"bytes": 25727
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/desktop.js",
|
||||
"bytes": 72311
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/drill-grant.js",
|
||||
"bytes": 1740
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/extract-asar.js",
|
||||
"bytes": 16091
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/hide-console.js",
|
||||
"bytes": 11804
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/hooks-deny.js",
|
||||
"bytes": 8054
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/host-clean.js",
|
||||
"bytes": 1043
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/host.js",
|
||||
"bytes": 652
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/identity.js",
|
||||
"bytes": 41793
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/index.js",
|
||||
"bytes": 108136
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/mnemon-settings.js",
|
||||
"bytes": 1661
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/net-scope.js",
|
||||
"bytes": 32199
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/official-update.js",
|
||||
"bytes": 1832
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/official-update.wscript",
|
||||
"bytes": 4815
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/phase1-keep.js",
|
||||
"bytes": 904
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/preset-default.js",
|
||||
"bytes": 6056
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/reapply-cli.js",
|
||||
"bytes": 177
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/reapply.js",
|
||||
"bytes": 13653
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/client.js",
|
||||
"bytes": 277332
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/evidence-ledger.mjs",
|
||||
"bytes": 19026
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/index.js",
|
||||
"bytes": 24345
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/ip-utils.js",
|
||||
"bytes": 6151
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/platform-config.js",
|
||||
"bytes": 22339
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/preset-composition.js",
|
||||
"bytes": 16861
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/report-replay.js",
|
||||
"bytes": 10490
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/role-section.js",
|
||||
"bytes": 6010
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/schema.js",
|
||||
"bytes": 33554
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/score-rules.js",
|
||||
"bytes": 63287
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/settings.js",
|
||||
"bytes": 3305
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/skill-availability.js",
|
||||
"bytes": 17783
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/skill-fixes.js",
|
||||
"bytes": 38363
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/store-core.js",
|
||||
"bytes": 340826
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/store.js",
|
||||
"bytes": 2969
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/tools.js",
|
||||
"bytes": 136501
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/ui.js",
|
||||
"bytes": 50785
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/redteam/validate.js",
|
||||
"bytes": 3538
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/restart-desktop.js",
|
||||
"bytes": 14602
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/restart-web.js",
|
||||
"bytes": 2903
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/rewind.js",
|
||||
"bytes": 20894
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/rules.js",
|
||||
"bytes": 17682
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/settle-policy.js",
|
||||
"bytes": 2024
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/skills.js",
|
||||
"bytes": 28551
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/surface.js",
|
||||
"bytes": 3838
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/table-key.js",
|
||||
"bytes": 915
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/table-read.js",
|
||||
"bytes": 567
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/uninstall-restart.js",
|
||||
"bytes": 1653
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/uninstall.js",
|
||||
"bytes": 8632
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/update.js",
|
||||
"bytes": 39046
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/web.js",
|
||||
"bytes": 3385
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/lib/zip-extract.js",
|
||||
"bytes": 5699
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-replay/original-purge/upstream/package.json",
|
||||
"bytes": 2623
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-sources/SOURCE.json",
|
||||
"bytes": 4159
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-sources/sources/action-checkout.yml",
|
||||
"bytes": 4593
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-sources/sources/action-setup-node.yml",
|
||||
"bytes": 2614
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-sources/sources/action-setup-python.yml",
|
||||
"bytes": 2364
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-sources/sources/action-upload-artifact.yml",
|
||||
"bytes": 1150
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-sources/sources/gitea-126-comparison.body",
|
||||
"bytes": 87919
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-sources/sources/gitea-126-quickstart.body",
|
||||
"bytes": 73731
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/ci-sources/sources/gitea-126-runner.body",
|
||||
"bytes": 158980
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/official-contracts/INDEX.json",
|
||||
"bytes": 1493
|
||||
},
|
||||
{
|
||||
"file": "tests/fixtures/official-contracts/registry-2026-11-07-ci1.json",
|
||||
"bytes": 113278
|
||||
},
|
||||
{
|
||||
"file": "tests/model-reservation.test.mjs",
|
||||
"bytes": 16831
|
||||
},
|
||||
{
|
||||
"file": "tests/official-contracts-ci.test.mjs",
|
||||
"bytes": 4207
|
||||
},
|
||||
{
|
||||
"file": "tests/original-purge.test.mjs",
|
||||
"bytes": 6031
|
||||
},
|
||||
{
|
||||
"file": "tests/python-api.test.mjs",
|
||||
"bytes": 11037
|
||||
},
|
||||
{
|
||||
"file": "tests/python-checks.test.mjs",
|
||||
"bytes": 18240
|
||||
},
|
||||
{
|
||||
"file": "tests/reasoning-settings.test.mjs",
|
||||
"bytes": 15820
|
||||
},
|
||||
{
|
||||
"file": "tests/reminder-api.test.mjs",
|
||||
"bytes": 3944
|
||||
},
|
||||
{
|
||||
"file": "tests/research.test.mjs",
|
||||
"bytes": 18504
|
||||
},
|
||||
{
|
||||
"file": "tests/source-package.test.mjs",
|
||||
"bytes": 5925
|
||||
},
|
||||
{
|
||||
"file": "tests/task-control.test.mjs",
|
||||
"bytes": 15716
|
||||
}
|
||||
],
|
||||
"forbiddenPaths": 0,
|
||||
"knownPrivateCredentialSignatures": 0,
|
||||
"scope": "Only changed committed files and known private-key/API-token signatures; no real configured credentials inspected."
|
||||
},
|
||||
"remoteBefore": "470d694831c315c034e3f36891e4f3faf94cbe0f",
|
||||
"incrementalCommits": [
|
||||
"a43262c11247247b50f4b4d23d69ca7c2016a0c7",
|
||||
"af42b949ef5fff11124dad7ae3a0eb4cf8954414",
|
||||
"0b4f1de4395e42999ade74fdaa7869919db883b6",
|
||||
"36c6be4ff8e298471eec2da5e2f01e128a787fba",
|
||||
"49dbc2f74ce99a6464746e31807f3e2c8ab6565b",
|
||||
"880c88c7fc1ddb81cd552cffa7680afacb68d63b",
|
||||
"50d799c6b494cb9a02cf3f8148485f480ed465b7",
|
||||
"d5207dc6cc6953c2b9dced2fe820645291c1f2a8",
|
||||
"87ca07517589612839eb492d3851ef82e40a6c97",
|
||||
"4a03ec528a40eb2c8edf22f3fd9d0bf6181f70fb",
|
||||
"1a9c51b15a653e91d99baa67127e5186eee19ef6",
|
||||
"70d1d8c7a3fd17ca654ab4c5232c4e971b777a1b",
|
||||
"3d649b3b697f1123155fdddd28f719f6f1c74ecf",
|
||||
"44b6a09b06570448690e61c8077537f7ccfea299"
|
||||
],
|
||||
"remoteAfter": "44b6a09b06570448690e61c8077537f7ccfea299",
|
||||
"remoteReachableCommitsMatched": [
|
||||
"a43262c11247247b50f4b4d23d69ca7c2016a0c7",
|
||||
"af42b949ef5fff11124dad7ae3a0eb4cf8954414",
|
||||
"0b4f1de4395e42999ade74fdaa7869919db883b6",
|
||||
"36c6be4ff8e298471eec2da5e2f01e128a787fba",
|
||||
"49dbc2f74ce99a6464746e31807f3e2c8ab6565b",
|
||||
"880c88c7fc1ddb81cd552cffa7680afacb68d63b",
|
||||
"50d799c6b494cb9a02cf3f8148485f480ed465b7",
|
||||
"d5207dc6cc6953c2b9dced2fe820645291c1f2a8",
|
||||
"87ca07517589612839eb492d3851ef82e40a6c97",
|
||||
"4a03ec528a40eb2c8edf22f3fd9d0bf6181f70fb",
|
||||
"1a9c51b15a653e91d99baa67127e5186eee19ef6",
|
||||
"70d1d8c7a3fd17ca654ab4c5232c4e971b777a1b",
|
||||
"3d649b3b697f1123155fdddd28f719f6f1c74ecf",
|
||||
"44b6a09b06570448690e61c8077537f7ccfea299"
|
||||
],
|
||||
"eachIncrementalCommitReachableFromVerifiedRemoteHead": true,
|
||||
"exactRemoteSha": true,
|
||||
"ordinaryIncrementalPush": true,
|
||||
"success": true,
|
||||
"sealedEvidenceUnchanged": true,
|
||||
"finishedAt": "2026-11-07T17:17:42.450Z",
|
||||
"records": [
|
||||
{
|
||||
"id": "branch",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"branch",
|
||||
"--show-current"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 60776,
|
||||
"startedAt": "2026-11-07T17:17:33.246Z",
|
||||
"finishedAt": "2026-11-07T17:17:33.284Z",
|
||||
"durationMs": 38,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\branch.stdout.log",
|
||||
"bytes": 25,
|
||||
"sha256": "8336ba986a24d729010501b2a78b41335dbd032b68ba30c2a4bc6cbf95477b08"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\branch.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "remote",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"remote",
|
||||
"get-url",
|
||||
"origin"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 38224,
|
||||
"startedAt": "2026-11-07T17:17:33.288Z",
|
||||
"finishedAt": "2026-11-07T17:17:33.327Z",
|
||||
"durationMs": 39,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\remote.stdout.log",
|
||||
"bytes": 61,
|
||||
"sha256": "b4b134ddb31b08e603822d3fd89781f9995a154c7d8d8e71aff9298ef3d0a804"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\remote.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "status",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"status",
|
||||
"--porcelain"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 53992,
|
||||
"startedAt": "2026-11-07T17:17:33.328Z",
|
||||
"finishedAt": "2026-11-07T17:17:33.395Z",
|
||||
"durationMs": 67,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\status.stdout.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\status.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "head",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"rev-parse",
|
||||
"HEAD"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 63796,
|
||||
"startedAt": "2026-11-07T17:17:33.396Z",
|
||||
"finishedAt": "2026-11-07T17:17:33.435Z",
|
||||
"durationMs": 39,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\head.stdout.log",
|
||||
"bytes": 41,
|
||||
"sha256": "f19874796b1b0a0976c2f4f8b7f09a0e0333fbe1e11ecbbae9b8fa1d45b32eba"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\head.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "changed-inputs",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"diff",
|
||||
"--name-only",
|
||||
"470d694831c315c034e3f36891e4f3faf94cbe0f",
|
||||
"44b6a09b06570448690e61c8077537f7ccfea299"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 67452,
|
||||
"startedAt": "2026-11-07T17:17:33.436Z",
|
||||
"finishedAt": "2026-11-07T17:17:33.475Z",
|
||||
"durationMs": 39,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\changed-inputs.stdout.log",
|
||||
"bytes": 6326,
|
||||
"sha256": "e253d7604fd4260399fe6f8a0788f1cc82b67399786102625f4df310c293b558"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\changed-inputs.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "remote-before",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"ls-remote",
|
||||
"--heads",
|
||||
"origin",
|
||||
"refs/heads/feat/task-agreement-1.25"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 50236,
|
||||
"startedAt": "2026-11-07T17:17:39.099Z",
|
||||
"finishedAt": "2026-11-07T17:17:39.833Z",
|
||||
"durationMs": 734,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\remote-before.stdout.log",
|
||||
"bytes": 77,
|
||||
"sha256": "c4da2ddf27c0e6fb7d6b9d5ccd7b8c5b21c96d5263a641bdf0161068e49f6d17"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\remote-before.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "ancestor",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"merge-base",
|
||||
"--is-ancestor",
|
||||
"470d694831c315c034e3f36891e4f3faf94cbe0f",
|
||||
"44b6a09b06570448690e61c8077537f7ccfea299"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 3236,
|
||||
"startedAt": "2026-11-07T17:17:39.834Z",
|
||||
"finishedAt": "2026-11-07T17:17:39.864Z",
|
||||
"durationMs": 30,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\ancestor.stdout.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\ancestor.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "commit-list",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"rev-list",
|
||||
"--reverse",
|
||||
"470d694831c315c034e3f36891e4f3faf94cbe0f..44b6a09b06570448690e61c8077537f7ccfea299"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 60680,
|
||||
"startedAt": "2026-11-07T17:17:39.869Z",
|
||||
"finishedAt": "2026-11-07T17:17:39.898Z",
|
||||
"durationMs": 29,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\commit-list.stdout.log",
|
||||
"bytes": 574,
|
||||
"sha256": "856c591252e8f3fdfefd0770d55a0d81a0e8b21953c3fde90712d84815df3e51"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\commit-list.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "push",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"push",
|
||||
"origin",
|
||||
"HEAD:refs/heads/feat/task-agreement-1.25"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 37848,
|
||||
"startedAt": "2026-11-07T17:17:39.898Z",
|
||||
"finishedAt": "2026-11-07T17:17:41.761Z",
|
||||
"durationMs": 1863,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\push.stdout.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\push.stderr.log",
|
||||
"bytes": 206,
|
||||
"sha256": "e8bdfc5b95c247f37e2734aac247a05fa35640794d54e4ddc4c4d48636e5a4b9"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "remote-after",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"ls-remote",
|
||||
"--heads",
|
||||
"origin",
|
||||
"refs/heads/feat/task-agreement-1.25"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 60316,
|
||||
"startedAt": "2026-11-07T17:17:41.762Z",
|
||||
"finishedAt": "2026-11-07T17:17:42.040Z",
|
||||
"durationMs": 278,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\remote-after.stdout.log",
|
||||
"bytes": 77,
|
||||
"sha256": "8e70dffb6d3b87220769891f415261c22f3af15934971fea4dd8717af0830e98"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\remote-after.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "fetch-verified-branch",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"fetch",
|
||||
"--no-tags",
|
||||
"origin",
|
||||
"refs/heads/feat/task-agreement-1.25"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 33868,
|
||||
"startedAt": "2026-11-07T17:17:42.040Z",
|
||||
"finishedAt": "2026-11-07T17:17:42.370Z",
|
||||
"durationMs": 330,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\fetch-verified-branch.stdout.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\fetch-verified-branch.stderr.log",
|
||||
"bytes": 120,
|
||||
"sha256": "abf2c4ffbfce48ccc01b3957f187e55fca9794614f42d7139014528c037d6c76"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "fetched-head",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"rev-parse",
|
||||
"FETCH_HEAD"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 37444,
|
||||
"startedAt": "2026-11-07T17:17:42.371Z",
|
||||
"finishedAt": "2026-11-07T17:17:42.410Z",
|
||||
"durationMs": 39,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\fetched-head.stdout.log",
|
||||
"bytes": 41,
|
||||
"sha256": "f19874796b1b0a0976c2f4f8b7f09a0e0333fbe1e11ecbbae9b8fa1d45b32eba"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\fetched-head.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": "fetched-commits",
|
||||
"executable": "git",
|
||||
"args": [
|
||||
"rev-list",
|
||||
"--reverse",
|
||||
"470d694831c315c034e3f36891e4f3faf94cbe0f..FETCH_HEAD"
|
||||
],
|
||||
"cwd": "D:\\Ai-project\\REA逆向\\Agent-Workbench",
|
||||
"pid": 39484,
|
||||
"startedAt": "2026-11-07T17:17:42.411Z",
|
||||
"finishedAt": "2026-11-07T17:17:42.449Z",
|
||||
"durationMs": 38,
|
||||
"exitCode": 0,
|
||||
"signal": null,
|
||||
"interrupted": false,
|
||||
"spawnError": null,
|
||||
"stdout": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\fetched-commits.stdout.log",
|
||||
"bytes": 574,
|
||||
"sha256": "856c591252e8f3fdfefd0770d55a0d81a0e8b21953c3fde90712d84815df3e51"
|
||||
},
|
||||
"stderr": {
|
||||
"file": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-ci-close-2026-11-08\\push\\run-contract-mw6o25kq-7fbe3800\\evidence\\fetched-commits.stderr.log",
|
||||
"bytes": 0,
|
||||
"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
@@ -0,0 +1,126 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"scope": "Authorized ordinary incremental source/acceptance push; no release or production install",
|
||||
"branch": "feat/task-agreement-1.25",
|
||||
"remote": "ssh://git@gitatlas.simosen.cn:2222/mosen/Agent-Workbench.git",
|
||||
"remoteWeb": "https://gitatlas.simosen.cn/mosen/Agent-Workbench",
|
||||
"sourceCommits": [
|
||||
"71f1805561e6082011ed4f0c70bb38f4546413c5",
|
||||
"9e1feeebb3eca5c82bfa9efb4f32e68a73628df4",
|
||||
"ee97aaae7389e0aff7f89eced8d06ecce7d4f982"
|
||||
],
|
||||
"acceptanceCommit": "1a8975769bf2430272dbfac6a82b8de4422394a4",
|
||||
"startedAt": "2026-11-07T12:37:51.092Z",
|
||||
"finishedAt": "2026-11-07T12:37:53.025Z",
|
||||
"node": "D:\\Ai-project\\REA逆向\\tools\\node-v24.22.1-win-x64\\node.exe",
|
||||
"localHead": "1a8975769bf2430272dbfac6a82b8de4422394a4",
|
||||
"remoteBefore": "bdfeb29c85b3231bfd7bc24b8b5f49d40a3beebf",
|
||||
"remoteAfter": "1a8975769bf2430272dbfac6a82b8de4422394a4",
|
||||
"exactRemoteSha": true,
|
||||
"ordinaryIncrementalPush": true,
|
||||
"force": false,
|
||||
"records": [
|
||||
{
|
||||
"args": [
|
||||
"branch",
|
||||
"--show-current"
|
||||
],
|
||||
"code": 0,
|
||||
"signal": null,
|
||||
"stdout": "feat/task-agreement-1.25\n",
|
||||
"stderr": ""
|
||||
},
|
||||
{
|
||||
"args": [
|
||||
"remote",
|
||||
"get-url",
|
||||
"origin"
|
||||
],
|
||||
"code": 0,
|
||||
"signal": null,
|
||||
"stdout": "ssh://git@gitatlas.simosen.cn:2222/mosen/Agent-Workbench.git\n",
|
||||
"stderr": ""
|
||||
},
|
||||
{
|
||||
"args": [
|
||||
"status",
|
||||
"--porcelain"
|
||||
],
|
||||
"code": 0,
|
||||
"signal": null,
|
||||
"stdout": "",
|
||||
"stderr": ""
|
||||
},
|
||||
{
|
||||
"args": [
|
||||
"rev-parse",
|
||||
"HEAD"
|
||||
],
|
||||
"code": 0,
|
||||
"signal": null,
|
||||
"stdout": "1a8975769bf2430272dbfac6a82b8de4422394a4\n",
|
||||
"stderr": ""
|
||||
},
|
||||
{
|
||||
"args": [
|
||||
"ls-remote",
|
||||
"--heads",
|
||||
"origin",
|
||||
"refs/heads/feat/task-agreement-1.25"
|
||||
],
|
||||
"code": 0,
|
||||
"signal": null,
|
||||
"stdout": "bdfeb29c85b3231bfd7bc24b8b5f49d40a3beebf\trefs/heads/feat/task-agreement-1.25\n",
|
||||
"stderr": ""
|
||||
},
|
||||
{
|
||||
"args": [
|
||||
"merge-base",
|
||||
"--is-ancestor",
|
||||
"bdfeb29c85b3231bfd7bc24b8b5f49d40a3beebf",
|
||||
"1a8975769bf2430272dbfac6a82b8de4422394a4"
|
||||
],
|
||||
"code": 0,
|
||||
"signal": null,
|
||||
"stdout": "",
|
||||
"stderr": ""
|
||||
},
|
||||
{
|
||||
"args": [
|
||||
"push",
|
||||
"origin",
|
||||
"HEAD:refs/heads/feat/task-agreement-1.25"
|
||||
],
|
||||
"code": 0,
|
||||
"signal": null,
|
||||
"stdout": "",
|
||||
"stderr": "remote: . Processing 1 references \nremote: Processed 1 references in total \nTo ssh://gitatlas.simosen.cn:2222/mosen/Agent-Workbench.git\n bdfeb29c..1a897576 HEAD -> feat/task-agreement-1.25\n"
|
||||
},
|
||||
{
|
||||
"args": [
|
||||
"ls-remote",
|
||||
"--heads",
|
||||
"origin",
|
||||
"refs/heads/feat/task-agreement-1.25"
|
||||
],
|
||||
"code": 0,
|
||||
"signal": null,
|
||||
"stdout": "1a8975769bf2430272dbfac6a82b8de4422394a4\trefs/heads/feat/task-agreement-1.25\n",
|
||||
"stderr": ""
|
||||
}
|
||||
],
|
||||
"success": true,
|
||||
"realModelRequests": 0,
|
||||
"realCredentialsRead": 0,
|
||||
"productionWrites": 0,
|
||||
"baseEvidenceRootWrites": 0,
|
||||
"evidenceRoot": ".qa/m2-native-cache-state-2026-11-07",
|
||||
"manifestSha256": "c887364e54fe3f68e5f432155db87517fb8413966f5786156bedceddadbf2217",
|
||||
"closingRoot": ".qa/m2-native-cache-state-close-2026-11-07",
|
||||
"receiptAndCompletionCommitThenOrdinaryPush": true,
|
||||
"subsequentFinalCheckStoredOutsideCommittedReceipt": true,
|
||||
"productionDeployed": false,
|
||||
"officialServiceHealthRechecked": false,
|
||||
"restartedProduction": false,
|
||||
"nativeCacheOrPurgeBenefitVerified": false
|
||||
}
|
||||
@@ -0,0 +1,41 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"at": "2026-11-11T16:46:45.130Z",
|
||||
"remote": "ssh://git@gitatlas.simosen.cn:2222/mosen/Agent-Workbench.git",
|
||||
"remoteWeb": "https://gitatlas.simosen.cn/mosen/Agent-Workbench",
|
||||
"branch": "feat/task-agreement-1.25",
|
||||
"force": false,
|
||||
"priorRemoteHead": "2f2b459c1860b4b3ba6670ce6c5a80dab3b4cb4e",
|
||||
"acceptanceCommit": "e132601b9f9b5421d36b63cf69b550059b06b7a1",
|
||||
"remoteHead": "e132601b9f9b5421d36b63cf69b550059b06b7a1",
|
||||
"fetchedHead": "e132601b9f9b5421d36b63cf69b550059b06b7a1",
|
||||
"verifiedCommits": [
|
||||
"30ad156abfe2cfaee92e8b990f519b3562add446",
|
||||
"0054ea282180b20061797b00cf5122ee8baa1df4",
|
||||
"7152d654e37aebb6f8ceeb6190a189b7bc5d3b35",
|
||||
"b8eca3b5a9ef3cdbbc935dcf8f97b2839535cd2e",
|
||||
"52accafe6bfee904d4351519e81d565cdbb7e852",
|
||||
"ba755b0d42963aec81d8977b79f02706521b41b7",
|
||||
"e132601b9f9b5421d36b63cf69b550059b06b7a1"
|
||||
],
|
||||
"evidenceManifestSha256": "558c9b548559798279d7f8e61deb3b7344920753a53dbf4e84eec374c6f3a3df",
|
||||
"sealedFiles": 541,
|
||||
"sealedBytes": 134884464,
|
||||
"realModelRequests": 40,
|
||||
"reportedTotalTokens": 629106,
|
||||
"unknownRequests": 0,
|
||||
"engineeringResults": [
|
||||
"node",
|
||||
"react",
|
||||
"python"
|
||||
],
|
||||
"oldClaudeIdentityMismatchRetained": true,
|
||||
"postSealHumanCorrection": "Separate fresh validation; old evidence and key settings not restored.",
|
||||
"realKeyReadOrResaved": false,
|
||||
"purgeExecuted": false,
|
||||
"productSourceChanged": false,
|
||||
"releaseOrRestart": false,
|
||||
"remoteRunnerResult": "unknown",
|
||||
"ignoredEvidenceRoot": ".qa/m4-live-baseline-2026-11-11",
|
||||
"note": "This receipt and completion pointers are committed and their final remote SHA verified separately; sealed evidence is not modified."
|
||||
}
|
||||
@@ -0,0 +1,29 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"at": "2026-11-10T19:45:47.107Z",
|
||||
"remote": "ssh://git@gitatlas.simosen.cn:2222/mosen/Agent-Workbench.git",
|
||||
"remoteWeb": "https://gitatlas.simosen.cn/mosen/Agent-Workbench",
|
||||
"branch": "feat/task-agreement-1.25",
|
||||
"force": false,
|
||||
"priorRemoteHead": "e306f40caef0592e6f0fb61c8aac007dc8685451",
|
||||
"sourceCommit": "35a6d6d9b25f2301bc9b8844b9e6af2a6d3bb94f",
|
||||
"acceptanceCommit": "1e3097004e63e07821700b924449b1d763789f0a",
|
||||
"remoteHead": "1e3097004e63e07821700b924449b1d763789f0a",
|
||||
"fetchedHead": "1e3097004e63e07821700b924449b1d763789f0a",
|
||||
"verifiedCommits": [
|
||||
"35a6d6d9b25f2301bc9b8844b9e6af2a6d3bb94f",
|
||||
"1e3097004e63e07821700b924449b1d763789f0a"
|
||||
],
|
||||
"formalVersion": "1.25.13",
|
||||
"formalHostPid": 63892,
|
||||
"formalInstallerRuns": 1,
|
||||
"formalInstallerExitCode": 0,
|
||||
"formalInstallerElapsedMs": 351789,
|
||||
"releaseIdentity": "docs/mnemo-1.25.13-release-identity.json",
|
||||
"evidenceManifestSha256": "611c4c40d9c28b09649d8a041505d1684242f436b7c8b972b7cfbb0c9c65260c",
|
||||
"realModelRequests": 0,
|
||||
"realKeyReadOrResaved": false,
|
||||
"remoteRunnerResult": "unknown",
|
||||
"sourceProgramSetupUploadedToGit": false,
|
||||
"note": "Ordinary incremental source, packaging fix and acceptance push verified. Program/source ZIP and Setup stay local, not in Git. This receipt and completion pointers are committed afterward and their final remote SHA is checked separately without editing the sealed QA evidence."
|
||||
}
|
||||
@@ -0,0 +1,20 @@
|
||||
# 智枢国内入口:Workbench 只读登记
|
||||
|
||||
2026-11-08,来自三项目协调入口的可选接口参考。主任务继续 [I27/I28 CI](../M2-CI-PLAN.md) 和 [M3 发行路线](../M3-RELEASE-PLAN.md),不因本资料启动模型请求、工程派发、身份注册或服务对接;本登记不向来源线程发送回复。
|
||||
|
||||
四个指定文件 SHA 全部匹配。离线从第三窗口和 closeout 核对:60 个 request_key 唯一、每个 submit_posts=1 且 passed=true;180 份采样、transport_errors=0;closeout.inputSha256 与该窗口原件 SHA 一致,60 对源任务/本机任务匹配、localSourceMatches 和 localContentDeleted 均为60。没有复跑来源方的 HTTPS、34 API、12 Edge 或900秒真实请求。
|
||||
|
||||
| 固定参考原件 | SHA256 |
|
||||
| --- | --- |
|
||||
| [DOMESTIC-ENTRY-acceptance.md](D:/Ai-project/models/docs/platform/reviews/DOMESTIC-ENTRY-acceptance.md) | `2585924921b4681724ed5b5eff2d0ddc73539fb87e6fd133cb488db5059cd446` |
|
||||
| [DOMESTIC_ENTRY.md](D:/Ai-project/models/docs/platform/DOMESTIC_ENTRY.md) | `040c3581b0dfd7264dd844f5894d6b2674260a4d6458e47172d00f427d37239e` |
|
||||
| [第三窗口](D:/Ai-project/models/logs/domestic-entry-soak-20261107-third.json) | `04148cd5fcd6866333edc7e680c176fb77237bcfe278c8a6c2bbebea1279825f` |
|
||||
| [第三窗口 closeout](D:/Ai-project/models/logs/domestic-entry-soak-20261107-third-closeout.json) | `f08b630f24618a77720805d54d625589fc51efbeba605b321f5d313bbe0d888c` |
|
||||
|
||||
来源 summary 为60/60、两种有效率1、p95 7.828秒;前后用量对象一致,month_micro=2957、unknown_calls=3,closeout.newModelSubmissions=0、identityRevoked=true、originBinding.schema=6。50元月预算来自来源方约定,不能变成工作台默认模型额度。原件保持在来源目录,本地只保存 SHA/字段名和必要计数/归属复算结果,不复制完整身份或请求正文。
|
||||
|
||||
可选 API 为 [ai-api.mnemo.cn/v1](https://ai-api.mnemo.cn/v1),控制台 [ai-console.mnemo.cn](https://ai-console.mnemo.cn)。资料声明仍转香港 schema6 唯一控制器/原键/账本/删除权威,本机仍 schema7;网关无 POST 自动重试,未知只按原键核对。新增 SDK profile 不自动改 Workbench/Harness 域名、模型 Key、选线、预算或权限;共享域名不合并三项目认证、数据库或项目身份。连接池退役和短 query-only WAL读取是来源方本次反例的修复,不证明工作台供应商、缓存或工程交付收益。
|
||||
|
||||
旧手机49/60发起、48确认、125/180、p95 48.437失败以及 SSE105/108缺 wire、新域真实移动网络待验、模型质量门槛和完整境内权威迁移未完成均保持;新窗口不覆盖它们。后续实际互操作必须另冻新合成范围和同请求证据,不重发旧模型/工程请求,不触发来源方手机票据。
|
||||
|
||||
[Mnemo 记忆桥只读反馈](MNEMO-PROJECT-MEMORY-BRIDGE-V1-READONLY-2026-11-07.md) 的 SHA `9efcc256836ef7ddd2ae24c22d7da4a60df2c8833b76778402e6faeef8e73796` 与协调方读取一致;仍只有七原件和五组离线向量,原生 SDK/服务、203/485/33回归和工作台桥接待验。本批记录在 `.qa/m2-ci-2026-11-07/pre-final-review-1/domestic-source-hashes.json` 与 `domestic-facts-review.json`,不改正在执行 CI 的冻结程序/测试、正式服务或历史 unknown。
|
||||
@@ -0,0 +1,57 @@
|
||||
# 三项目协调:有界摘要与任务边界只读登记
|
||||
|
||||
2026-11-10(Asia/Shanghai)。接收协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的合并依赖更新。只读核对公开报告、元数据和来源 SHA,修正上次说明中的状态错误;没有执行来源测试、模型、清理或恢复。本次不改变 [M3 统一发行](../M3-RELEASE-PLAN.md)顺序,也不新增发行或服务操作。
|
||||
|
||||
## 原件与核对范围
|
||||
|
||||
- [智枢有界摘要与团队上下文回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-BOUNDED-COMPACTION-acceptance.md):33189 字节,SHA256 `c01de21ce90a644afcac77f97def0c172d7e7b580f2074f2d38f6921815e7cfd`,与来信及协调引用一致。
|
||||
- [Mnemo 动态协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-09T1851.json):按一次读取的原字节快照绑定,具体长度、SHA、completedAtUtc 和投递状态见本次 registration;不推断其后续最终字节或接收确认。
|
||||
- 14 份公开元数据报告的长度、SHA 全部匹配;30 条来源 SHA 绑定核对一致,涉及 23 个不同路径。报告内来源字段、计划 SHA 和对应路径一并核对;绑定的源脚本只计算摘要,没有执行。
|
||||
- 原任务/请求键、云端/本机/清理映射复算一致。本轮 20 条长窗加 14 条诊断共 **34 个唯一新任务映射**;26 条诊断清理范围包含旧 12 条及本轮 14 条,不能全部算作本轮新增。
|
||||
|
||||
证据在 [registration.json](../../.qa/coordination-readonly-2026-11-10T0257/registration.json) 和独立 [MANIFEST.json](../../.qa/coordination-readonly-2026-11-10T0257/MANIFEST.json),包含协调/验收原字节、14 份公开元数据、更正前文档和核对脚本。19 个清单成员均已复算;旧 M3/B4b 清单及上次协调的 12 个封存成员未变。
|
||||
|
||||
这里只核对公开报告和元数据,未读取私有会话、原始 SSE、来源数据库或真实 Key。来源报告中的 checks 标记通过不等于本次重新运行了这些检查。checkpoint manifest 的 SHA 已匹配,3092 文件/70443856 字节及旧 2146 份证据保持仍按来源报告登记,没有重算全部外部成员。
|
||||
|
||||
## 改善证据与未通过项
|
||||
|
||||
来源已安装有界摘要拥有者提示和条件 Team 上下文,保留原八节、1024 摘要硬限、58 工具及 `MAX_TOKENS` 停止保护;原摘要没有后处理。七个新提示来源绑定诊断的公开判定均通过:第五原前缀 373 token,五摘要链依次为 270/264/264/275/338 token,最终原 14 字段/类型/值且无新增字段。首候选三请求中的重复 JSON 失败仍保留;旧前缀本轮也曾以 826 token 完成,不能称原截断稳定复现。
|
||||
|
||||
Team 旧提示 1 组、新提示 3 组均答对;无条件 `Use list_agents` 已改为当前任务需要协调时才使用,但旧错误未稳定复现,不能据此认定唯一根因已解决。诊断是 `stream=false`、零实际工具执行,不替代普通 Desktop 完整长任务验收。
|
||||
|
||||
**BC-03 仍为 `failed_quality`,P8 仍进行中**;BC-01/02/04 为 passed,ISS-029/ISS-038 及来源六项 partial、八个阻断保留。
|
||||
|
||||
| 窗口 | 公开任务与持久历史元数据 | 原任务质量判定 | wire 证据范围 |
|
||||
| --- | --- | --- | --- |
|
||||
| v7 | 10 原任务全部 succeeded / attempt1;1 份 221 token 摘要;原任务 32 项、历史 29 项 | 应答 REVIEW_02 却调用 list_agents;被原串行 scope 拒绝,返回工具未执行 | 专用 72 项认证全部上游有序事件;不认证被拒完成帧成功下发 |
|
||||
| v8 | 10 原任务全部 succeeded / attempt1;1 份 347 token 摘要;原任务 32 项、历史 29 项 | 应答 REVIEW_02 却答历史 REVIEW_03;末次 28294 prompt_tokens | 103 项完整 cloud/node/local ordered wire |
|
||||
|
||||
两个窗口都未到原 24 段、至少两份摘要和最终精确事实门槛。原标准仍为 24×6000 字、14 事实、17 bundles/58 工具、32K 上下文/2048 输出、4096 保留、0.875 阈值和 1024 摘要硬限。这些是来源冻结实验条件,不是 Workbench 新增额度。传输 passed 不能改签质量,七个摘要诊断通过也不代表完整长窗口通过或原第五摘要已在 Desktop 重新验收。
|
||||
|
||||
## 审计、清理和累计保护
|
||||
|
||||
来源报告 v7/v8 均先审计、后清理;本次仅核对其公开回执,没有观察或重做运行时顺序。v7 通用 wire 工具因拒绝行缺 response_sha256 发生 KeyError 的失败保留,后续专用审计的范围如上;v8 完整审计另列。两窗清理 verify-only 各 23 项、零 DELETE;合并诊断 verify-only 95 项,26 条任务的 deletePosts 复算为 0,未重发删除。
|
||||
|
||||
**v6 完整逐帧审计缺证继续保留,不能由 v7/v8 补签。** 历史 AB-04 `failed_context`、NL-03 `failed_quality`、NL-04 `partial` 和原网络/unknown 均不改写,不恢复已删正文补证。
|
||||
|
||||
本轮新增为 **34 任务/5 身份**;来源累计保护为 **160 任务/18 身份**,是另一个范围。保护报告 20 项检查标记 passed,18 身份 revoked 字段全部为 1,原 12 表保持;这些是公开元数据结论,不认证外部副本删除。香港 schema6 单一权威、本机 schema7 独立、月 50 元、历史 2957 micro CNY 和 3 unknown 保持。新增厂商付费 0 不清零历史费用或调用。
|
||||
|
||||
普通 Desktop 的 8 配置哈希及 8 项恢复核验按来源报告登记,没有重新启动或读取配置正文。159/150/442 三组相关测试可能重叠,不相加;14 Python、9 构建门、207 集成通过加 1 skip、11 许可、43 文档检查及 lint 退出 0 也是来源结论。本轮首候选重复 JSON、文档/核验器失败、原 Windows/POSIX 夹具失败和 lint 既有警告均保持,不能称全平台快照全过。
|
||||
|
||||
源方下一步是固定 v8 当前任务与历史边界的小规模来源对照,取得区分证据后才另冻完整窗口;不盲跑 v9。本协调不要求 Workbench 执行该步骤。长期业务质量、30 秒 SLA、完整境内迁主和其他 P8 门槛仍未认证。
|
||||
|
||||
## 上次登记的文字更正
|
||||
|
||||
[上次说明](MNEMO-LONG-AB-COORDINATION-READONLY-2026-11-10.md)把捕获前读取的协调状态误写成最终封存状态。实际封存协调为 16906 字节,SHA256 `be567d2a7da185cf0933aa1dea620dc9b64c65906d88f3c1057e8a0f61a4b92d`;其 `completedAtUtc=2026-11-09T16:58:39.132Z`,投递状态为 `tool_accepted_recipient_acknowledgment_pending`,不是 null / `dispatch_attempt_in_progress`。
|
||||
|
||||
已核对原 registration 与快照一致,只更正文档。旧快照、脚本、清单、registration 及收尾回执全部保留;更正前 7487 字节文档以原 SHA `a619f99f2d6d5e8fbc0768b7f30ee1770c1badeca73807999e5ef883d8cacefc` 存入新证据根。旧收尾回执仍对应更正前文档,新收尾另绑定更正后的说明。工具接受不等于接收确认,也不表示核验后续最终协调字节;没有重跑来源验收。
|
||||
|
||||
## Mnemo 与 Workbench 接续
|
||||
|
||||
Mnemo 普通开发 W1d、W1e 已完成各自列明的零模型/隔离范围,按协调观察登记,不独立重验。W1e 为同一认证 Core 的原文与原生元数据只读预览,协议 `mnemo.core-evidence-preview.v1`,默认关闭,当前 4320/4321 后端未切换;不推导通用命题、双时间、自动提取或真实模型质量。下一 W1f 需固定活来源、同模型/路由/tokenizer、两臂总预算及未消费留出,缺项不发请求;这里仅登记准备条件,不启动 W1f。
|
||||
|
||||
Workbench [B4b 七契约及 25 个合成案例](../QA-B4B-MNEMO-NATIVE.md)、[M3 隔离迁移和兼容恢复](../QA-M3-MIGRATION-RECOVERY.md)的工程结论及未尽事项不变。M3 最终产品 commit/版本、源码程序包、Inno Setup、独立安装和正式升级仍待;同源恢复副本不代签不同旧业务恢复。上次 916 个冻结构建输入核对结果仅沿用其原时点,本次没有重新运行构建或产品测试。
|
||||
|
||||
三产品身份/数据库/权限/偏好独立;unrestricted/passthrough/preserve 和持久 ask/auto/full 不变。历史 W/AD/AE/AG 成功/网络/unknown、R/W 真机边界、普遍重复读取/缓存与 Purge 收益 unknown 完整保留。
|
||||
|
||||
本轮只更正旧说明并新增本登记及独立 `.qa` 证据;零模型/工程/删除请求,未重放 unknown,未查正式 health、改产品/配置或其他项目,未提交/推送/安装/重启/发行、委派或跨线程回信。
|
||||
@@ -0,0 +1,71 @@
|
||||
# 三项目协调:维护互斥、第十一长窗与摘要诊断依赖登记
|
||||
|
||||
2026-11-10(Asia/Shanghai)。承接协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的新增公开依赖,仅完成只读复核和本仓库登记。[B4b 默认关闭的记忆桥](../QA-B4B-MNEMO-NATIVE.md)及 [M3 统一发行顺序](../M3-RELEASE-PLAN.md)保持,没有据协调来信启动新实现、模型、工程任务或服务操作。
|
||||
|
||||
本次将维护互斥更新为**来源工程已部署并通过验收**;v11 和两组摘要候选仍为**质量失败**。旧说明保留原时点,不把工程通过、任务成功或离线解码通过合并成模型能力通过。
|
||||
|
||||
## 固定来源与独立复核范围
|
||||
|
||||
| 公开来源 | 字节 | SHA256 |
|
||||
| --- | ---: | --- |
|
||||
| [维护互斥回顾](D:/Ai-project/models/docs/platform/reviews/RUNTIME-COORDINATION-acceptance.md) | 11659 | `96d004ff46558079da697d421c260f3c1df0aa0a4d62c67366b3f48c426c6b9c` |
|
||||
| [第十一长窗回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-COORDINATED-LONG-acceptance.md) | 11322 | `efc3603e96716665baed0ef7e84e782f7eaafb154071880126451ba8f05b9775` |
|
||||
| [两组摘要与 codec 诊断](D:/Ai-project/models/docs/platform/reviews/HARNESS-PINNED-RECORD-diagnostic.md) | 10184 | `e0fb13d48f49087d24b1e64f3f02527efbace5b3db910c84b0d66df318632be2` |
|
||||
|
||||
[动态协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-10T0653.json)本次按单次原字节捕获:87079 字节,SHA256 `51c3d430c33edb731b6a589365a30d5ebc1842bfdcefff8971f87b4dce11156b`。固定快照的 `completedAtUtc=2026-11-10T07:29:18.694Z`,投递状态 `tool_accepted_recipient_acknowledgment_pending`、`toolAccepted=true`、`recipientAcknowledged=false`、`finalCurrentRecordBytesAcknowledged=false`。这是捕获时的来源记录,不声称后续最终字节或接收确认已核验,也不回写协调文件。
|
||||
|
||||
Workbench 本次核对三份回顾和 **23 份公开 JSON 元数据**的长度/SHA;34 条来源绑定对应19个不同路径,全部与报告 `sources` 声明及实际文件摘要相符。Windows 路径分隔符按绝对路径规范比较,没有修改原 SHA。另捕获两份诊断报告,共29份来源快照。只读取绑定文件用于摘要或公开元数据核对,没有执行来源脚本,也不是完整源码审计。
|
||||
|
||||
独立证据见 [registration.json](../../.qa/coordination-readonly-2026-11-10T1532/registration.json) 和 [MANIFEST.json](../../.qa/coordination-readonly-2026-11-10T1532/MANIFEST.json)。31个清单成员已逐一复算;清单 SHA256 `c7a975a8bc8b7a7d2d72e538a509b1abee756bf0b85611f0820c2d562d16ed8c`。下文的运行、工具、质量评分及测试通过均为**来源报告**;本次独立工作限于原字节、公开元数据、映射及顺序核对,没有重新运行来源验收。
|
||||
|
||||
## MC:来源维护互斥已通过,整体状态仍待复核
|
||||
|
||||
来源 `sourceStatus=review_pending` 与 `engineeringVerdict=passed` 并存;AC-MC-01/02/03 均 passed,不能代改来源整体状态。这个新时点替代[上一登记](MNEMO-RECEIPT-PROBE-COORDINATION-READONLY-2026-11-10.md)中“仍在开发”的当前性,旧登记原文不改。
|
||||
|
||||
来源已部署稳定工作区落盘屏障,由 WSL 单权威管理,**无 TTL、无按时间自动释放**。覆盖完整长窗的准备、请求间隙、退出、撤销与预算处理;Scheduler 从 claim 到 finish,NodeAgent 从 pull 到原 receipt 均在合作入口内保护,空闲等待在锁外。
|
||||
|
||||
嵌套持有人、崩溃或退出 unknown 持续阻断;须核原 SHA、全部持有人停止及资源状态后显式恢复,不能删锁或按过期时间解锁。现场五个写入口返回503/not_started,匿名401,读取保持。这不控制任意外部 GPU 程序或源码编辑,也不唯一解释旧 v10 的准入失败。
|
||||
|
||||
原协议与 `/v1`、schema、身份、receipt/事件游标和推理 POST 语义保持。元数据探测补读不等于推理重试。97项 WSL(24+73)、Windows 11项探测/2项桥接/5项 PowerShell、6项归档夹具均仅按来源报告登记;本次未重跑。MC 阶段来源报告零模型/DELETE。此前 CP 备份约11分钟停机的问题没有因此得到优化或 SLA 认证。
|
||||
|
||||
## v11:63次成功交付,最终事实质量失败
|
||||
|
||||
原实验保持24×6000字资料、14字段的类型和值、58工具、32K上下文、2048输出和1024摘要硬限。这是来源冻结实验条件,不是 Workbench 新增的默认限额。
|
||||
|
||||
63个原请求全部 attempt1/succeeded。24段资料准确、23份持久摘要和四次摘要后工具检查按来源通过;本次核对63条原 key/云 task/node job/本机 task,以及原任务、wire摘要、持久历史、清理和独立复核的映射相符。
|
||||
|
||||
但 **CL-02=failed_quality**,CL-01/03=passed;`engineeringVerdict=passed` 与 `qualityVerdict=failed_quality`、`sourceStatus=review_pending`并存。第三份摘要的原输入尚有完整14字段,输出已由最近检查点文件的目录包装替代;最终仅有 `<path>/<type>` 两个包装字段,原14字段全缺。**最终事实检查实际执行且失败**,不是未执行,也不是本窗传输失败。
|
||||
|
||||
来源原任务192项、wire634项、持久历史187项、清理130项和独立复核130项记录保持。清理前证明所引三份审计 SHA/检查数与本次捕获一致,其结束时间及证明时间均早于清理开始;最终独立复核在清理结束后。这里只复核公开证明,没有重读原始 wire 或重新评分。
|
||||
|
||||
## 两组摘要候选与 codec:未采用,Session 权威未验收
|
||||
|
||||
两组各8个非流式请求,共16个,全部一次成功交付;每组另2个依赖请求因前置摘要未通过而未派发。两候选整体质量均失败、均未采用,正式 Harness 和默认模型保持。
|
||||
|
||||
原评分器对数组、多 JSON、重复对象及嵌套类型的漏判已由来源修复;原报告与离线重评分开保留。PR-01A/C/D 为 passed,**PR-01B=failed_quality,PR-02/03=not_started**。来源的19项解码/评分及2项状态保护测试仅证明相应工程范围。
|
||||
|
||||
离线 codec 能恢复原14字段,但它只是带捕获 SHA、消息索引、UTF-8 起止范围及范围 SHA 的数据提案:`pinAuthorityGranted=false`、`sessionEventProvenanceVerified=false`、`candidateAdopted=false`。本次没有打开其引用的私有 request/response capture、消息正文范围或原 SSE,不能据公开 codec 回执补签用户 pin 权限、Session 事件来源或真实质量。
|
||||
|
||||
来源正在推进 Session 生命周期原型;权限、版本、更新、撤销、重放及1024总容量**尚未验收**。协调中的开发进度和结构化 `not_started` 分开记录,不把原型在途写成已实现或另起完整长窗。
|
||||
|
||||
两组清理前审计138项;16条 task/key/job/local 映射与两份原诊断、云/本机公开元数据及清理记录相符。最终独立 verify56项在清理结束后,逐任务零 DELETE;早期与清理重叠的 verify 不作为最终顺序证据。
|
||||
|
||||
## 清理、费用、历史异常及封存保持
|
||||
|
||||
本次63+16为 **79个唯一新增任务**,四类身份映射均无重复。来源每个任务各 DELETE 一次,最终独立复核全部零 DELETE;本次没有发任何删除请求。MC累计272任务/23撤销身份,v11后335/24,两诊断后351/26,与本次79分开。受管清理不证明供应商或外部副本普遍删除,也不证明 Purge 收益。
|
||||
|
||||
月上限50元、历史2957 micro CNY及4个 unknown 保持。旧 v9末项 `actual_micro=null` 和较早 wire 的 `local_task_id=null` 原字节仍在,不以较晚完整映射补写旧记录;新增付费0不等于 unknown 已结算0。
|
||||
|
||||
v9 CT-03 failed_transport、v10 RR-02 failed_admission及原409正文永久缺证、BC-03/AB-04/NL-03/NL-04与 v6完整 wire 缺证均保留。P8仍六 partial/八阻断,不能用这批工程收尾改成整体通过。
|
||||
|
||||
三份来源 checkpoint manifest 的长度/SHA 均匹配协调所引值;没有逐一重哈希其中归档成员,也没有读取备份数据或恢复数据。来源正常 Desktop 九检查/八配置、资源 idle 与维护屏障释放只在其完成时点有效,本次没有查询正式4318 health或重新认证来源运行状态。
|
||||
|
||||
上一登记的18份捕获及20个清单成员已逐SHA复核,其中20条绑定记录原样保持,9个保护摘要一致;本次还将其说明、清单和独立完成回执加入保护链,共12个保护摘要。旧 B4b 清单 `3dfdc2b6…`、M3清单 `78ce1df7…`不变。工作区原62个待提交文件(含M3+B4b及既有登记)的状态和摘要另行固定,未暂存、提交或改写这些已有改动。
|
||||
|
||||
## Workbench 接续
|
||||
|
||||
Mnemo C5a合成 claim-resolution 已完成,C5b task-sourced procedure 仍在来源普通开发中;只记录协调快照,没有追读动态 status或执行其工作。W1k 的 engineering verified 仍与 quality blocked 并存,未产生派发许可、默认切换、提供方模板、独立留出或真实质量验收。三产品身份/数据库/权限/偏好/费用独立,Mnemo Core SQLite、J2封存和147冻结保持。
|
||||
|
||||
下一产品步骤仍是 M3 最终产品 commit/版本、源码包/程序包/Inno Setup及独立安装验收,再另冻正式升级和新真实样本;这些没有因来源工程通过而完成。本次不启动新阶段。B4b默认关闭和七契约/25合成案例不变;开放透传及持久 ask/auto/full 不变。历史W/AD/AE/AG成功/网络/unknown、R/W真机边界、普遍重复读取/缓存收益、Purge收益、长期及物理iPhone未验均保留。
|
||||
|
||||
本轮仅新增本说明和独立`.qa`证据:零模型/工程/DELETE请求,未执行来源脚本或产品测试、未读真实Key/来源数据库/私有会话/原SSE/备份数据、未重放unknown,未改产品/配置、部署/恢复/安装/重启/发行、提交/推送、委派或跨线程回信。接收登记不构成新增执行授权。
|
||||
@@ -0,0 +1,58 @@
|
||||
# 三项目协调:当前任务顺序与节点回执只读登记
|
||||
|
||||
2026-11-10(Asia/Shanghai)。按协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的合并来信,在既有安全边界登记公开依赖证据。没有启动新开发、模型或工程委托,不向来源线程专门回信;[B4b 工程范围](../QA-B4B-MNEMO-NATIVE.md)和 [M3 发行顺序](../M3-RELEASE-PLAN.md)不变。
|
||||
|
||||
## 来源与本次实际核对
|
||||
|
||||
- [智枢当前任务接纳与第九长窗回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-CURRENT-TASK-acceptance.md):30654 字节,SHA256 `d329a78c123415d71a08dad73af679d3bcbb7f395a0a19a9b24c8aa7cf61477e`,与来信一致。来源完成回合 `01a1c2af-b894-7051-a0d9-e4c9d911aab2`、最终消息 `msg_012c044ecbd7eafe016af2728ae40087d099c36d1d9e1a7293` 仅按来信及协调记录标识,没有另读线程或催问。
|
||||
- [本次动态协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-10T0248.json):以一次原字节快照绑定。快照 `completedAtUtc=2026-11-10T02:53:13.513Z`,投递状态为 `tool_accepted_recipient_acknowledgment_pending`;工具接受、接收确认和后续最终协调字节核验分开,不互相推断。
|
||||
- [Mnemo W1i 回顾](D:/Ai-project/Mnemo/docs/integration/2026-11-10-shared-context-review.md):8437 字节,SHA256 `e0dd2f339a71ffd5f86c1f8c0788bd89093d938e5c011d6b37a16433e929f148`,与协调引用一致;不借此核验动态 status 或其他在途开发。
|
||||
|
||||
本次核对协调引用的 10 份公开元数据,并补充 3 份部分清理、原任务、持久历史及通用 wire 报告,共捕获 16 份公开 JSON;23 条来源 SHA 绑定全部匹配。62 条原 key、云 task、node job、本机 task 与历史/清理映射相符。绑定的 Python 文件只计算 SHA,没有执行;没有读取来源数据库、私有会话或原始 SSE。
|
||||
|
||||
独立证据见 [registration.json](../../.qa/coordination-readonly-2026-11-10T1053/registration.json) 和 [MANIFEST.json](../../.qa/coordination-readonly-2026-11-10T1053/MANIFEST.json)。M3、B4b、前两次协调清单及已更正的说明均保持原 SHA。本次不重新构建或复跑产品测试;检查条目 passed 是公开回执事实,不代表本次重做运行验收。
|
||||
|
||||
## 已报告的修复及其边界
|
||||
|
||||
来源已安装 canonical Session 新输入稳定分组:Harness 补充信息先于本轮直接用户任务;两组各自内容、顺序和原 ID 保留,旧历史不重排。此次没有修改默认采样、Session v4、API/schema、工具权限或生产权威。
|
||||
|
||||
NodeAgent 的成功 receipt 先持久,再按原 `event_cursor/sequence` 补送事件;事件 ACK 丢失复用同序号,结果 ACK 丢失复用同 receipt。执行中的 `transport_or_response_unknown` 只继续查询同一本机任务,不重新 POST 推理。租约、删除屏障、明确取消/停止及原 deadline 仍阻止发布。82 项 WSL、26 项 Windows 相关回归和部署记录为来源报告,本次未重跑、部署或重启。
|
||||
|
||||
这与 v9 原失败分别记录:旧内层异常没有保留当时诊断,具体公网超时原因未被唯一证明;不能据修复说明反推唯一根因,也不能称已通过修复后的完整长窗。来源已另有原标准复验在途,本条没有最终结果,不查询、催问或重复启动。
|
||||
|
||||
## v9 仍是 CT-03 failed_transport
|
||||
|
||||
公开元数据保持 **62 原请求,前 61 次完整交付**;23 段、23 份持久摘要及四次摘要后工具通过按来源报告登记。最后一次本机 succeeded / attempt1,云端 failed / `node_result_unknown`,node job stopped;客户端末条没有成功 DONE,最终 14 项精确事实未执行。
|
||||
|
||||
原标准仍为 24×6000 字、14 精确事实、17 bundles/58 工具、32K/2048、0.875 压力阈值、4096 保留和 1024 摘要容量。这些属于来源冻结实验,不是 Workbench 新增额度。CT-01/02/04 为 passed,**CT-03 为 failed_transport,完整质量未通过**;P8 六项 partial、八个阻断及 ISS-029/ISS-038 保留。
|
||||
|
||||
| 报告 | 公开回执与本次核对结果 | 不可推导的结论 |
|
||||
| --- | --- | --- |
|
||||
| 交付专项 | 318 项标记 passed,`windowPassed=false`;61 完整交付加 1 unknown | 不等于长任务或最终事实通过 |
|
||||
| 原通用任务 | 187 项中 3 项失败,原件保持 | 不把 null 费用、两端终态差异或新增 unknown 改成成功 |
|
||||
| 原通用 wire | 245 个已记录检查、61 received,整体 false,末条缺成功字段后停止 | 不把已记录检查全为 true 当作完整审计通过 |
|
||||
| 持久历史 | 185 项标记 passed,23 摘要 | 不替代最终公网交付和事实质量 |
|
||||
|
||||
本次登记首轮误假定专项审计内较早 wire 快照的 `local_task_id` 在所有 62 条都完整,因此失败。实际只有末条为 null,最终原任务与清理回执才给出完整关联。已按不同时点保留字段,只通过原 key/云 task/node job 与最终映射连接,不给旧快照补值。首版脚本、失败日志和修正后回执均保留;这不改变来源失败判定。
|
||||
|
||||
旧 [BC-03 failed_quality、AB-04 failed_context、NL-03 failed_quality、NL-04 partial](MNEMO-BOUNDED-COMPACTION-COORDINATION-READONLY-2026-11-10.md)及 **v6 完整逐帧审计缺证**继续保留,不能由 v9 或新复验补签。
|
||||
|
||||
## 清理、费用和时点
|
||||
|
||||
四份来源清理回执的 DELETE 分别为 20、3、35、4,按任务逐条复算为 **62 次、62 个不同任务、重复 0**;独立 verify-only 128 项标记 passed,顶层及逐任务 DELETE 均为 0。本次没有发送删除。前三次部分失败、回执丢失/只读耗尽和维护收尾都保留,已尝试的删除不因 resume 重发。
|
||||
|
||||
受管 TLS 隧道只用于维护收尾,其回执明确 `ordinaryPublicAcceptance=false`,不能替代普通公网验收,也不能覆盖 DNS/TCP 异常。这里核对受管清理记录,不认证所有外部副本删除或 Purge 收益。
|
||||
|
||||
本阶段 30 诊断加 62 长窗共 **92 个唯一任务**,与累计保护的 **252 任务/22 身份**分开;22 身份 revoked 元数据均为 1。月上限 50 元、历史 2957 micro CNY 保持;unknown 从 3 增至 **4**,末项 `actual_micro=null`,不改成成功或已结算 0。来源新增付费 0 不抹除历史费用和未知计费。
|
||||
|
||||
来源 Desktop 九项检查、八配置哈希及 checkpoint 仅在对应已完成时点有效,不作为当前 idle 或新复验状态。checkpoint manifest SHA `8eaf12e95a72245f687943d1c6d68510fcbe9ca99b27db50fd2516e23d9285e4` 已核对;3332 文件/80320787 字节、27 项独立检查按公开报告登记,没有重算全部归档成员。
|
||||
|
||||
## Mnemo 与 Workbench 接续
|
||||
|
||||
Mnemo W1h/W1i 已记录各自零模型工程范围,合成 Task/Job 等夹具确有写入,不能称全过程无写。W1i 在相同完整 Runtime 中,10 条合法池原选 6 来源;4 条完整消息的 CURRENT 与扩展候选字节相同,均为 739 本地 tokens;seed 598 仍漏所需经过。没有候选材料收益或答案质量收益证据。
|
||||
|
||||
全池独立候选召回/联合排序、提供方模板一致性、中央完整 Runtime 预检、未消费独立留出和真实质量仍未验;本地 tokens 不代替供应商实际计费。Core SQLite 权威、原后端和默认关闭保持,协调不改状态或冻结。下一开发包按 Mnemo 自身范围推进,不转化为 Workbench 执行请求。
|
||||
|
||||
B4b 七契约/25 合成案例及 M3 待办不变;三产品身份、数据库、权限、偏好和费用边界独立。Workbench 默认 unrestricted/passthrough/preserve、持久 ask/auto/full 保持;历史 W/AD/AE/AG 成功/网络/unknown、R/W 真机边界、普遍重复读取/缓存与 Purge 收益 unknown 完整保留。
|
||||
|
||||
本次仅新增本说明及独立 `.qa` 登记证据:零模型/工程/DELETE 请求,未重放 unknown,未查正式 health、改产品/配置或其他项目;未提交/推送/安装/重启/发行、委派或发送跨线程消息。
|
||||
@@ -0,0 +1,241 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"recordedAtUtc": "2026-11-11T17:44:27.815Z",
|
||||
"recordedLocalDate": "2026-11-12",
|
||||
"timezone": "Asia/Shanghai",
|
||||
"scope": "Read-only public source metadata and named result receipts; current product phase and existing authorization remain independent.",
|
||||
"sourceThreadId": "01a1ae3d-66a7-78b1-945e-b0c1d42c28ae",
|
||||
"sources": [
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/docs/integration/2026-11-12-goal-native-worker-checks-review.md",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/review.md",
|
||||
"bytes": 6227,
|
||||
"sha256": "71f24c31ed4c11cd931d338fa1ccf3eaa1953fcf194b6006d0780a0dc1eb137d"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/docs/integration/goal-native-worker-checks-validation.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/source-validation.json",
|
||||
"bytes": 13905,
|
||||
"sha256": "2c5a6a9058dd3140d609dc200f83506b5a7ab08ada864d7c52e54b33418b696d"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-11T1729.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/coordination.json",
|
||||
"bytes": 32164,
|
||||
"sha256": "6768561acf99ded187af307957f07d1c7e731d1e412cddb075745f70f0c1443a"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/output/goal-native-worker-checks/2026-11-11-user-continue/independent-http-FVYkLu/result.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/actualEvidence.json",
|
||||
"bytes": 39370,
|
||||
"sha256": "41d994a81a2d0b8778053884c19d8d699d834ad41e16235e47630553dd5a3a8a"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/output/goal-native-worker-checks/2026-11-11-user-continue/tests-07-nAJKiX/result.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/relatedNewTestRun.json",
|
||||
"bytes": 2854,
|
||||
"sha256": "34b3f7f981ecca566f38fb097ce18090aac47e18bebdeb04b52d983d1dbcdf7a"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/output/goal-native-worker-checks/2026-11-11-user-continue/goal-regression-01-lmA40f/result.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/relatedRegressionRun.json",
|
||||
"bytes": 2951,
|
||||
"sha256": "f78243a86a12484996ad2df600b1ba0a43ce4a11a4f17d6cf42b0bdab4bbee3c"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/output/goal-native-worker-checks/2026-11-11-user-continue/types-07-iSdiNm/result.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/typecheckRun.json",
|
||||
"bytes": 2703,
|
||||
"sha256": "7c366f9ee110cb2d5faa8f2116d0cb5a4aa62bd5c93f1418acb6e43a2ff7de48"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/output/goal-native-worker-checks/2026-11-11-user-continue/integrity-final.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/finalIntegrity.json",
|
||||
"bytes": 58872,
|
||||
"sha256": "4ae23efff41de7a8030fcc747e8c0c43552bec0ed6b2a4ba7d2aafd991e5e244"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/output/goal-native-worker-checks/2026-11-11-user-continue/current-host-readonly-probe.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/source-host-probe.json",
|
||||
"bytes": 335,
|
||||
"sha256": "2a213a12bc51a2fcc72a6b45fbdddeaf7b7e8af8218357d286f313c58fa21a17"
|
||||
}
|
||||
],
|
||||
"claude": {
|
||||
"connectionId": "9ae348a5-d0a6-42ed-8664-0da5f9effe34",
|
||||
"connectionRevision": 9,
|
||||
"credentialRevision": 2,
|
||||
"catalogModels": 14,
|
||||
"selectedListed": true,
|
||||
"requests": 2,
|
||||
"unknownRequests": 2,
|
||||
"actualTotalTokens": null,
|
||||
"actualTotalCost": null,
|
||||
"upstreamHttpStatus": 403,
|
||||
"upstreamType": "permission_error",
|
||||
"message": "该分组仅允许使用 Claude Code 官方客户端。 This group is restricted to Claude Code clients (/v1/messages only)",
|
||||
"textTaskId": "22f44c69-5b5e-4d23-9298-9207e7550e34",
|
||||
"toolTaskId": "5decb3cb-ea08-4d14-aa28-2f697c37ebb0",
|
||||
"completeHttpErrorAndUnknownUsageSeparate": true,
|
||||
"oldMismatchResponses": 2,
|
||||
"oldCatalogAbsenceRetained": true,
|
||||
"modelCompatibilityQualified": false,
|
||||
"repeatRequested": false,
|
||||
"newScopeOnlyAfterActualClientPermissionChange": true
|
||||
},
|
||||
"mnemo": {
|
||||
"status": "listed_finite_existing_worker_goal_scope_passed",
|
||||
"nativeIdentity": {
|
||||
"projectId": "4070c3e0-a292-45ba-bdbc-bcb4160f0fd1",
|
||||
"sessionId": "91d910d3-7849-4f43-a0d5-1cfa019c34d7",
|
||||
"taskId": "42aacbd1-f1fd-418c-80a9-9e1af9ff42c8",
|
||||
"runId": "b330b966-848a-496b-a232-6b4438383ccc",
|
||||
"workspaceId": "11771ddf-9dc1-4ba4-987f-34cc3d80c2ce",
|
||||
"workerId": "2539daaf-ae13-4506-ac95-227032f606b0",
|
||||
"baseCommit": "c213ad68236af508bf9b31e9739483caeda4ef0c"
|
||||
},
|
||||
"newChecks": 19,
|
||||
"existingRelatedChecks": 94,
|
||||
"relatedChecks": 113,
|
||||
"independentChecks": 69,
|
||||
"independentHttpRequests": 31,
|
||||
"originalCodeReexecuted": false,
|
||||
"originalTestsPreviouslyPassed": 9,
|
||||
"originalIndependentChecksPreviouslyPassed": 57,
|
||||
"originalModelRequests": 8,
|
||||
"originalReportedTotalTokens": 109888,
|
||||
"originalPriceVerified": false,
|
||||
"modelCallsThisSourceIncrement": 0,
|
||||
"readOnlyInspectCalls": 3,
|
||||
"closedLoopbackHosts": 2,
|
||||
"closedStoreCopies": 11,
|
||||
"frozenInputsUnchanged": 147,
|
||||
"initialFailedValidationAttemptsRetained": 11,
|
||||
"sameGoalFiniteScopeCompleted": true,
|
||||
"explicitCompletionCheckpointVerified": true,
|
||||
"completeW2": false,
|
||||
"producerAuthenticated": false,
|
||||
"nativeFullSourceFingerprintVerified": false,
|
||||
"parentDeliveryVerified": false,
|
||||
"semanticQualityVerified": false,
|
||||
"nativeDispatchReceiveQualified": false,
|
||||
"normalHostRouteAvailableAtSourceObservation": false,
|
||||
"normalHostReloaded": false,
|
||||
"sourceHostProbe": {
|
||||
"path": "D:/Ai-project/Mnemo/output/goal-native-worker-checks/2026-11-11-user-continue/current-host-readonly-probe.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-12-goal-worker/source-host-probe.json",
|
||||
"bytes": 335,
|
||||
"sha256": "2a213a12bc51a2fcc72a6b45fbdddeaf7b7e8af8218357d286f313c58fa21a17"
|
||||
},
|
||||
"rawProbeFieldNames": [
|
||||
"status",
|
||||
"url",
|
||||
"method",
|
||||
"at",
|
||||
"modelCalls",
|
||||
"body",
|
||||
"mutations",
|
||||
"authenticated"
|
||||
],
|
||||
"historicalTaskRedispatched": false,
|
||||
"sourceNext": {
|
||||
"id": "U2-W2-next-dependency-supported-actual-task-review",
|
||||
"status": "no_new_dependency_ready_task_selected_by_this_increment",
|
||||
"implementationStarted": false,
|
||||
"scope": "Only review newly available tasks/adapters/resources in the existing W2 plan; do not repeat this finite existing-worker task.",
|
||||
"completionCondition": "核对新增、明确登记且未消费的实际任务,固定其事前判据与原生适配/manifest/资源/预算/原键;只实施已授权且依赖具备的部分。常驻服务载入需服务变更新授权,真实模型/embedding/付费/权重仍关闭。没有新具备依赖的待办就保持现状,不重复原九项/build、本地摘录或C5o~C5y。"
|
||||
}
|
||||
},
|
||||
"counting": "113=19+94; independent69 includes selected-pin checks and is not added to31 HTTP or3 source inspections. Original9/build/57 and8 model requests are historical; this source increment made0 model calls.",
|
||||
"taskSeparation": {
|
||||
"adTaskId": "6be88bee-1995-424a-b15a-2e0e54207d1a",
|
||||
"agTaskId": "42aacbd1-f1fd-418c-80a9-9e1af9ff42c8",
|
||||
"m4TaskIds": [
|
||||
{
|
||||
"kind": "node",
|
||||
"taskId": "8249fea9-3613-4830-b016-53446019c476"
|
||||
},
|
||||
{
|
||||
"kind": "react",
|
||||
"taskId": "dddcbdf2-588d-4fee-a26c-236e110a925d"
|
||||
},
|
||||
{
|
||||
"kind": "python",
|
||||
"taskId": "875afb7b-ea0b-41bf-8434-5897f5c800f0"
|
||||
}
|
||||
],
|
||||
"allDistinct": true,
|
||||
"oldAdInternalReportInProgressRetained": true,
|
||||
"oldAgNetworkUnknownAndPurgeBenefitRetained": true
|
||||
},
|
||||
"sealedWorkbench": {
|
||||
"originalM4Requests": 40,
|
||||
"originalM4UnknownRequests": 0,
|
||||
"originalM4Files": 541,
|
||||
"originalM4ManifestSha256": "558c9b548559798279d7f8e61deb3b7344920753a53dbf4e84eec374c6f3a3df",
|
||||
"correctionFiles": 41,
|
||||
"correctionManifestSha256": "9b8a1cfcc4a8842e5c751990fdfc4a274831bf4dcafe73c557514c992490f2e5",
|
||||
"rootsWritten": false,
|
||||
"rawMembersReopened": false,
|
||||
"verificationScope": "Existing public indices and manifest metadata only in this coordination; previous member verification remains separately recorded."
|
||||
},
|
||||
"ownReferences": [
|
||||
{
|
||||
"path": "docs/QA-CLAUDE-PROXY-CORRECTION-2026-11-12.md",
|
||||
"bytes": 4881,
|
||||
"sha256": "32087435370b79aa0939ffec16d7bd99edf41e3a9d55fc8b362c7645617eee8f"
|
||||
},
|
||||
{
|
||||
"path": "docs/claude-proxy-correction-2026-11-12.json",
|
||||
"bytes": 14308,
|
||||
"sha256": "7c4cbc3f2897584f5601c846218e3713fb8f89a76b24f9a10412a9e192dfafab"
|
||||
},
|
||||
{
|
||||
"path": "docs/claude-proxy-correction-seal-2026-11-12.json",
|
||||
"bytes": 958,
|
||||
"sha256": "9747526c8eab2c393069143656213a81b96091cdd80e7d8e69a90c99a0e298fc"
|
||||
},
|
||||
{
|
||||
"path": "docs/QA-M4-LIVE-BASELINE.md",
|
||||
"bytes": 6286,
|
||||
"sha256": "ca2b27789a52060fa6bc2f273af48108f7eb610485e891bd727e48181bd4c75d"
|
||||
},
|
||||
{
|
||||
"path": "docs/m4-final-2026-11-11.json",
|
||||
"bytes": 167824,
|
||||
"sha256": "e0ea050f24774d61c190d400a1e7bc6dcdb233e4c2d3076e10928dd4dd7046c3"
|
||||
},
|
||||
{
|
||||
"path": "docs/m4-live-seal-2026-11-11.json",
|
||||
"bytes": 622,
|
||||
"sha256": "c5ac5623023f0a7f89a2c8f95dbb1e19b76103e39b8f734f21c990f878e45cea"
|
||||
},
|
||||
{
|
||||
"path": "docs/integration/WORKBENCH-AD-NODE-READONLY-SOURCE-2026-11-11.json",
|
||||
"bytes": 25974,
|
||||
"sha256": "a0a1fdb246a07b3a64c160c307545bd1e00c2f182adcebaa50ab9f6f0f30c9f6"
|
||||
}
|
||||
],
|
||||
"verification": {
|
||||
"ninePublicCopiesExact": true,
|
||||
"fivePinnedSourceReceiptHashesMatched": true,
|
||||
"related19And94ReceiptCountersMatched": true,
|
||||
"independent69ReceiptPassedMatched": true,
|
||||
"typecheckExitCodeMatched": true,
|
||||
"selectedAGIdentityMatchesPriorSource": true,
|
||||
"distinctFromADAndM4": true
|
||||
},
|
||||
"boundaries": {
|
||||
"modelRequests": 0,
|
||||
"engineeringExecutions": 0,
|
||||
"externalHttpRequests": 0,
|
||||
"credentialsReadOrResaved": false,
|
||||
"otherProjectSourceWritten": false,
|
||||
"formalServiceReadOrRestarted": false,
|
||||
"oldSamplesReplayed": false,
|
||||
"messagesSent": 0,
|
||||
"subagentsOrAutomationsCreated": false,
|
||||
"authorizationExpanded": false,
|
||||
"sourceAuthenticationDatabasesPermissionsAndModelPreferencesInherited": false
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,15 @@
|
||||
# Mnemo 同 Goal 旧 Worker 工件连接:新增只读结论
|
||||
|
||||
2026-11-12(上海时间)接收三项目协调来源,按当前记录范围只读核对;机器索引:[MNEMO-GOAL-WORKER-READONLY-2026-11-12.json](MNEMO-GOAL-WORKER-READONLY-2026-11-12.json)。九份公开来源/结果回执原字节副本在 `.qa/coordination-readonly-2026-11-12-goal-worker/`,五份发布的回执bytes/SHA、19+94检查计数、69独立结果及类型退出码已核对。未执行来源检查、HTTP、模型请求或工程,也未查询/重启正式服务、读Key或回复来源线程。
|
||||
|
||||
Claude修后现有两用例已经结束:revision9/credentialRevision2,完整14型号目录包含claude-opus-5-5;文本Task22f44c69、工具Task5decb3cb各一请求,均上游403 permission_error,仅允许Claude Code官方客户端。完整188字节HTTP错误与usage未知分别成立,实际tokens/费用null;旧两次自报GPT、旧目录缺项和原M4四十请求/541成员封存均保持。当前只核已有公开索引及清单元数据,没有重开原封存成员、再次验收或重POST。本协调不请求重试/Purge;实际标准Messages客户端权限变化后也需按当前用户授权另冻新用例。
|
||||
|
||||
Mnemo已完成人类指定旧AG ba-off的有限同Goal工件连接/显式complete检查点:Task `42aacbd1-f1fd-418c-80a9-9e1af9ff42c8`、Run `b330b966-848a-496b-a232-6b4438383ccc`、子工作区及HEAD c213ad68与原登记逐字段一致。相关113=新19+原94;独立69包含选定pin核对,31次隔离HTTP、3次只读inspect和两个关闭宿主单列,不能累加成更多独立测试。11次原失败保留。原九项/build/57结果未重跑,旧8请求/109888报告tokens及未核实价格保留,本增量模型调用0。
|
||||
|
||||
这证明既有原件连接和隔离本地服务守卫的有限范围。常驻4321的新入口只有来源当时未认证GET的404观察,未载入/重启本包;本轮没有重新探测。生产者认证、完整原生来源指纹、原生派发接收、父交付/完整交付报告、语义质量和完整W2仍未验证,文件复核也不提供跨进程原子快照。Core SQLite权威、147冻结与原质量失败保持。
|
||||
|
||||
原AD的Task6be88bee、这个旧AG与M4三个新工程均是不同对象,不能把一方结果补签到另一方。原AD内置报告in_progress、原AG三次网络失败/unknown及唯一ba-off工程成功、Purge收益未验证等原结论不改签;当前M4三工程成功也不替代Mnemo生产或记忆语义验收。
|
||||
|
||||
后续M7原生集成引用这份限定证据;M8新原版Purge仍按独立冻结计划。三产品认证、数据库、权限、模型偏好及质量门槛不互相继承。本消息没有新增执行授权,本次仅登记来源并沿用已有本地小提交纪律,待后续开发阶段统一增量推送。
|
||||
|
||||
来源:[Mnemo回顾](D:/Ai-project/Mnemo/docs/integration/2026-11-12-goal-native-worker-checks-review.md)、[机器索引](D:/Ai-project/Mnemo/docs/integration/goal-native-worker-checks-validation.json)、[本轮协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-11T1729.json)。此前[NRO只读登记](MNEMO-NRO-READONLY-2026-11-11.md)与Workbench原验收不修改。
|
||||
@@ -0,0 +1,59 @@
|
||||
# 三项目协调:长会话对照与摘要截断保护只读登记
|
||||
|
||||
2026-11-10(Asia/Shanghai)。承接[上次长会话登记](MNEMO-LONG-SESSION-COORDINATION-READONLY-2026-11-08.md),接收协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的公开依赖更新。这里只核对报告、公开元数据、文件摘要及 Workbench 冻结输入,没有重跑来源测试、模型或工程任务。主路线仍为 [M3 统一发行](../M3-RELEASE-PLAN.md),本次登记不启动发行或改变产品默认行为。
|
||||
|
||||
## 来源及实际核对范围
|
||||
|
||||
- [智枢源绑定对照与截断保护回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-NATIVE-LONG-AB-acceptance.md):20776 字节,SHA256 `f4c46e40a291c9cefccaa39105513af48875ec308599d70af1fdcc6a6f576d3c`。
|
||||
- [Mnemo 协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-09T1650.json):文件可继续更新,本次以读取的原字节快照绑定。实际封存为 16906 字节,SHA256 `be567d2a7da185cf0933aa1dea620dc9b64c65906d88f3c1057e8a0f61a4b92d`;`completedAtUtc` 为 `2026-11-09T16:58:39.132Z`,投递状态为 `tool_accepted_recipient_acknowledgment_pending`,`toolAccepted=true`、`recipientAcknowledged=false`。这只证明该快照的工具接受状态,不代表接收确认或核验后续最终协调字节。
|
||||
- 8 份公开 JSON 元数据报告的字节数、SHA256 全部与该协调快照一致;报告内 12 条来源绑定也全部匹配,涉及 10 个不同路径,不能说成 12 个独立文件。绑定的 Python 文件仅计算摘要,没有执行。
|
||||
- Workbench 当前 916 个构建输入与 B4b 最终冻结逐项字节数/SHA256 一致;M3、B4b 两个旧封存清单的 SHA256 及上次验收快照均保持。
|
||||
|
||||
2026-11-10 文字更正:本项此前误用了捕获前读到的 null / `dispatch_attempt_in_progress`,与实际封存快照不符,现按原件更正。原快照、registration、清单及原收尾回执均不改;更正前文档另有原字节快照,新核对见[本次有界摘要依赖登记](MNEMO-BOUNDED-COMPACTION-COORDINATION-READONLY-2026-11-10.md)。
|
||||
|
||||
本次证据在 [registration.json](../../.qa/coordination-readonly-2026-11-10T0057/registration.json),包括协调及验收报告原字节快照、8 份公开元数据快照、来源绑定和复算结果;[MANIFEST.json](../../.qa/coordination-readonly-2026-11-10T0057/MANIFEST.json) 为独立清单。没有向旧封存根追加文件。
|
||||
|
||||
本次核对了原任务/请求键、云端/本机映射、状态和公开报告中的检查条目,未读取来源数据库、私有会话或原始 SSE。报告标记的检查通过只作为元数据事实,不能称为 Workbench 独立重跑通过;checkpoint 的 2842 文件/63123336 字节和旧 2024 份证据保持也仅核对公开报告,未重算所有外部成员。
|
||||
|
||||
## 长任务仍未通过
|
||||
|
||||
来源 P8 仍为 `in_progress`:AB-01/02/03/05 为 passed,**AB-04 为 `failed_context`**,ISS-029/ISS-038 保留阻断。旧 NL-03 `failed_quality`、NL-04 `partial` 及历史窗口结论不改写。
|
||||
|
||||
12 个源绑定诊断对照中,来源报告的新提示四摘要链和最终事实保持原 14 字段;旧串线未稳定复现,最小上下文对照仍多出行号。人工拆分仅用于诊断,不等于生产修复。诊断为 `stream=false`,除首臂外 `temperature=0`,未后处理模型原文;这些是来源实验条件,不是 Workbench 新增限制。
|
||||
|
||||
第六个真实长窗口的公开元数据复算为 **28 原任务、27 succeeded / 1 failed,全部 attempt1**;失败码为 `context_budget_exceeded`。4 份持久摘要、7 个工具映射一致;前五次 review/read 回填正确是来源报告结论。
|
||||
|
||||
第五份摘要达到固定 **1024 token / finish length**,残缺摘要没有提交;旧客户端继续使用旧历史派发 REVIEW_06 后超限。原标准为 24×6000 字、17 bundles、58 工具、32K 上下文、2048 输出、4096 保留及 1024 摘要容量;没有完成全部资料和最终事实,不能用 27 次成功替代完整任务通过。这些数值属于来源冻结实验,Workbench 的默认透传策略不变。
|
||||
|
||||
**v6 完整 cloud/node/local ordered SSE 审计在正文清理前没有完成,明确缺证。** 已有任务映射和持久历史报告不替代该项;不恢复已删正文补证,不拿上轮 253 项检查补签本轮逐帧一致性,也不把缺证改写成 wire 失败或通过。
|
||||
|
||||
## 工程保护修复与容量复验分开
|
||||
|
||||
来源报告已修改 `pre-step`:遇到 `MAX_TOKENS` 重新抛出并停止当前轮,不接纳 pending 输入、不提交残缺摘要,不继续派发 conversation/step/replacement,已有持久历史保留。真实 Loader/AgentLoop 回归和移除保护的负控制通过,之后精确恢复源码 SHA;负控制公开回执为预期失败退出码 1。本次仅核对其报告摘要,没有执行负控制、恢复源码或构建。
|
||||
|
||||
来源列明 159 包测试、14 Python 检查、9 构建门、207 集成通过加 1 skip、11 许可和 43 文档检查;正常 Desktop 恢复及 8 配置文件保持也是来源结论。**真实第五摘要容量没有重新验收**,保护截断后停止派发不代表容量问题或完整长任务质量已解决。Session 格式、schema、API 和路由未改、unknown 原键只查不重发按来源报告保留。
|
||||
|
||||
## 数量、费用及保护边界
|
||||
|
||||
| 范围 | 本次核对的公开元数据 |
|
||||
| --- | --- |
|
||||
| v6 长窗口 | 28 唯一映射;历史报告 79 项、原任务报告 85 项检查条目均标记 passed |
|
||||
| 诊断对照 | 原首臂 1 条加 v2 的 11 条,与收尾 12 唯一映射相符;独立 verify-only 44 项 |
|
||||
| 本次新增映射合计 | 28 + 12 = 40,任务 ID 去重后仍为 40 |
|
||||
| v6 清理核对 | verify-only 59 项;该核对报告声明零模型提交、零 DELETE |
|
||||
| 累计保护 | 另一个范围:126 任务、13 身份、12 原表;18 项检查条目标记 passed,13 个身份的 revoked 字段均为 1 |
|
||||
| checkpoint | 24 项检查条目标记 passed;manifest SHA256 `4f5650bed34fe14b0ae5cadea7040d3616aa1d4d97172c6767dcd2d5221b8fc0` |
|
||||
|
||||
126 不是本次新运行数量,13 也不是本次新建身份数量。清理/撤销事实只核对来源公开元数据,不独立认证外部副本删除。
|
||||
|
||||
香港 schema6 单一权威、本机 schema7 独立、月 50 元、历史 2957 micro CNY 及 3 个 unknown 均按来源保持。新增厂商付费 0 不表示没有历史费用或调用。本次 Workbench 登记新增模型/付费/工程请求均为 0,未重放 unknown。长期质量、30 秒聊天 SLA、完整境内迁主仍未认证。
|
||||
|
||||
## 对 Workbench 后续的影响
|
||||
|
||||
Mnemo 协调观察仍为 U2-B / Core SQLite:普通开发已列明零模型隔离 W0/W1a/W1b/W1c,当前 W1d `candidate-answer-delivery` 在途。人工语义接纳不等于自动提取或真实模型质量,协调未认证正式默认能力。本次没有要求 Workbench 执行新接口,也没有把观察状态补签为阶段完成。
|
||||
|
||||
Workbench [B4b 七契约及 25 个合成工程案例](../QA-B4B-MNEMO-NATIVE.md)的范围不变;真实模型、长期、物理 iPhone、普遍重复读取/缓存收益及 Purge 收益仍未验。后端 30 Run / 37 尝试 / 31 HTTP、含两端 UI 的 34 Run / 41 尝试 / 35 HTTP 继续引用上次登记,不将本次来源长任务混入这些计数。
|
||||
|
||||
[M3 隔离迁移及兼容恢复](../QA-M3-MIGRATION-RECOVERY.md)已有工程证据,最终统一产品 commit/版本、源码程序包、Inno Setup、独立安装和正式升级仍待;同源恢复副本尚未证明回到不同旧业务实现。先完成 M3,再另冻真实兼容与自主工程样本,历史 W/AD/AE/AG 成功/网络/unknown、R/W 真机剔除边界和 Purge 收益未知全部保留。
|
||||
|
||||
本次仅新增本说明及独立 `.qa` 登记证据:未改产品/配置或其他项目,未查正式 4318 health,未读真实 Key,未启动产品测试、模型或工程任务,未提交/推送/发布/安装/重启、委派或发送跨线程消息。默认 unrestricted/passthrough/preserve 和持久 ask/auto/full 不变。
|
||||
@@ -0,0 +1,40 @@
|
||||
# 三项目协调:长会话质量与 B4b 工程边界只读登记
|
||||
|
||||
2026-11-08(Asia/Shanghai)。接收 Mnemo 协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的新增公开报告,离线核对来源和本地计数。主路线保持 [M3 统一发行](../M3-RELEASE-PLAN.md)。本次只新增这份说明和独立登记证据,不修改产品、配置、正式服务或两个已封存证据根,不向来源线程回复或群发。
|
||||
|
||||
## 来源与核对范围
|
||||
|
||||
- [智枢长会话验收](D:/Ai-project/models/docs/platform/reviews/HARNESS-NATIVE-LONG-acceptance.md):SHA256 `ebfcb0e0dd6c2483fa3ce8a2a20adc229ce51500f834d00cc6aada7f05f48c64`。
|
||||
- [智枢收尾保护报告](D:/Ai-project/models/logs/native-long-protection-20261108-final.json):SHA256 `b8006180480d2519f33da50bbf28825ac876f6294c83c76036b73465c9fe0800`。
|
||||
- [Mnemo 本次协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-08T0227.json):该协调文件仍可更新,本地保留本次读取的原字节快照、摘要和当时状态,不覆盖原报告。
|
||||
|
||||
协调记录引用的11份公开来源文件 SHA 全部匹配。保护报告的25项检查均为 passed,8窗口/78原任务、10个临时身份均已撤销的公开元数据一致。Workbench 当前916个构建输入与 B4b 最终冻结一致。这里核对的是原报告和公开元数据,没有重新运行智枢长会话、读取源数据库、私有会话或原始 SSE,也没有独立重做外部删除。
|
||||
|
||||
登记和快照在 `.qa/coordination-readonly-2026-11-08T0227/registration.json`,独立清单为同目录 `MANIFEST.json`;M3/B4b 已封存根未追加。
|
||||
|
||||
## 智枢结果保持原判定
|
||||
|
||||
P8 仍进行中:NL-01/02/05 为 passed,NL-03 为 `failed_quality`,NL-04 为 `partial`。
|
||||
|
||||
第五窗口的25原请求、4份持久摘要和253项 SSE/任务一致性检查完成,但第三摘要后,真实 read 回填仍误答资料标记 `REVIEW_06`。两次同会话重开保留14个原字段值和类型,却均多5个字段,严格JSON回答质量继续失败。持久历史、摘要数量和HTTP成功分别保留其技术证据,严格长任务质量门槛保持未通过。
|
||||
|
||||
单次终帧断流只有1个 POST,客户端持久保留 unknown,未追加成功回答、派发工具或自动重发。服务端原键只读核对为成功,客户端没有据此补签。来源声明新增付费0;这与本次零新增模型请求登记分开记录,不把既有费用和 unknown 清零。
|
||||
|
||||
8窗口/78原任务的清理映射和守卫元数据经来源报告检查,10临时身份 revoked;不认证外部副本删除、长期质量、30秒聊天SLA或完整境内权威迁移,也不认定摘要投影或模型能力为唯一根因。
|
||||
|
||||
## Workbench 与后续边界
|
||||
|
||||
本地[最终 B4b 机器记录](../b4b-mnemo-native-2026-11-08.json)及网页/隐藏Electron逐请求回执复算一致:
|
||||
|
||||
| 范围 | Run | 账本尝试 | 实际HTTP |
|
||||
| --- | ---: | ---: | ---: |
|
||||
| 后端 N01–N24 | 30 | 37 | 31 |
|
||||
| 加入网页与隐藏Electron两端各4组 | 34 | 41 | 35 |
|
||||
|
||||
6个发送前拒绝属于账本尝试,不计作实际HTTP。25/25是单项目合成原生工程验收,真实模型质量、物理iPhone、工程委托、长期效果和正式发行分别保持待验。
|
||||
|
||||
M3隔离 schema1→2 迁移及新增记录后的兼容恢复完成;最终产品commit/版本、源码和程序包、安装器与正式升级未做,使用同源兼容恢复副本的证据也未证明回到不同旧业务实现。相关状态见 [M3报告](../QA-M3-MIGRATION-RECOVERY.md) 与 [B4b报告](../QA-B4B-MNEMO-NATIVE.md)。
|
||||
|
||||
Mnemo 下一 D0 仅为零模型质量准备,本次协调不启动该阶段。三个产品的身份认证、数据库、模型偏好和预算保持独立;Workbench unrestricted/passthrough/preserve 及持久 ask/auto/full 不变,历史 W/AD/AE/AG 成功/网络/unknown、重复读取普遍收益、R/W 真机边界和 Purge 收益未验保持。
|
||||
|
||||
本次新增模型/付费/工程请求0,未重放 unknown、读取真实Key、修改配置、提交、推送、发布或发送跨会话消息。
|
||||
@@ -0,0 +1,96 @@
|
||||
# 三项目协调:自然授权、摘要事务与接纳时序依赖登记
|
||||
|
||||
2026-11-10(Asia/Shanghai)。承接协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的新增公开格式/恢复参考,仅只读核对并登记。本轮不启动新实现、模型/工程实验、迁移、发布或跨线程转发,[B4b 默认关闭记忆桥](../QA-B4B-MNEMO-NATIVE.md)与 [M3 统一发行顺序](../M3-RELEASE-PLAN.md)保持。
|
||||
|
||||
三增量均为**隔离工程通过、整体 partial / review_pending,未发布到日常 Desktop**;真实模型整体收益和原标准完整窗仍待验。它们更新[上轮登记](MNEMO-RECORD-LIFECYCLE-READONLY-2026-11-10.md)中自然授权/摘要事务的在途范围,旧文档按其原时点保持,不补签旧质量或部署通过。
|
||||
|
||||
## 固定来源与复核范围
|
||||
|
||||
| 来源回顾 | 字节 | SHA256 |
|
||||
| --- | ---: | --- |
|
||||
| [RC 自然授权](D:/Ai-project/models/docs/platform/reviews/HARNESS-NATURAL-RECORDS-prototype.md) | 33076 | `89522ac5b5b45d9c30bb32f219d71c09ce326e19255029ccf8170bdd622676dd` |
|
||||
| [RT 摘要事务](D:/Ai-project/models/docs/platform/reviews/HARNESS-RECORD-TRANSACTIONS-prototype.md) | 31894 | `45bd214a20e939475cf1a1e519191379c06d777767dfc3b66fda179130397993` |
|
||||
| [RA 接纳修复](D:/Ai-project/models/docs/platform/reviews/HARNESS-RECORD-ADMISSION-prototype.md) | 28383 | `b0a6fefa441060283befc0a5dbb8d29575612abba486dc6d3bf93d125c5165b2` |
|
||||
|
||||
配套三份契约也按协调所引原字节捕获:
|
||||
|
||||
- [HARNESS_NATURAL_RECORDS.md](D:/Ai-project/models/docs/platform/integrations/HARNESS_NATURAL_RECORDS.md):3230字节,SHA256 `37a3dc9fe23ef08d6e088c59be2940bd9ac47b538d836ec4e43a25421f0d1196`。
|
||||
- [HARNESS_RECORD_TRANSACTIONS.md](D:/Ai-project/models/docs/platform/integrations/HARNESS_RECORD_TRANSACTIONS.md):3048字节,SHA256 `67cb2ae522726fde5dbde9dcd592c0dcdc6421ef4cf620e787de3ad1e5d177a8`。
|
||||
- [HARNESS_RECORD_ADMISSION.md](D:/Ai-project/models/docs/platform/integrations/HARNESS_RECORD_ADMISSION.md):1791字节,SHA256 `aabd1cd8cde94d5b0b235d66579cdbe11edd5b9b49183f7ae2dd9bbebcaae65e`。
|
||||
|
||||
[动态协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-10T1057.json)按单次原字节捕获:43317字节,SHA256 `92e98b374fde01782aac7e50d1568bca972a1a992982b160fc4e4851084dc510`。固定版本 `completedAtUtc=2026-11-10T11:06:18.280Z`、投递状态 `tool_accepted_recipient_acknowledgment_pending`、`toolAccepted=true`、`recipientAcknowledged=false`;未提供最终字节接收确认字段,不补推后续状态。
|
||||
|
||||
Workbench 本次复核**13份公开元数据、15条来源SHA引用(6个不同来源)及3份checkpoint manifest**,均匹配。共20份来源捕获,见 [registration.json](../../.qa/coordination-readonly-2026-11-10T1907/registration.json)与 [MANIFEST.json](../../.qa/coordination-readonly-2026-11-10T1907/MANIFEST.json);22个清单成员逐SHA核对,清单SHA256 `1474690b1e5df5a01a67e29974d791d2003502dc0ca841c6a232557ef167112d`。
|
||||
|
||||
下文测试、运行、来源重建及质量结论均为**来源报告**。本次没有执行来源脚本/产品测试、打开私有Session、raw请求/响应、评分答案、来源数据库、真实Key或备份数据。绑定脚本只计算摘要,没有执行;不将公开元数据核对写成重新运行或完整源码审计。
|
||||
|
||||
## RC:有界自然授权与原read来源
|
||||
|
||||
RC-01/02工程通过,RC原回顾仍保留RC-03 pending、RC-04 partial、RC-05 passed;随后RT独立完成RC-03工程集成。两时点分别登记,不改早先状态。
|
||||
|
||||
服务显式 `enableNaturalLanguage` 后,以有界中英文JSON保留/更新/撤销规则识别同Session的原直接用户请求,原用户任务原文保持。它不是任意自然语言意图理解,不从模型、工具、引用/条件文本或评分答案取得授权,也不读取消息所提磁盘文件。
|
||||
|
||||
retain/update核对授权后唯一成功完整read的原call/result、路径、turn/step、从1连续行号、EOF、UTF-8范围及SHA。相对/模糊路径、部分/截断/重复或有损来源拒绝;同basename不同绝对路径不合并,active须明确update并重读,撤销保留无正文墓碑。64KiB来源上限还受完整记录状态和1024启发式总预算约束,不代表可把64KiB直接塞进摘要。
|
||||
|
||||
新增必需 `compaction/records-natural`,rule为 `json-file-retention/v1`,追加及重放均核原事件。writer仍V4,same-version格式ack独立登记;14模块旧读者的公开回放回执明确为 `SessionFormatUnsupportedError`。本次仅复核该回执,未执行旧读者或打开fixture。
|
||||
|
||||
## RT:内存整批发布与磁盘部分日志拒绝
|
||||
|
||||
RT-01~04工程通过。显式 `integrateCompaction` 才把记录服务接入BasicCompactionEngine;默认关闭。必需 `compaction/records-prepared` 紧接原compaction/start,绑定原Session、start、revision、活动记录ID/version及完整预算。原raw模型输出与最终有效checkpoint分开保留,原输入不补评分oracle。
|
||||
|
||||
`Session.appendBatch` 用原末序号CAS、整批预验证,提交所有内存事件/投影后才通知观察者,**不是磁盘原子事务**。JSONL仍沿原缓冲/flush写入,提交顺序为:
|
||||
|
||||
1. `compaction/summary`;
|
||||
2. 空正文replacement;
|
||||
3. 必需 `compaction/records-checkpoint`。
|
||||
|
||||
只有最后解释器核验后才形成有效上下文。停在prepared、summary或replacement的部分日志必须拒绝恢复/续办,不能静默补齐或重新派发。普通截断失败保持原历史,可另开新的合法摘要;不能把失败原件改写为成功事务。
|
||||
|
||||
writer仍V4,旧读者缺必需解释器时明确拒绝;公开14模块回放回执首先拒绝 `compaction/records-prepared`。本次不以该回执冒称逐一实测所有新事件,也不实施云schema、`/v1`或Workbench迁移。
|
||||
|
||||
记录一旦更新/撤销,自由正文可能残留旧值,当前策略保守拒绝后续压缩,要求显式审查新Session;不是任意正文清洗、自动复制或隐私删除。完整包装/记录/续办/消息合计≤1024沿原TokenMeter启发式,溢出拒绝,不截断、不增限;它不是provider精确tokenizer,也不是Workbench新增额度。
|
||||
|
||||
## RA:低压力多轮的授权接纳时序
|
||||
|
||||
RA-01~03修复了RT原型的实际跨回合缺口:原read完成后若尚未到压缩阈值,未提交授权会被下一直接用户消息取代。修前真实Loader/Loop复现失败和关闭负控制原件保留,不能靠第三摘要后补字段或重写原任务掩盖。
|
||||
|
||||
集成开启、auto=true的pre-step先等 `next()` 返回enter并确认未取消,再以ready模式提交已具完整read来源的授权;这发生在压力计量和后续直接用户append之前。ready只允许尚未完成且未失败的匹配来源等待;已返回但损坏/模糊/失败的匹配证据拒绝。原指令先读另一不存在文件、再读完整事实JSON的错误恢复仍可继续。真正压缩保留required模式;取消/拒绝proposal不捕获。auto=false的自定义调用方须在接纳新用户前显式捕获。
|
||||
|
||||
本次核对公开来源重建回执11项及其映射:原用户seq35、read call44/result45、call ID `ujyqRrgRztc8g6J2My5L6If30eig0xYC`、14字段与协调相符。这是来源使用已构建服务的重建证据,本次没有打开其原Session、第三摘要捕获、transcript或事实正文,未独立重演UTF-8数据提取;重建追加的事件也不声称原历史已经存在。原日志、第三摘要输入和评分oracle边界保持。
|
||||
|
||||
RA复用既有必需事件、投影、CAS和事务,没有新增schema或格式。香港schema6/本机schema7及原`/v1`不变;安装隔离公开回执仍记录日常bundle不含该原型。
|
||||
|
||||
## 检查、失败与checkpoint范围
|
||||
|
||||
| 来源检查 | 保留的结论与边界 |
|
||||
| --- | --- |
|
||||
| RC源码948项、最后98项 | 重叠,不合计1046;首次回归/夹具/类型及文档失败保留 |
|
||||
| RT相关30项、扩大Windows回归 | 扩大回归1576通过、1符号链接EPERM失败、10跳过;未提权或新增跳过,不称全套通过,两组不累计 |
|
||||
| RA相关248项、边界8项 | 重叠不相加;修前失败、来源11项与状态保护8项分开记录 |
|
||||
| 两SDK | scripted事件投影;Python RC2/RT4为直接断言,不是pytest或真实server/模型验收 |
|
||||
| 文档/构建 | RT/RA最终43文档及相关构建按来源通过;RC原doc-sync失败和快速检查范围保留 |
|
||||
| Desktop | RC/RT第九全局idle因并行GPU失败保持;RA九项/八配置只证明其 `2026-11-10T10:54:20.935443+00:00` 观察时点,不覆盖旧失败或认证持续idle |
|
||||
|
||||
RC投影CPU局部测量不证明完整Session、模型或产品性能收益。RT首次长路径复制失败和未完成pending目录保留;短目录重导出不能抹去首失败。公网DNS/TCP/TLS及健康短读按来源范围理解,不认证手机、多运营商、全天或本轮正式服务状态。
|
||||
|
||||
三份checkpoint manifest均与公开独立回执相符:
|
||||
|
||||
| 增量 | 来源报告成员/字节 | manifest SHA256 |
|
||||
| --- | --- | --- |
|
||||
| RC | 4256 / 108921286 | `904bf0b5ae1629900866a8fd43473a4232e19013c1ab0b73ae53279810bef009` |
|
||||
| RT | 4520 / 116286076 | `deaad418ce3ccaebff0064442540d2eaf6a638f8a0fedfb1dddee59e43dee7f2` |
|
||||
| RA | 4752 / 123666471 | `7e4c0b9909ac0b7eb46f3c10c0ff342afc828a348389f0c4831463a53b7f10c9` |
|
||||
|
||||
22/23/23项独立checkpoint检查和178/196/197份导出均仅来源报告;本次没有逐个重哈希归档/导出成员或重做凭据扫描,不将重叠归档成员相加成独立证据总量。
|
||||
|
||||
## 历史、工作区与接续
|
||||
|
||||
截至固定来源,三增量零新增模型/DELETE/费用。351历史任务/26撤销身份、12旧表、月50元、2957 micro CNY及4unknown保持,旧v9末项 `actual_micro=null`和较早wire `local_task_id=null`仍按原字节保留。v9传输、v10准入、v11质量、PR-01两候选失败未采用、v10原409正文/v6完整wire缺证、BC-03/AB-04/NL-03/NL-04及P8六partial/八阻断均不改。
|
||||
|
||||
上轮11份捕获、13个清单成员及15保护摘要复核一致,5条绑定记录原样保留;加入其说明、清单和完成回执后,本次18项保护链保持。原64个待提交文件状态/摘要不变,HEAD仍 `f7705d96347cb19ea4ceab745bb2408061ef7621`,未暂存、提交或推送。
|
||||
|
||||
协调快照中Mnemo C5d/C5e/C5f仅按来源记录登记;C5B-RED-001、C5D-DEP-001、C5E-RED-001仍开放,原Core未改。C5g为来源计划ready,不据此唤醒或执行。三产品身份/数据库/权限/偏好/费用独立,Core SQLite、J2封存/147冻结、W1资源/模板/独立留出及完整池缺口保持。
|
||||
|
||||
下一Workbench产品步骤仍为M3最终产品身份/版本、源码包/程序包/Inno Setup及独立安装,之后另冻正式升级和新真实样本。本批不启动新阶段,不改变B4b/M3顺序。开放透传与持久ask/auto/full、历史W/AD/AE/AG成功/网络/unknown、R/W真机边界、普遍重复读/缓存、Purge收益、长期及物理iPhone未验保持。
|
||||
|
||||
本轮仅新增本说明及独立`.qa`证据:零模型/工程/DELETE请求,不执行来源脚本或产品测试,不读真实Key/私有Session/raw请求响应/评分答案/来源数据库/备份,不查正式health、不改产品或配置、不重放unknown,不迁移/部署/恢复/安装/重启/发行、提交/推送、委派或跨线程回信。依赖接收不是新增执行授权。
|
||||
@@ -0,0 +1,55 @@
|
||||
# 三项目协调:明确失败事件尾页修复与候选取舍
|
||||
|
||||
2026-11-11(Asia/Shanghai)。根据协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的心跳来信,只读登记智枢 NFD/ISS-041 已部署修复和 EQD 候选取舍。本轮不启动模型/工程请求、产品测试、配置变更、发布或转发,不回信或改动协调任务。Workbench 仍沿 [1.25.13 交付](../QA-MNEMO-RELEASE-1.25.13.md) 和 M4 顺序推进,本条不是 Workbench 同类缺陷或现网行为的验证。
|
||||
|
||||
## 固定公开来源
|
||||
|
||||
五份来源已保存本地原字节,长度和 SHA256 均与来信一致;后续只读本地捕获。
|
||||
|
||||
| 来源 | 字节 | SHA256 |
|
||||
| --- | ---: | --- |
|
||||
| [NFD 验收回顾](D:/Ai-project/models/docs/platform/reviews/NODE-FAILURE-TAIL-acceptance.md) | 10359 | `0d4a7d85cf2a9b1b724ca32f36134f63f73017b91d1ade1fff56a17a0edd7cc5` |
|
||||
| [尾页契约](D:/Ai-project/models/docs/platform/NODE_FAILURE_TAIL.md) | 2253 | `a49f9a4bc5b3fbeb9edccbe0a81c8d568141452248258a40893bcdc7691511fe` |
|
||||
| [NFD 检查点 manifest](D:/Ai-project/models/docs/platform/archives/20261111T124445.301314Z-NFD-checkpoint-5c644323/manifest.json) | 20750 | `167f474d039487f0656650e1178e80da2aee208644e1dcf6404c58d87dda2eeb` |
|
||||
| [部署核对回执](D:/Ai-project/models/logs/nfd-20261111-a/deployed-verification.json) | 14022 | `6f9ad294e166ec554b4592ddf9e31ff1edfce5c7eaeb1989dc0db566cb93169d` |
|
||||
| [EQD 离线诊断回顾](D:/Ai-project/models/docs/platform/reviews/EXECUTOR-FAILURE-DIAGNOSTIC-review.md) | 9164 | `af067baa209d72173e85833f3a600054487b908c414836bc7c9d44c633e0a42c` |
|
||||
|
||||
本轮只核公开文档、manifest 和回执元数据。没有打开包/归档成员、原模型响应、私有 Session、评分答案、数据库、备份或凭据,没有运行来源脚本、重新评分或复验来源运行时。
|
||||
|
||||
## NFD 的确切修复范围
|
||||
|
||||
来源 `Nodes.dispatch` 每页读取 128 条持久事件。原明确 failed 分支只读一页便抛原错误,257/256/129 条分别漏尾 129/128/1 条;成功路径原本会排空。修复仅在**原 failure 存在、节点明确 failed、当前页满 128**时按原 sequence 继续读取,短页或空页后抛相同错误。failed 已不接纳新事件,排空对象是已有终态前缀,不是继续生成或重新派发。
|
||||
|
||||
原错误保持,不补成功 finish 或 `[DONE]`;unknown/expired 仍沿原即时处理,取消/删除、原身份/项目、租约、原键及计费守卫不变。它不是旧 CT03/`node_result_unknown` 或模型质量失败的根因证明,不倒改历史失败。每页 128 是来源事件分页实现,不是模型上下文/输出额度,也不设置成 Workbench 本地限制。
|
||||
|
||||
## 来源部署和证据边界
|
||||
|
||||
来源现网 release 为 `network-failure-tail-20261111T123434Z`,包 SHA256 `0aeebd5ff604f0461c280fc45ff68810ef6377e3dd567c7a4ef29ca206046d8c`。本轮只核该包身份在捕获 manifest 中的引用,没有重哈希或读取包内容。香港 schema6 无迁移、本机 schema7 保持;来源报告只改 `server/platform/network_nodes.py` 分支,其他 61 生产成员不变。
|
||||
|
||||
| 来源结果 | 本轮登记范围 |
|
||||
| --- | --- |
|
||||
| 105 项精确兼容回归 | 来源报告通过,未重跑 |
|
||||
| Windows/WSL 同组 10 项 | 包含 loopback HTTP/SSE、页边界、原错误、取消与 unknown;同组不能相加为 20 |
|
||||
| 14 项部署回执 | 公开 `passed`/全部检查项、现网 release、62 个成员的核对标记、schema6/9 文件备份/原服务用户可读等元数据相符,不等于本轮连云复验 |
|
||||
| 26 项保护、9 项 Desktop | 按来源回顾登记,不重读其私有现场;本机生命周期 12 项与云包 11 项重叠,不累计为 117 独立样本 |
|
||||
| 117 成员检查点/23 项独立核验 | manifest 原字节相符,成员总字节 4578221;成员正文及独立核验按来源结果,未复验 |
|
||||
|
||||
manifest 保留留档时的 `phaseStatus=in_progress` 和当时回顾 SHA;最终回顾及有效指针在核验后追加。当前回顾 SHA 与快照时的 SHA 不同,不能据此改写旧 manifest 或声称所有成员都在本轮复验。
|
||||
|
||||
首次发布因权限 700 导致原非特权用户无法读取,健康失败后已回退旧代码;显式 755 修正及原服务用户读取检查、两候选包成员相同和后续成功均按来源记录保留。纯代码回退没有恢复旧账本或更改删除权威,首失败不能删掉。回执中的云端 `billing_unknown=4` 与本机当前活动计数是不同范围,不据后者为零将历史 unknown 清零。
|
||||
|
||||
普通 Windows 无代理/隧道/凭据的 API 短 GET200 与 console 两地址 TLS 超过本次 3 秒限时分别保留;这是来源短测,不认证持续公网、手机或完整业务,也不是本轮发送的请求。现网部署/短测/保护结论均只说明各自来源时点,未查询当前运行态。
|
||||
|
||||
## EQD 候选取舍与未验证事项
|
||||
|
||||
EQD 为零新模型的离线诊断,结论 `passed_offline_diagnostic_candidate_not_adopted`;当前 35B32K 执行候选已结束、不采用。EQT 严格 7/8(正向 3/4、负向 4/4)、首请求 168.429 秒和原质量失败保持;不剥围栏、放宽评分或将离线通过写成模型质量修复。不同缓存/冷热/时点的历史时延不能当同步随机对照或持续性能证据。
|
||||
|
||||
默认 4B、普通 35B4K、生产工具 allowlist 和正常 Desktop 不变,不把结束执行候选解释成删除已有权重、环境或产品能力。本条不要求 Workbench 切模型、改提示或重开任何实验。ISS-029/038/023、EQT/HCV/CPC/RS 及 P8 六 partial/八阻断保持;[上一轮 RS 的 7/7 交付、0/4 续办质量与来源保留缺口](MNEMO-RESUME-SUCCESSOR-READONLY-2026-11-11.md)不因 NFD 的局部修复改签。
|
||||
|
||||
三产品认证、数据库、全局权限、偏好及费用独立;unknown 只按原身份/原键查询,不重发或补造成功。来源月 50 元/2957 micro CNY/四 unknown、完整窗及其32K/2048/1024冻结条件保持,不移植成 Workbench 额度。记忆桥默认关闭及凭据仅宿主内存、真实 Mnemo 生产绑定/语义收益、长期、物理 iPhone、普遍重复读取/缓存与 Purge 收益未验证等边界不变。
|
||||
|
||||
## 本轮本地登记
|
||||
|
||||
开始时 HEAD `30ad156abfe2cfaee92e8b990f519b3562add446`,工作树干净,origin 跟踪仍为 `2f2b459c1860b4b3ba6670ce6c5a80dab3b4cb4e`;没有联网核远端、查正式 health、重建或安装。原发行身份、1.25.13 选择性证据 manifest、RS 登记和其 manifest 的字节摘要保持。
|
||||
|
||||
独立捕获位于 `.qa/coordination-readonly-2026-11-11-nfd/`,九个清单成员/75939 字节;[登记](../../.qa/coordination-readonly-2026-11-11-nfd/registration.json)、[清单](../../.qa/coordination-readonly-2026-11-11-nfd/MANIFEST.json),清单 SHA256 `882175846a879830c39489d3f8ba3d852aad2f12c812fb3da51ebfc170d94d12`,本轮逐件重读核对。只新增本说明、路线参考入口和限定本地文档提交,不推送/发布、改配置或程序,不调用模型/工程/DELETE,不运行产品测试/来源脚本、不委派/回信/新增心跳;M4 及既有开发次序保持。
|
||||
@@ -0,0 +1,244 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"observedAt": "2026-11-11T15:52:43.878Z",
|
||||
"incomingSourceThreadId": "01a1ae3d-66a7-78b1-945e-b0c1d42c28ae",
|
||||
"scope": "Read-only existing source/protocol metadata at the current M4 phase boundary.",
|
||||
"sources": [
|
||||
{
|
||||
"path": "D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-11T1531.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-11-nro-m4/coordination.json",
|
||||
"bytes": 34914,
|
||||
"sha256": "099435ceecdc8b1d7aca3931d327cde7fea23cbfd0a56f3a94450d0ae9612753"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/models/docs/platform/reviews/NETWORK-RECOVERY-acceptance.md",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-11-nro-m4/acceptance.md",
|
||||
"bytes": 15151,
|
||||
"sha256": "7df5be780207cdf8bbb61f021beaf1f81e61762cb00e8fa9616b1f9c6843f8fe"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/models/docs/platform/NETWORK_RECOVERY.md",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-11-nro-m4/network-recovery.md",
|
||||
"bytes": 3327,
|
||||
"sha256": "8e361cbbaa535685b144c869e5047242796d74d9298edcc9a6e239294b080ec0"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/models/logs/nro-20261111-a/deployed-verification.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-11-nro-m4/deployed-verification.json",
|
||||
"bytes": 13924,
|
||||
"sha256": "8c944b6ad1a9929caed44445cca26bb9e2e3a316473284eff4f6c07f2d8f1e64"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/models/docs/platform/archives/20261111T153928.158953Z-NRO-checkpoint-bb0bca42/manifest.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-11-nro-m4/checkpoint-manifest.json",
|
||||
"bytes": 40489,
|
||||
"sha256": "1d7bec284761aa1148a2eca64058eb4a9e54ab9dcc67d0be86cd8cc20349e2c1"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/models/logs/nro-20261111-a/checkpoint-independent-a3.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-11-nro-m4/source-independent.json",
|
||||
"bytes": 15344,
|
||||
"sha256": "d42736c9dff1d8f92da09b0df43061b97d4e2b2a851fb08f47f40f78915389bb"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/models/logs/nro-20261111-a/checkpoint-safety-a3.json",
|
||||
"copyPath": ".qa/coordination-readonly-2026-11-11-nro-m4/source-safety.json",
|
||||
"bytes": 40383,
|
||||
"sha256": "dd1da74d7c0231109cc6532d3cf5bcb60bbd9d077df82d1010fbb156d7b506ed"
|
||||
}
|
||||
],
|
||||
"sourceNro": {
|
||||
"status": "completed",
|
||||
"reviewVerdict": "passed",
|
||||
"acceptance": {
|
||||
"AC-NRO-01": "passed",
|
||||
"AC-NRO-02": "passed",
|
||||
"AC-NRO-03": "passed",
|
||||
"AC-NRO-04": "passed"
|
||||
},
|
||||
"release": {
|
||||
"scope": "http_response_framing_and_optional_node_request_id",
|
||||
"release": "network-recovery-20261111T145334Z",
|
||||
"releaseSha256": "582d3d79aaecd15fb03db9d2076af5907b8020497eced5fab8d06d6f847d675e",
|
||||
"manifestSha256": "728486fce605502480a272ca006f2dbe57f16b297e8a2e5a57645dd2907615cc",
|
||||
"schemaVersion": 6,
|
||||
"schemaMigration": false,
|
||||
"changedProductionSources": [
|
||||
"server/platform/network.py"
|
||||
],
|
||||
"unchangedProductionMembers": 61,
|
||||
"sdkConsumer": {
|
||||
"path": "examples/network_client.py",
|
||||
"beforeSha256": "4d0556cd17f3ceaa614c13de55b920eec9ac8e3b94cf0f54a1e5913f240f7017",
|
||||
"sha256": "e1e21be5c5e84ccbff78e10d3375dacb3b5d8e88c4ff82faf8703d31be921bcc",
|
||||
"includedInProductionTar": false
|
||||
},
|
||||
"windowsChecks": 12,
|
||||
"wslChecks": 12,
|
||||
"windowsRelatedChecks": 41,
|
||||
"exactSchema6Checks": 130,
|
||||
"deploymentChecks": 14,
|
||||
"protectionChecks": 26,
|
||||
"protectedTasks": 418,
|
||||
"protectedIdentities": 35,
|
||||
"protectedLocalTasks": 2875,
|
||||
"protectedLocalNonProcessTables": 35,
|
||||
"protectedOtherSources": 153,
|
||||
"desktopChecks": 9,
|
||||
"confirmedRemoteDeployments": 1,
|
||||
"rejectedReservationAttempts": 2,
|
||||
"managedLocalReloads": 1,
|
||||
"totalLocalStartCalls": 2,
|
||||
"totalLocalStopCalls": 1,
|
||||
"normalDesktopRestored": true,
|
||||
"causeOfInitialRuntimeStop": "unestablished",
|
||||
"modelPosts": 0,
|
||||
"deleteRequests": 0,
|
||||
"newPaidMicroCny": 0,
|
||||
"newDownloads": 0,
|
||||
"recordsComplete": true,
|
||||
"recordsFinalizationPending": false,
|
||||
"checkpointFinalizationPending": false,
|
||||
"completedAt": "2026-11-11T15:41:59.800278+00:00"
|
||||
},
|
||||
"wire": {
|
||||
"incompleteContentLengthDespiteValidJsonOrGzip": "original transport_or_response_unknown; discard connection; original identity/key GET only, never rePOST",
|
||||
"providerRequestId": "optional audit metadata; missing writes NULL, existing preserved; native/usage remain required",
|
||||
"publicProtocolVersionChanged": false,
|
||||
"schemaChanged": false,
|
||||
"originalLedgerRewritten": false
|
||||
},
|
||||
"counting": "Windows/WSL same12; source130 contains12/41, not additive independent counts; source27 is immutable-checkpoint verification including limited frozen12/old6 replay.",
|
||||
"checkpoint": {
|
||||
"manifest": {
|
||||
"path": "D:/Ai-project/models/docs/platform/archives/20261111T153928.158953Z-NRO-checkpoint-bb0bca42/manifest.json",
|
||||
"bytes": 40489,
|
||||
"sha256": "1d7bec284761aa1148a2eca64058eb4a9e54ab9dcc67d0be86cd8cc20349e2c1"
|
||||
},
|
||||
"digestMatchesState": true,
|
||||
"members": 230,
|
||||
"bytes": 6076658,
|
||||
"sealedPhaseStatus": "review_pending",
|
||||
"sealedReviewSha256": "2d72b753bf3360466f4da261fe8741b1a69ce0bc6158ea9e66b97ac8498eb35c",
|
||||
"currentFinalReviewSha256": "7df5be780207cdf8bbb61f021beaf1f81e61762cb00e8fa9616b1f9c6843f8fe",
|
||||
"sourceIndependent": {
|
||||
"path": "D:/Ai-project/models/logs/nro-20261111-a/checkpoint-independent-a3.json",
|
||||
"bytes": 15344,
|
||||
"sha256": "d42736c9dff1d8f92da09b0df43061b97d4e2b2a851fb08f47f40f78915389bb"
|
||||
},
|
||||
"sourceIndependentChecks": 27,
|
||||
"sourceIndependentAllPassed": true,
|
||||
"sourceSafety": {
|
||||
"path": "D:/Ai-project/models/logs/nro-20261111-a/checkpoint-safety-a3.json",
|
||||
"bytes": 40383,
|
||||
"sha256": "dd1da74d7c0231109cc6532d3cf5bcb60bbd9d077df82d1010fbb156d7b506ed"
|
||||
},
|
||||
"sourceSafetyKnownCredentialMatches": 0,
|
||||
"sourceSafetyValuesExported": false,
|
||||
"coordinatorMembersOpened": 0,
|
||||
"coordinatorCredentialScanPerformed": false,
|
||||
"coordinatorFrozenReplayPerformed": false,
|
||||
"ownMemberBytesOpened": 0,
|
||||
"ownRuntimeReplay": false,
|
||||
"ownCredentialScan": false
|
||||
},
|
||||
"public": {
|
||||
"sourceAnonymousApiStatus": 200,
|
||||
"apiSeconds": 4.453,
|
||||
"sourceConsoleDnsTcpTls": true,
|
||||
"sourceConsoleHeaderTimeoutSeconds": 8,
|
||||
"sustainedPublicQualified": false,
|
||||
"mobileQualified": false,
|
||||
"coordinatorLiveProbe": false
|
||||
},
|
||||
"oldQuality": {
|
||||
"strict35bSevenOfEightFailed": true,
|
||||
"candidateEndedNotAdopted": true,
|
||||
"P8SixPartialEightBlockedRetained": true,
|
||||
"oldRD04CauseClosed": false,
|
||||
"originalPlatformEqtHcvCpcRsQualityFailuresRetained": true
|
||||
},
|
||||
"initialStopRootCause": "unestablished"
|
||||
},
|
||||
"verification": {
|
||||
"sevenPublicMetadataCopiesExact": true,
|
||||
"fourAcceptanceValuesMatched": true,
|
||||
"deploymentFourteenReceiptMatched": true,
|
||||
"manifest230CountAndShaMatched": true,
|
||||
"independent27AndZeroMatchReceiptBound": true,
|
||||
"ownADIndexHashesExact": true
|
||||
},
|
||||
"adReference": {
|
||||
"identity": {
|
||||
"projectId": "674354be-ffb2-4f2f-b2ce-8b2c4960b008",
|
||||
"sessionId": "af80ce72-1289-4323-8be2-fafbaa210e01",
|
||||
"taskId": "6be88bee-1995-424a-b15a-2e0e54207d1a",
|
||||
"runId": "2ff20f93-7a85-4117-9615-b71c976624f8"
|
||||
},
|
||||
"allowedFileCount": 46,
|
||||
"structuredAcceptance": [],
|
||||
"missingDescriptors": [
|
||||
{
|
||||
"artifactId": "841b3dd5-e766-42f3-b372-0afdb7e77c82",
|
||||
"reason": "Check log descriptor absent from the sealed session artifact preview; independent-final logs remain in the read scope."
|
||||
},
|
||||
{
|
||||
"artifactId": "e96c4509-32ba-419b-8248-59cc5bb23342",
|
||||
"reason": "Check log descriptor absent from the sealed session artifact preview; independent-final logs remain in the read scope."
|
||||
}
|
||||
],
|
||||
"terminalGitObserved": true,
|
||||
"internalDeliveryReportStatus": "in_progress",
|
||||
"nativeEngineeringEvidenceV2Envelope": false,
|
||||
"selectedAGIdentity": {
|
||||
"projectId": "4070c3e0-a292-45ba-bdbc-bcb4160f0fd1",
|
||||
"sessionId": "91d910d3-7849-4f43-a0d5-1cfa019c34d7",
|
||||
"taskId": "42aacbd1-f1fd-418c-80a9-9e1af9ff42c8",
|
||||
"runId": "b330b966-848a-496b-a232-6b4438383ccc",
|
||||
"workspaceId": "11771ddf-9dc1-4ba4-987f-34cc3d80c2ce",
|
||||
"workerId": "2539daaf-ae13-4506-ac95-227032f606b0",
|
||||
"baseCommit": "c213ad68236af508bf9b31e9739483caeda4ef0c"
|
||||
},
|
||||
"differentFromSelectedAG": true,
|
||||
"producerRepackaged": false,
|
||||
"consumerAcceptanceRerun": false,
|
||||
"ownIndexReferences": [
|
||||
{
|
||||
"path": "D:/Ai-project/REA逆向/Agent-Workbench/docs/integration/WORKBENCH-AD-NODE-READONLY-SOURCE-2026-11-11.md",
|
||||
"bytes": 2255,
|
||||
"sha256": "16bb2baaa459c696efff64f72ea13cf51374d38b3f2347fd43551f0080c18ed3"
|
||||
},
|
||||
{
|
||||
"path": "D:/Ai-project/REA逆向/Agent-Workbench/docs/integration/WORKBENCH-AD-NODE-READONLY-SOURCE-2026-11-11.json",
|
||||
"bytes": 25974,
|
||||
"sha256": "a0a1fdb246a07b3a64c160c307545bd1e00c2f182adcebaa50ab9f6f0f30c9f6"
|
||||
}
|
||||
]
|
||||
},
|
||||
"selectedAGNotice": {
|
||||
"source": "Incoming coordination message only; no consumer runtime was rerun here.",
|
||||
"reportedLimitedPrecheckAndNineTests": true,
|
||||
"reportedSameGoalGuardInExistingContinuation": true
|
||||
},
|
||||
"workbench": {
|
||||
"m4Requests": 40,
|
||||
"m4ReportedTotalTokens": 629106,
|
||||
"m4UnknownRequests": 0,
|
||||
"existingPermissions": "unrestricted/passthrough/full; persisted settings unchanged",
|
||||
"providerOrEngineAdaptationWritten": false
|
||||
},
|
||||
"boundaries": {
|
||||
"checkpointMembersOpened": 0,
|
||||
"modelRequests": 0,
|
||||
"externalHttpRequests": 0,
|
||||
"credentialsRead": false,
|
||||
"otherProjectSourceWritten": false,
|
||||
"servicesRestarted": false,
|
||||
"oldSamplesReplayed": false,
|
||||
"messagesSent": 0,
|
||||
"authorizationExpanded": false,
|
||||
"mnemoProductionBindingVerified": false,
|
||||
"semanticBenefitVerified": false,
|
||||
"purgeBenefitVerified": false
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,11 @@
|
||||
# Mnemo 协调入口的 NRO 来源只读登记
|
||||
|
||||
2026-11-11,在当前 M4 阶段边界接收已有来源和协议变化。记录:[MNEMO-NRO-READONLY-2026-11-11.json](MNEMO-NRO-READONLY-2026-11-11.json)。七份公开元数据原字节副本位于 `.qa/coordination-readonly-2026-11-11-nro-m4/`,各 SHA/bytes 与来源索引一致;没有打开检查点成员、重放来源检查、读凭据、运行外部请求或回复来源线程。
|
||||
|
||||
NRO 来源 phase completed、review passed、四项 AC-NRO-01–04 均 passed;已部署 schema6 兼容包只改可选请求 ID 一处,其他 61 生产成员保持。节点成功仍要求 native/usage,provider_request_id 可选,缺项保存 NULL、已有值保持。SDK 以独立消费 SHA 登记,未包含在生产包中;固定 Content-Length 未读完即使 JSON/gzip 完整也保留 transport_or_response_unknown,丢弃连接并只沿原身份/原键 GET,不重 POST。此同步不改变协议、schema、账本或模型质量资格。
|
||||
|
||||
来源的部署 14 项回执、230 条清单及 SHA、27 项独立核验和凭据零匹配回执绑定同一 manifest,已核对元数据。Windows/WSL 为同组 12,130 包含相关 12/41,不能相加;27 是来源独立核验及其限定冻结重放,本轮没有重跑。检查点保留当时 review_pending,最终回顾 passed 是后续追加,两者不互相改签。
|
||||
|
||||
现成 AD 只读入口的两份索引 SHA 与协调记录相同;46 允许文件、structured acceptance 空、两个原检查日志描述符缺项,以及终端 Git 成功/内置报告 in_progress 保持。人类另选 AG Task `42aacbd1-f1fd-418c-80a9-9e1af9ff42c8` / Run `b330b966-848a-496b-a232-6b4438383ccc`;AD 仍是不同任务参考。本消息报告该 AG 的有限预检/原九测试及现有续办的同 Goal 守卫,不作为本轮消费者运行验收,也不重复补包或重派。
|
||||
|
||||
当前 M4 原连接、参数、权限及四十请求结果继续独立核验。来源 P8 六 partial/八阻断、旧质量、公网、手机与初始停机原因未知保留;不将来源工程通过升级成 Mnemo 生产绑定/语义收益或 Workbench/Purge 收益,不继承来源的本地额度。没有新增模型、工程、Git、配置、服务或消息授权;本轮仅在现有记录范围内登记。
|
||||
@@ -0,0 +1,49 @@
|
||||
# Mnemo 单项目记忆桥 v1:Workbench 只读反馈
|
||||
|
||||
2026-11-07,当前工作仍为 [I27/I28 CI](../M2-CI-PLAN.md)。本次只读合同与已有挂点,不接通服务、不修改配置/用户项目/旧发送工件,不作真实联调、模型质量或 Git 工程交付声明。本反馈仅保存本地,不因协调消息向其他对话发送回复。
|
||||
|
||||
已核对源文件 [Mnemo 契约](D:/Ai-project/Mnemo/docs/integration/project-memory-bridge/README.md),SHA256 与指定值逐字一致:`8bb95af9bdc2c6142c6c6c0ad3bb2c9b5a371417552bc08f881cefad7f2cc41f`。目录当前仅此 README,尚未取得共享 digest 实现/规范测试向量或消费端本次实际验证原件。协议是新增的 `mnemo.project-memory.v1`,不是已经接通的原生能力。
|
||||
|
||||
该核对是最初读取的时点观察。稍后准备只读封存时,源 README 已被并发更新,固定 SHA 核对失败即停止;新原文单独保存为 `.qa/m2-ci-2026-11-07/mnemo-project-memory-bridge-current-drift.md`,对应机器记录 `mnemo-readonly-drift.json`。未默默采用新版本或把最初观察改成当前仍一致,接入需新固定合同及共享向量后另验。
|
||||
|
||||
| 契约项 | 已有 Workbench 与需补的差异 |
|
||||
| --- | --- |
|
||||
| producerId + projectId | 项目已有持久 ID;服务重启 instanceId/文件路径不能作为 producerId。需单独持久 UUID 和显式单项目绑定,默认关闭。 |
|
||||
| 用户确认 / candidate | 本地记忆 `status=current` 与来源核对不能单独证明用户确认。需保存确认主体/时间;模型产出只映射 candidate,不能从已有状态推断 `by=user`。经验采用有自身修订,不能混同普通记忆确认。 |
|
||||
| sequence / version | 本地记忆 ID、createdAt/updatedAt 与经验修订不是桥的持久递增 sequence。需来源级事件序号、原版本与原键,并原子持久化;不得用墙上时钟或重启补造。 |
|
||||
| fingerprint / manifest | `ProjectMemoryManager.list/context` 已有实际来源和工程检查指纹。只复用本次完整核对的原值;远端摘要不能充当本地 sourceFingerprint。不能复用旧 context 工件生成新 manifest。 |
|
||||
| digest / 原键 lookup | README 指定规范 JSON 的 SHA256,但没有完整字段顺序、Unicode/null/遗漏规则的共享向量。接入前需固定该实现/向量 SHA;超时先按持久原键 lookup,unknown 不重发不同正文。 |
|
||||
| 状态 / 删除 | 已有本地归档、expired/unverified 和来源变化检测;需新增 withdrawn/deleted 持久事件与永久墓碑关联。删除不能通过新绑定/高序号复活,旧工件仍保留其原时点内容。 |
|
||||
| recall / validate | 已有 runtime 在 requestPlanner.prepare 前取得本地 projectMemory。远端需完整 manifest→recall→validate,最终核对须绑定本次物理请求及当次本地来源,召回后任何变化/到期/撤权都不使用旧远端条目。该核对不是新增模型执行审批。 |
|
||||
| 标签 / 历史 | 现有 context 标为 source-checked-project-memory。远端应独立 `authorized_project_mirror`,记录 Mnemo/来源 ID、修订、摘要、授权代次及当前请求绑定,不能伪装本地核对或改写历史发送原字节。 |
|
||||
| 凭据 / 存储 | 需独立桥凭据槽和引用,不能借模型 Key 或把凭据写日志、备份、请求工件。只能消费该项目接口,不扩大任务、模型、工具或费用权限。 |
|
||||
|
||||
具体源码挂点为 [项目记忆](../../src/core/memory/project-memory.mjs) 的 `context` 与 [Runtime](../../src/server/runtime.mjs) 的 `memoryManager().context → requestPlanner().prepare`。已有非完整扫描返回 null 的语义可以保持;失效远端记忆同样应省略,工程修复和用户完整历史继续正常发送,不能降级到旧缓存。
|
||||
|
||||
后续独立切片需要固定共享 digest/请求响应 schema/错误分类与消费端结果,增加持久绑定、来源事件及发送前二次核对。首验收只用新合成项目与受控 HTTP,覆盖确认同步和新会话实际请求正文、候选/项目/凭据隔离、原键冲突与 unknown、变更/归档/到期/删除/撤权、recall 后变化及旧发送 SHA exact;不重放历史工程、不发送付费模型、不以测试工程提交 Git。备份/恢复及 Mnemo 147 冻结由对方实际证据另列。
|
||||
|
||||
当前证据是上述源文件固定 SHA、对应源码只读观察及本地反馈;尚无消费端联调或 Workbench 原生接入通过证据。继续原 CI 及 M3/M4 路线,历史网络/unknown、缓存/重复读/真实交付/Purge 收益结论不变。
|
||||
|
||||
## 验收后补充规范的只读接续
|
||||
|
||||
随后收到 U2-B4a 消费端补充。重新固定读取七份指定文件,全部 SHA 匹配,并在 `.qa/m2-ci-2026-11-07/mnemo-readonly-final-1/` 保存本次原字节与 `registration.json`。其中初始 `README.sent-2026-11-07.md` 的 `8bb95af9...` 也取得原副本,不再以后来变化的 README 冒充初始正文;前文最初工具读取与漂移拒绝仍保留。
|
||||
|
||||
| 本次固定原件 | SHA256 |
|
||||
| --- | --- |
|
||||
| README.sent-2026-11-07.md | `8bb95af9bdc2c6142c6c6c0ad3bb2c9b5a371417552bc08f881cefad7f2cc41f` |
|
||||
| README.md | `777d0617f08bb30a4f1df9eb0b9d30c4ccb1246a1c1246c69846f913da24ceb7` |
|
||||
| PROTOCOL.md | `7689cf34ab14d4c5608aa87ee95ea4fd4b33bb65eac46fe44be6d916d3d34379` |
|
||||
| USER-GUIDE.md | `9f0e63efe7c70e25dfb8992982f16d989ea8890ddcc20c0cdd04ea572c5859d6` |
|
||||
| producer-project.example.json | `c2b9d21943c67664b0495ffcccd72975ae1809c457c96d07fb76e2984eb0e494` |
|
||||
| digest-vectors.json | `2ed45ff142981131b9442c76ebddc41625ee0270b4fc95a15fea5acb90bad7e0` |
|
||||
| project-memory-validation.json | `ace7d420248815a4b5bb1ad28ee7466d5b84b796cbf5f84b89593fe8d5e3f1cb` |
|
||||
|
||||
摘要的字段顺序、UTF-8/显式 null、ASCII 来源排序、checkedAt 排除及独立期限现在有明确规范。按原向量独立复算五组:三个 source/event 与原 JSON 字节、manifest 的排序/换核对时间、context/refs 都匹配,记录 `offline-vector-review.json`。这是只读字节分析,固定示例时间没有验证实时 freshness、旧快照期限或服务状态;未运行 Mnemo SDK/服务,也未复跑对方的 203/485/33 项和强杀验收。
|
||||
|
||||
再次核对当前源码:`ProjectMemoryManager.context` 仍只生成 `source-checked-project-memory`,含真实来源指纹/核对时间及本地条目,没有桥的来源 sequence/version/明确用户确认、连接授权代次、原键/永久墓碑。`Runtime` 的 `memoryManager().context → requestPlanner().prepare → provider.complete` 取得文档、封存请求计划并进入 Provider;发送回调只记录提醒,尚无桥的 recall/validate 绑定。这些挂点已存在,原生适配仍缺。
|
||||
|
||||
后续 B4b 需同时绑定本次完整本地来源、manifest、recall/snapshot、validate/contextDigest、Run/请求计划和实际物理发送 ID。连接描述本身不授权,正文不得从本地 `current` 推断用户确认;远端记忆独立标记 `authorized_project_mirror`。私有候选、永久墓碑、原键 unknown→lookup、续期/撤权、召回后变化、备份旧包未知删除均按新固定合同单独验。桥自身的字段/体积上限不能变成模型上下文或输出额度,也不能静默裁剪本地原正文;失效远端条目不回退旧副本,用户原对话和既往发送字节继续保持。
|
||||
|
||||
本次没有接通桥、改变工作台配置、读取凭据、请求模型、派发工程或回复源线程。仍先完成 I27/I28 CI,再按既定路线推进,Mnemo 联合验证另开新的合成范围。
|
||||
|
||||
2026-11-08 接续只复核上述本地七份固定原件与五组离线记录,首次只有 README、后续漂移拒绝和最终共享向量属于各自时点,未以当前文件替换原观察。此反馈单独提交;CI 最终冻结仍是 `4a03ec52`,文档提交不代表桥接服务或远端消费端通过。
|
||||
@@ -0,0 +1,90 @@
|
||||
# 三项目协调:RD04实际准入失败、中断审计与诊断边界登记
|
||||
|
||||
2026-11-11(Asia/Shanghai)。按协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 转来的新增依赖,只读核对公开证据并登记。本条不新增实现、实验、模型请求、迁移、发布或跨线程转发授权;[B4b默认关闭记忆桥](../QA-B4B-MNEMO-NATIVE.md)和[M3统一发行顺序](../M3-RELEASE-PLAN.md)保持。
|
||||
|
||||
**RD04-v2已实际执行并消耗,结果为 `failed_admission`,续办质量未评分。** 来源父级RD-01/02/03工程通过、RD-04 pending,与本次modelComparison的实际失败分别记录;不覆盖[上一轮准备/恢复兼容登记](MNEMO-RECORD-RETIREMENT-READONLY-2026-11-10.md)或[RQ的 failed_availability](MNEMO-RECORD-QUALITY-READONLY-2026-11-10.md)。正常Desktop退休原型仍默认关闭、未发布。
|
||||
|
||||
## 来源身份与一次捕获
|
||||
|
||||
来信[中断回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-RECORD-RETIREMENT-AB-interrupted.md)固定13653字节,SHA256 `49fb3726d590516c79a9c5eb87658b39d51de9c95bfcf9faaf17eee16d5f8b08`。本次当前全文14025字节、SHA256 `fc17b85e6e04798b1f0bf106b2b691e033c1072a29c3dad11b63b52228e83b42`;前13653字节逐SHA相符,末尾372字节为后续ND加载说明。完整捕获和派生固定前缀同时保留,派生文件不冒充独立捕获的历史全文,也不将全文变化隐去。
|
||||
|
||||
固定[协调快照](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-10T1658.json)76084字节,SHA256 `582c965a9dc6b8177d611de9ecdd2431210c4f85820667618259452651554348`。捕获的完成时间为上海2026-11-11 01:06:55.994(原 `2026-11-10T17:06:55.994Z`),投递状态 `tool_accepted_recipient_acknowledgment_pending`、`toolAccepted=true`、`recipientAcknowledged=false`;最终字节确认字段缺省,登记null,不推断后续确认。动态文件只读一次,后续核对均使用本地快照。
|
||||
|
||||
协调所读state为上海00:59:53对应版本(原 `2026-11-10T16:59:53.230978+00:00`),226924字节、SHA256 `f0d09c7e55b796dbf79d0bef7e415c868e47dec44913daad341ade26b8e1c875`。该身份仅来自固定协调记录;本次没有重新读取state或追读ND回顾。ND回顾在协调首次读取时ENOENT、增量仍in_progress,不能写成已完成收尾。新增文末指针也没有作为新验收授权或完成证据。
|
||||
|
||||
## 实际请求、失败与永久消耗
|
||||
|
||||
来源在自然idle及自有原acceptance预约内执行原v2:
|
||||
|
||||
| 事实 | 固定结果 |
|
||||
| --- | --- |
|
||||
| 模型提交POST | 2次,各一次 |
|
||||
| 已准入推理并交付 | 1次更新后摘要,attempt1、零价 |
|
||||
| 派发前拒绝 | 1次HTTP409 `no_eligible_target`,无task ID |
|
||||
| 首摘要 | 109.047秒;prompt23181 / completion288;完整445启发式token |
|
||||
| 后续 | 立即停止,剩余5请求未执行;续办质量未评分 |
|
||||
|
||||
`newModelRequests=1`指已准入推理,不把2次POST写成2次模型交付,也不把摘要可持久化写成续办或整体收益通过。AC-RD04-01仍passed、AC-RD04-02为failed_admission、AC-RD04-03为本次中断范围partial,原7请求完整对照未成立。
|
||||
|
||||
原HTTP409完整正文没有保留,`rawRejectedResponseBytesVerified=false`,拒绝细因未确认。后来的诊断不能补签原响应或追认根因。原report的来源报告SHA `9ca578b2f76005ae97326918b7f0f58e7b73d69e997d963da3d1b03c4e85cb43`、execution claim的来源报告SHA `bc99f2b075ad90f3142a7424f54f1e655551828ceb5784ddd5a5b350e453cc7c`永久保持;本次没有读取或重哈希这两份原件。
|
||||
|
||||
原v2永久消耗:不删claim、不换输出名、不复用原键重跑,查不到任务也不授予重发。新独立验证的输入、资源、声明和原预约由来源方按其另冻计划处理,本协调不执行或催办。
|
||||
|
||||
## 中断审计、单次清理与流式边界
|
||||
|
||||
公开中断审计15项通过;已准入任务的call/task/job/local/身份/原键映射、attempt1、input/native/modelVersion和零价相符。本次从公开审计、清理及独立复核元数据逐字段重算对应关系,未沿导出路径读取原请求响应、私有Session或评分答案。
|
||||
|
||||
原header加41个持久事件真实重开为42个,唯一新增边界是seq41空 `session/end-seed`。初次误要求计数全等而失败、修正及严格负例均保留;本次只登记来源审计的精确边界,没有重新打开事件内容或重跑Loader。
|
||||
|
||||
先完成审计,再对唯一归属合成任务执行1次DELETE、9项清理检查;随后独立verify-only8项、0次DELETE。三个报告的先后时间、同一任务/身份/原键、删除manifest及原writer回读均核对一致,临时身份已撤销。被拒键不删除,不重复DELETE,不触碰真实业务数据。
|
||||
|
||||
三端存储流的摘要相同且均0 chunks;这只证明该存储观察,不能证明原生SSE交付。备份按明确保留期限处理,恢复须核当前删除账本;provider副本不由本平台控制,不宣称所有副本已擦除,不认证Purge收益。
|
||||
|
||||
## 终态确认修复与实际加载分开
|
||||
|
||||
来源已证实成功任务收尾循环409 `already_complete`。修复仅在原认证job/lease、`stop_local`已证明停止,且额外GET返回同一原local task ID及终态后清本机journal;其他HTTP状态、stale lease、unknown、缺ID、读取失败或未停均保留。正常stop仍须 `accepted=true`,不改云结果/账本或重派。原journal此前由既有审计清理释放,不倒写为新修复的运行效果。
|
||||
|
||||
Windows18与WSL39相关回归为来源通过结果,范围有重叠,不相加成57个独立案例。WSL初次导入错误、混合编码包装器失败及修正回执保持。
|
||||
|
||||
关键[运行时/源码时序回执](D:/Ai-project/models/logs/node-diagnostics-loaded-runtime-split-20261111.json)1328字节、SHA256 `fe38acc8eceff68ada2f493c9cc2ac060c53400e3993a9dcd4c498bad4ef8cdd`已核对。上海00:55:51来源时点,PID18074早于新SDK、晚于终态修复源码,故 `terminalAckSourceLoaded=true`、`sdkDiagnosticSourceLoaded=false`。修复随已有并行重启加载,来源记录 `newRestartPerformed=false`;本协调没有查询实时进程、重启或卸载模型。
|
||||
|
||||
旧中断回顾上海00:21(原16:21Z)的“未部署”保持历史含义,后续00:55时序证据单列,不改旧文。终态收尾缺陷是否造成原 `no_eligible_target` 仍unconfirmed,不能建立未经证明的因果。
|
||||
|
||||
## 有界脱敏诊断、身份与负例资格
|
||||
|
||||
[SDK源码](D:/Ai-project/models/examples/network_client.py)31987字节,SHA256 `4d0556cd17f3ceaa614c13de55b920eec9ac8e3b94cf0f54a1e5913f240f7017`与来信相符;本次仅摘要读取,不更新或执行。新增 `CallError.diagnostic.admission_rejections` 仅用于HTTP409 / `no_eligible_target`,`schema_version=1`、`source=schema6_message`,收白名单reason和target_index,最多16项、原message≤8192字符;`format_recognized`、`unrecognized_targets`和`truncated`显式保留。不导出目标名、地址或原文,不授予retry。
|
||||
|
||||
Windows/WSL各38项为同组检查,38个相关案例,不能加为76;3状态检查与6探测门禁另按来源范围登记。本次没有运行。诊断新SDK尚未加载常驻执行器;fresh Python client具备诊断不等于常驻运行时已加载,两者分开。
|
||||
|
||||
两次admin到公网API的403证实service/node身份边界,不是服务身份请求成功;早先仅归因受管路径的解释已由来源纠正。console普通HTTPS负例返回409 `data_policy`,只证明该实际返回的脱敏解析;原预期 `invalid_max_tokens` 失败保持。
|
||||
|
||||
关键[console诊断资格回执](D:/Ai-project/models/logs/node-diagnostics-console-live-qualified-20261111.json)1732字节,SHA256 `b7c9f01d6848be53d49c3458908bb5a43cfa67b608f96e01a70d036229183ec3`已匹配。其 `parserLiveQualification=true` 与 `originalProbePassed=false` 并存;原键无calls/tasks、`resubmitSafe=false`,没有补做POST。来源本诊断含3次负例POST、0新增已准入推理,不把这些网络探测记成本次协调操作。
|
||||
|
||||
该资格不认证RD04原细因、实际服务身份/手机/推理/持续业务可用性,也不排除来源TUN影响。诊断仍in_progress,ND回顾首次未落盘的状态保持;本次不追读或唤醒来源。
|
||||
|
||||
## 证据、保护与历史保持
|
||||
|
||||
本次独立核对12份公开元数据、32条源码/公开回执SHA引用(29个不同路径)、9份检查日志摘要和2份manifest。重复引用主要来自终态逻辑和SDK在多组来源中的绑定,不能写成32份独立源码。原raw模型报告、execution claim、私有Session、原HTTP正文、评分答案、凭据、来源数据库和备份没有打开;公开审计内的raw路径只作指针,不跟随。
|
||||
|
||||
| 检查点 | manifest SHA256 | 成员 / 字节(来源报告) |
|
||||
| --- | --- | --- |
|
||||
| 恢复RD04R | `52ff5a82f4fbc76d4da3bf98ed026713339ff4c0a8c227b835331bd4706ccfef` | 41 / 2406886 |
|
||||
| 实际RD04T | `855591bdc877bc662d30cd4475ae040689543bcaad05a03fa4b6b5e785d13e0d` | 107 / 4150120 |
|
||||
|
||||
manifest原字节/SHA相符,16项/15项独立检查仍为来源报告,未重哈希归档成员或重做凭据扫描。历史无claim观察发生于真实执行之前,不能用它声明当前计划未消耗或允许执行。
|
||||
|
||||
来源保护从363任务/27身份追加为364任务/28身份,仅本次已审计的1任务/1身份;12旧表、月50元、2957 micro CNY、4unknown和旧nullable费用保持。旧363/27排除错误、路径canonical化早拒、首93成员档案相对链接错误、扫描问题及更正版107成员全部保留。账本旧unknown不因新增零价变0。
|
||||
|
||||
正常Desktop旧bundle仅按来源SHA保持,本轮未读取实际bundle/配置、不查4318 health。原RQ failed_availability、既有质量/网络/idle/EPERM及v9/v10/v11、PR-01、BC-03/AB-04/NL-03/NL-04、v6完整wire/v10原409正文缺证沿前轮登记保持。V4匹配reader、必需事件、部分落盘拒绝恢复、P8六partial八阻断、24×6000/14字段精确类型值、原生SSE/手机/持续P95/长期/普遍重复读与缓存/R-W真机边界/Purge收益均不代签。32K/2048/1024和来源预算只是冻结实验条件,不恢复为Workbench本地限制。
|
||||
|
||||
## Workbench与三项目边界
|
||||
|
||||
上轮21捕获/30清单成员、27条绑定、24保护摘要及54746字节固定协调版本核对一致;两个原长度前缀的派生身份继续保留。新增上轮说明/清单/完成回执后,本轮27项保护摘要相符。本轮起始67个待提交文件(含此前66项及上一轮登记说明)原状态/字节/SHA全部保持,HEAD仍 `f7705d96347cb19ea4ceab745bb2408061ef7621`,未暂存、提交或推送。
|
||||
|
||||
独立证据在 `.qa/coordination-readonly-2026-11-11T0107/`:17份来源捕获、1份明确派生的固定前缀,加登记脚本及元数据,共25清单成员,见[登记](../../.qa/coordination-readonly-2026-11-11T0107/registration.json)及[清单](../../.qa/coordination-readonly-2026-11-11T0107/MANIFEST.json)。清单SHA256 `99dc2235f9adf8743894e7e60f71bef565c3f13cb1b14a45a301280ab34d67d5`,旧封存根均未追加或改写。
|
||||
|
||||
协调快照中的Mnemo C5j/C5k仅在合成提取/维护历史范围完成,41表partial预检/5非空阻塞不代表完整C3生产adapter。C5lL0为来源在途工作,本轮不执行;C5B-RED-001、C5D-DEP-001、C5E-RED-001、C5G-MATRIX-001保持。Mnemo状态从234877字节/SHA `817c53b7e84b6e68baab4179a088b336d2eaaf17ea214a635e4bd931597ab890` 合法并行变化到240377字节/SHA `7fbca1a3b84f4fd6faf416e7285c4a655a162656590948fcfc07d2f68845a1c2` 仅按协调记录登记,没有追读或覆盖。三产品身份、数据库、权限、偏好、费用独立,Core SQLite、J2封存/147冻结及各自发行门槛保持。
|
||||
|
||||
Workbench下一仍为M3最终身份/版本、源码包/程序包/Inno Setup及独立安装,再另冻正式升级和真实样本;本次来信不启动这些动作。开放透传、完整历史、持久ask/auto/full与历史W/AD/AE/AG成功/网络/unknown完整保留。
|
||||
|
||||
本轮仅新增本说明及独立`.qa`证据:零模型/工程/DELETE请求,不运行来源脚本/产品测试、不读真实Key/私有数据、不查正式health、不改产品/配置,不重派unknown,不迁移/部署/恢复/安装/重启/发行、提交/推送、委派或跨线程回信。
|
||||
@@ -0,0 +1,56 @@
|
||||
# 三项目协调:第十长窗准入、节点探测与恢复依赖只读登记
|
||||
|
||||
2026-11-10(Asia/Shanghai)。按协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的合并来信登记公开依赖参考,没有启动来源实验或新工作。[B4b 七契约与25合成案例](../QA-B4B-MNEMO-NATIVE.md)、[M3 统一发行顺序](../M3-RELEASE-PLAN.md)及其未尽事项保持。
|
||||
|
||||
## 固定来源与核对范围
|
||||
|
||||
- [事件回执修复后第十长窗回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-RECEIPT-REVALIDATION-acceptance.md):22213 字节,SHA256 `82ebb2e3ab5b423ca69531f3f0b3511311f89c9736c4cacc563d48c6f7763dcd`。
|
||||
- [节点协议探测与实际备份回顾](D:/Ai-project/models/docs/platform/reviews/NODE-CONTRACT-PROBE-acceptance.md):11487 字节,SHA256 `92fac3a2695fea43c89d6cfff7dbbba7885eacb200e35abf031b75db3163c5ee`。
|
||||
- [动态协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-10T0451.json):绑定本次单次读取的原字节快照,其 `completedAtUtc=2026-11-10T05:00:21.814Z`、投递状态为 `tool_accepted_recipient_acknowledgment_pending`。这不代表接收确认或核验后续最终协调字节。
|
||||
- Mnemo 的[完整池选材回顾](D:/Ai-project/Mnemo/docs/integration/2026-11-10-full-pool-selection-review.md)与[完整 Runtime 预检回顾](D:/Ai-project/Mnemo/docs/integration/2026-11-10-runtime-preflight-review.md)均按协调所引 SHA 捕获,未核对动态 status、重跑工程或释放冻结。
|
||||
|
||||
13 份公开 JSON 元数据的长度/SHA 全部匹配,20 条来源 SHA 绑定与报告内来源字段一致;20 条原 key/云 task/node job/本机 task 及历史、清理映射核对相符。检查条目通过仅作为公开回执事实,不是本次重新运行验收。
|
||||
|
||||
证据在 [registration.json](../../.qa/coordination-readonly-2026-11-10T1300/registration.json) 和独立 [MANIFEST.json](../../.qa/coordination-readonly-2026-11-10T1300/MANIFEST.json),20 个清单成员均已复算。绑定脚本只计算摘要,没有执行;未读取来源数据库、私有会话、原始 SSE、真实 Key、备份数据或恢复根。
|
||||
|
||||
## v10:20 个实际任务,另有两次准入拒绝
|
||||
|
||||
来源记录 **22 次 HTTP、20 个实际模型任务**。20 个均 succeeded / attempt1 并完整交付;4 段资料、3 份持久摘要和三次摘要后工具通过按来源报告登记。其余两次是不同逻辑键,分别为 summary1024 和 main2048,各单 POST 返回 HTTP409 `no_eligible_target / node_tool_contract_unavailable`,`retryable=false`、无 task_id。
|
||||
|
||||
公开拒绝证明的写入端只读查询记录两键在 network_calls 及原任务表均为零行,本次仅核对该证明,未重新查源数据库。拒绝不造任务,也不造模型执行或删除回执,不能把22次HTTP写成22次推理。
|
||||
|
||||
**RR-02 仍为 `failed_admission`**,24段与最终14字段/类型/值没有完成;RR-01/03为passed。原24×6000字、17 bundles/58工具、32K/2048、0.875阈值、4096保留及1024摘要硬限保持。这些属于来源冻结实验,不是 Workbench 新增限额。
|
||||
|
||||
清理前,原任务63项、已接纳任务的成功 ordered wire 204项、持久历史63项均标记passed。但旧采集器只保留两次拒绝的解析错误和SHA,**原HTTP409正文未保存,永久缺证**。不把JSON重新编码当原响应,不补签旧失败wire。新有界原字节采集器只服务未来窗口,不能补证v10;`rawRejectedResponseBytesVerified=false`保留。
|
||||
|
||||
v10故障与独立视频/运行时维护有时间交叠,尚未证明唯一根因。不用过期成功声明绕过准入,不重放拒绝键或unknown,不降原标准。旧v9 CT-03、BC-03/AB-04/NL-03/NL-04失败、v6完整wire缺证及[较早wire末项local_task_id=null与最终映射](MNEMO-CURRENT-TASK-COORDINATION-READONLY-2026-11-10.md)的时点差异全部保持。
|
||||
|
||||
## 节点探测修复与在途维护互斥
|
||||
|
||||
来源CP工程增量已部署并按其回顾判定passed;协调的原结构化状态仍记`review_pending`,不替来源改状态。探测只补元数据GET,至多2次、共享2秒预算;仅短暂传输及502/503/504允许补读,权限/格式/契约错误直接拒绝。当前失败撤下legacy/v2/容量声明,不复用旧成功声明;有界脱敏contract诊断随心跳发布。
|
||||
|
||||
推理POST、receipt、事件游标、租约、Session/API/schema不变。Windows59、WSL59及备份相关27项是来源报告,本次未重跑。短普通HTTPS读取不认证全天/手机/多运营商,也不唯一解释旧v10。
|
||||
|
||||
**维护窗口与长窗的持久跨进程互斥仍在来源方开发,尚未验收。** 不把正在实现的停止新准入、持久阻断或嵌套保护当现有能力。自然idle只是采样时点,不保护请求间隙;来源下一完整窗由其既定门槛处理,本次不催问或另起实验。
|
||||
|
||||
## 清理、费用与真实恢复的边界
|
||||
|
||||
20个任务各DELETE一次、去重仍20;独立verify-only44项、顶层及逐任务DELETE均为0。本次没有发删除。累计保护为**272任务/23撤销身份**,与本轮20实际任务及2无task_id拒绝分开;原12表、香港schema6单一权威、本机schema7独立按公开报告保持。
|
||||
|
||||
月上限50元、历史2957 micro CNY及**4个unknown**保持,旧v9末项`actual_micro=null`未改。新增付费0不等于未知项已结算0,也不清零历史费用。受管删除核对不认证外部副本普遍删除或Purge收益。
|
||||
|
||||
来源备份/独立恢复的公开回执均为8381文件、1120913193字节,manifest SHA `1fc95fee3ce8a1048aa4a7ace17aa5f735bae453f413748787d557b87bf6514c`一致。26处明确可重装合成运行依赖被省略,绑定包SHA;原链接/数据未移删,默认及业务/未知符号链接仍拒绝。恢复后历史合成运行需重装依赖,不把此快照写成包括所有运行依赖的完整副本。
|
||||
|
||||
恢复未切storage根、未改配置或启动第二权威,当前删除权威覆盖按来源登记;本次没有打开实际备份成员、源数据库或恢复数据。停机约11分钟为来源报告,尚需优化,不能认证公众低停机或真实业务迁移。
|
||||
|
||||
正常Desktop九检查/八配置、保护21项与自然idle只在对应来源完成时点有效,早先视频运行中的资源失败原件保留。RR/CP两个checkpoint的manifest SHA已匹配;全部归档成员、备份成员和服务停机时长未在本次独立重验。无效review哈希候选、并行状态保护拒绝和资料扫描时点失败按来源回顾保留,不用最终工程收尾覆盖它们。
|
||||
|
||||
## Mnemo 与 Workbench 后续
|
||||
|
||||
W1j仅在原六来源有空位时补一条字面经历/纠正,保留六留五反证和三项满额缺口;开发材料覆盖9→11/14不代表答案收益。W1k已接完整Runtime信封、原模型桥纯body和中央原键五态只读核对;`engineeringStatus=verified`仍与`quality blocked`并存,不产生发送许可或默认切换。
|
||||
|
||||
提供方模板、未消费独立留出、真实质量均未验;下一C5为合成`memory_quality`语义预检,不据本条执行。三产品身份/数据库/权限/偏好/费用独立,Mnemo正式Core与冻结保持。Workbench开放透传和持久ask/auto/full不变,历史W/AD/AE/AG成功/网络/unknown、R/W真机边界、普遍重复读取/缓存与Purge收益unknown保留。
|
||||
|
||||
本次同时核对上一登记的19捕获文件、22清单成员、23绑定和6保护摘要均一致,不改旧字段或旧原件。M3产品commit/版本、程序与安装验收等未尽事项不因来源工程passed而完成。
|
||||
|
||||
本轮仅新增本说明及独立`.qa`证据:零模型/工程/DELETE请求,未查正式health、改产品/配置或其他项目,未部署/恢复/安装/重启/发行、提交/推送、委派或跨线程回信。接收记录不构成新的发送授权。
|
||||
@@ -0,0 +1,72 @@
|
||||
# 三项目协调:显式记录生命周期的格式与恢复依赖登记
|
||||
|
||||
2026-11-10(Asia/Shanghai)。按协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的新增公开依赖,只读复核 PR-02A 原型及恢复兼容参考。本次没有启动实现、模型、工程任务、实验、迁移或服务操作,[B4b 默认关闭的记忆桥](../QA-B4B-MNEMO-NATIVE.md)与 [M3 统一发行顺序](../M3-RELEASE-PLAN.md)保持。
|
||||
|
||||
来源 **RL-01~04 工程原型 passed**,整体仍 **partial / review_pending**。没有部署到日常 Desktop,默认 BasicCompactionEngine 和正式摘要均未启用;自然语言授权、真实 compaction 事务、撤销后自由摘要策略与模型收益仍未验收。这个新时点更新[上一登记](MNEMO-COORDINATED-LONG-PINNED-READONLY-2026-11-10.md)中的“原型在途”范围,不改旧原件,也不将工程通过写成产品质量通过。
|
||||
|
||||
## 固定来源与本次复核
|
||||
|
||||
| 来源 | 字节 | SHA256 |
|
||||
| --- | ---: | --- |
|
||||
| [原型回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-RECORD-LIFECYCLE-prototype.md) | 34588 | `345a6ff373dc028d8e8c5ebb895f856aeab6577f827473563741bd1ba03c8b2b` |
|
||||
| [接口与恢复约定](D:/Ai-project/models/docs/platform/integrations/HARNESS_RECORD_LIFECYCLE.md) | 4085 | `3f8dbfd2fe5aaca131c0736f64288b0651dd4b99e8992ab676923cc31818943a` |
|
||||
|
||||
[动态协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-10T0849.json)按单次原字节捕获:26447字节,SHA256 `e0ebbbc48002e09553c4e7e805b55340318412d3ab1f4742e9f95d39d002df8b`。固定快照的 `completedAtUtc=2026-11-10T08:55:08.379Z`,投递状态为 `tool_accepted_recipient_acknowledgment_pending`,`toolAccepted=true`、`recipientAcknowledged=false`;未提供 `finalCurrentRecordBytesAcknowledged` 字段,不补推其值或后续最终接收状态。
|
||||
|
||||
本次核对两文档、8份公开元数据的长度/SHA,5条来源绑定与各报告 `sources` 及实际摘要一致。固定11份来源快照,证据见 [registration.json](../../.qa/coordination-readonly-2026-11-10T1656/registration.json)与 [MANIFEST.json](../../.qa/coordination-readonly-2026-11-10T1656/MANIFEST.json)。13个清单成员逐SHA核对,清单 SHA256 `0078e373d659ac3f3d547c78e62d2463e80b918df4056952343550d0c919cb16`。
|
||||
|
||||
下面的测试、回放、SDK、运行和质量结论均为来源报告。本次独立复核限于公开原字节、元数据、来源绑定及前后记录一致性;没有执行来源脚本或重跑验收,也没有打开私有 Session、原 SSE、来源数据库、真实 Key 或备份数据。
|
||||
|
||||
## V4 的必需事件与旧读者恢复
|
||||
|
||||
来源 Session writer 仍为 **V4**,新增必需事件 `compaction/records`,保存完整变更后状态,`eventIgnorable=false`。可选 `CompactionRecords` 服务与纯 `records-projection` 入口不等于默认启用。匹配解释器须在追加与重放时验证完整状态和来源;缺解释器的旧读者必须明确拒绝,不能忽略事件或丢掉记录后继续恢复。
|
||||
|
||||
持久化 acknowledgment 的五处变化与协调所引字段逐项一致:
|
||||
|
||||
| 根 | 变化 |
|
||||
| --- | --- |
|
||||
| `event:compaction/records` | 新增必需事件根 |
|
||||
| `event:agent/inbox/spliced` | inserted 消息来源新增 `compaction-records` union 分支 |
|
||||
| `event:developer/message` | message 来源新增该分支 |
|
||||
| `event:user/message` | 消息来源新增该分支 |
|
||||
| `event:session/title-llm-request` | messages union 变化 |
|
||||
|
||||
来源报告已用实际旧 storage reader 及14个冻结模块独立回放,明确得到 `SessionFormatUnsupportedError`,不是用新读者代替旧读者。Workbench 本次核对公开回放回执的模块数、错误及 acknowledgment,未再次运行模块或读取 Session fixture。这是来源 Session 格式变化,**不是云 schema 或 `/v1` 升级,也没有实施 Workbench 迁移**;香港 schema6、本机 schema7保持。
|
||||
|
||||
回退参考为不挂载可选服务,同时保留已生成的合成日志和匹配读者;不能把必需事件改为可忽略、覆盖旧 Session 代际或修账本消除失败。当前旧原生工程接入范围没有因此扩大。
|
||||
|
||||
## 用户来源、版本与撤销语义
|
||||
|
||||
当前原型只接受同一不可变 Session 内、原始 append、`source.kind=user` 的单文本完整结构化操作,含 Session 身份、记录 ID、`expectedVersion`、put/revoke及原工具来源。创建须 `expectedVersion=0`,更新/撤销匹配当前版本;操作一次消费,较新的直接用户消息使待执行旧操作失效。
|
||||
|
||||
解释器使用原工具 call/result、call ID、turn/step、成功状态、文本块索引、UTF-8范围及SHA重验提案;来源必须早于操作。模型、工具、替换副本、引用、旧摘要和继承 fork 均不能自授用户权威。同名标签不合并,跨 Session 需要另行显式导入。
|
||||
|
||||
来源 codec 支持声明范围内的 JSON、从1连续编号的 JSON及 file-json;file-json只解析已记录的 content,不读取其中 path。当前 JSON仅支持可 `JSON.stringify` 往返的词法表示,重复键、有损数字、负零及其他未支持表示明确拒绝,不宣称所有合法JSON均兼容。
|
||||
|
||||
撤销留下**无正文版本墓碑**并停止后续模型可见记录传播;原始用户/工具 append-only 日志仍保留。因此撤销不是隐私删除或外部副本清除,也未证明任意自由摘要已清除撤销内容。自然语言授权识别与真实摘要事务正在来源方推进;不能用本原型的新 JSON操作替换原长窗的自然语言验收指令。
|
||||
|
||||
## 容量与验收边界
|
||||
|
||||
来源 `composeCheckpoint` 当前只做有界组合:完整记录、ID/版本/标签包装、完整续办正文、checkpoint前后包装及消息开销合计≤1024,沿既有 TokenMeter启发式计价。溢出明确拒绝,不截断、不增限,不提交新摘要或失败的记录事件。它不是 provider精确tokenizer,也不是 Workbench新增的上下文/输出上限;开放透传默认保持。
|
||||
|
||||
来源886项工程检查后定向42项重验存在重叠,**不合计成928项**;两SDK只是 scripted事件投影,Python为直接断言,不是pytest或真实server/agent-loop验收。首次 doc-sync 34通过9失败、再38通过5失败及后续定向通过分别保留,不冒称最后全跑43项。
|
||||
|
||||
最终浏览器源码打包正例和故意导入Host的 `node:path` 负例按来源通过;最初误读旧lib的负控制、旧读者构造器误用、离线依赖/Node16入口拒绝、CLI并发SIGKILL及修复前失败保持。相关修复没有在本仓库重做或扩大执行范围。
|
||||
|
||||
**最终 Desktop passed=false**:前8项归属/配置/HTTP等通过,8份配置摘要保持;第9项 `local_globally_idle_resource_guard_clear`失败,报告同时记录并行GPU工作、idle=false及租约held。此前空闲时点的通过不能覆盖此失败,本次未打断并行工作或申请维护,也不声明当前全局idle。
|
||||
|
||||
167份导出元数据的条数、路径唯一性及长度和5529523字节一致;仅复核清单信息,没有逐件重哈希167份导出。新 checkpoint manifest长度807449字节、SHA256 `f0afde03014b9c96bbceb373437433fc0aefe10766ece6034567e0ac2658cb67`相符;4012成员/102486864字节及19项独立核验按来源报告登记,本次未遍历成员或重做凭据扫描。
|
||||
|
||||
## 历史与工作区保持
|
||||
|
||||
本增量按公开保护回执为零模型/DELETE/新增费;累计351历史任务/26撤销身份、12原表、月50元、2957 micro CNY和4个unknown保持。旧v9末项 `actual_micro=null`及较早wire `local_task_id=null`仍按原字节保留,不补写或重放。
|
||||
|
||||
v9传输、v10准入、v11质量失败,PR-01两候选失败未采用,v10原409正文与v6完整wire缺证,BC-03/AB-04/NL-03/NL-04及P8六partial/八阻断全部保持。撤销、受管清理和checkpoint收尾均不认证Purge收益、外部副本清除或完整窗质量。
|
||||
|
||||
上轮29份捕获/31个清单成员已逐SHA复核,34条绑定记录原样保留,12个保护摘要一致;加入其说明、清单及完成回执后,本次保护链15项。工作区原63个待提交文件(M3+B4b及既有登记)保持状态与摘要,HEAD仍为 `f7705d96347cb19ea4ceab745bb2408061ef7621`,未暂存或提交。
|
||||
|
||||
协调快照中的Mnemo C5b源缺陷仍开放;C5c为另建合成清理候选,未改原Core,C5d普通开发在途。没有据此追读动态status或执行Mnemo工作。三产品身份/数据库/权限/模型偏好/费用独立,Mnemo Core SQLite、J2封存与147冻结保持。
|
||||
|
||||
下一Workbench产品步骤仍为M3最终版本/产品commit、源码包/程序包/Inno Setup及独立安装,再另冻正式升级和新真实样本。本次只登记依赖,没有开始新阶段或改变B4b/M3顺序。持久ask/auto/full、历史W/AD/AE/AG成功/网络/unknown、R/W真机边界、普遍重复读/缓存、Purge收益、长期及物理iPhone未验保持。
|
||||
|
||||
本轮只新增本说明及独立`.qa`证据:零模型/工程/DELETE请求,不读真实Key、不执行来源脚本或产品测试、不查正式health、不改产品/配置或其他项目、不重放unknown,不迁移/部署/恢复/安装/重启/发行、提交/推送、委派或跨线程回信。新增依赖记录不构成执行或发布授权。
|
||||
@@ -0,0 +1,73 @@
|
||||
# 三项目协调:记录事务有界收益与可用性阻塞登记
|
||||
|
||||
2026-11-10(Asia/Shanghai)。按协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的新增公开依赖,仅只读核对并登记。本轮不新增实现、实验、模型调用、迁移、发布或跨线程转发,[B4b 默认关闭记忆桥](../QA-B4B-MNEMO-NATIVE.md)与 [M3 统一发行顺序](../M3-RELEASE-PLAN.md)保持。
|
||||
|
||||
来源已有**一轮真实有界续办收益**;整体采用仍 **failed_availability**,RQ-02 partial、state review_pending、modelQualityPassed=false,正常Desktop未启用。更新/撤销后的压缩仍在派发前拒绝,不能把安全拒绝写成业务成功,也不能把这次局部收益写成完整窗或生产资格。
|
||||
|
||||
## 固定来源与核对范围
|
||||
|
||||
| 来源 | 字节 | SHA256 |
|
||||
| --- | ---: | --- |
|
||||
| [RQ质量回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-RECORD-QUALITY-diagnostic.md) | 7683 | `554fa46ee0789c60b058fb226851ed54fc34cfe56cd4002e73b9f7116734906a` |
|
||||
| [有界对照契约](D:/Ai-project/models/docs/platform/integrations/HARNESS_RECORD_AB.md) | 3096 | `4b0947aa134617f266eed7e98c3ca4cf516c3b50f944033b043126be9cfe226e` |
|
||||
|
||||
[动态协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-10T1254.json)按单次原字节捕获:37258字节,SHA256 `090e055895bff4ef0a41a1124e4a977fd54a587364c248b8336b4ba6237ab260`。固定版本 `completedAtUtc=2026-11-10T12:59:49.867Z`,投递状态 `tool_accepted_recipient_acknowledgment_pending`、`toolAccepted=true`、`recipientAcknowledged=false`;最终字节接收确认字段缺省,不补推后续状态。早读时尚未完成的状态没有冒充本次固定版本。
|
||||
|
||||
本次复核10份公开元数据、22条SHA引用(19个不同路径)、12条原key/call/task/job/local映射及新checkpoint manifest,均相符。22条引用明确包含各报告 `sources`/计划 `sourceBindings` 的20条,以及模型报告→计划、计划→复用源码manifest两条;不是22份独立源码。
|
||||
|
||||
13份来源快照、独立 [registration.json](../../.qa/coordination-readonly-2026-11-10T2101/registration.json)及脚本共15个清单成员,见 [MANIFEST.json](../../.qa/coordination-readonly-2026-11-10T2101/MANIFEST.json),已逐SHA核对。清单SHA256 `d604ff978aa00cb5698e29b2705df07a9395eea135b9822516508f1e43a1c244`。
|
||||
|
||||
下文模型、质量评分、实际提交/重开及来源检查均为**来源报告**;本次独立工作限于公开原字节、摘要、元数据映射、顺序和数量复算。未执行来源脚本、重跑验收或重新评分,未打开私有Session/raw请求响应/评分答案/归档成员。公开计划中的assessment-only字段未用于输入构造或重评分,也没有沿报告中的capture路径读取正文。
|
||||
|
||||
## 四来源、同raw两路径的真实有界对照
|
||||
|
||||
来源完成12个新零价请求,全部一次交付:每个来源各1次新摘要,同一raw分别经普通路径和记录事务真实提交、JSONL flush、另一Loader重开,再各1次真实后续召回。四来源各3次,共12个不同请求,不是重复重试;也不是四轮独立完整长窗。
|
||||
|
||||
| 来源 | 普通路径召回 | 记录事务召回 | 完整启发式摘要用量:普通/事务 |
|
||||
| --- | --- | --- | --- |
|
||||
| 原第三摘要14字段 | 失败,返回JSON但字段不符 | 全部字段/类型/值精确 | 301 / 458 |
|
||||
| 已完成普通任务 | 通过,精确新响应标记 | 通过 | 223 / 230 |
|
||||
| 同名不同来源 | 通过,两原值及顺序 | 通过 | 384 / 522 |
|
||||
| 工具注入 | 通过,原值未改写 | 通过 | 313 / 385 |
|
||||
|
||||
这些数值来自真实模型报告的四份bridge元数据,未混用较早CPU准备夹具的摘要用量。两路径实际durableReopenIdentical均为true,八次召回中的事务四项与普通三项按来源评分通过;原14字段普通路径失败保持。确定性记录附加本身不代替后续模型质量证据。
|
||||
|
||||
原用户/read字节、tool call ID、此前摘要与第三摘要逻辑输入保持,只重排seq及引用;插入的是本轮来源决策,不声称旧日志已经存在。评分oracle仅用于独立评分,不进入bridge/输入构建,不从事实磁盘补字段。本次未重新打开这些原始输入或答案,所以不把公开通过回执写成独立内容重验。
|
||||
|
||||
完整摘要包括记录、包装、raw续办正文和消息开销,均≤原1024启发式总限;原system、58工具、采样、32K/2048条件保持,不截断、不缩工具或增限。该限额是来源冻结实验条件,不是Workbench新增上下文/输出限制,也不是provider精确tokenizer。
|
||||
|
||||
## 仍阻塞采用的可用性与验证范围
|
||||
|
||||
更新/撤销来源在本地工程预检即保守拒绝压缩,模型派发0。它避免旧版本自由正文继续传播,但不能满足原会话更新/撤销后的可续办需求,故RQ-02仍partial、整体failed_availability。容量、截断/空/工具摘要负例继续拒绝且保持原表面,不能通过扩预算或修改原标准制造可用性通过。
|
||||
|
||||
RQ-01/03 passed与RQ-02 partial、engineeringVerdict=passed、modelQualityPassed=false并存。正常Desktop未启用、未发布,P8六partial/八阻断保持。这次非流式诊断不认证原生SSE、原24×6000完整窗、手机/多运营商、持续/P95、一般长期质量、并发或生产资格。
|
||||
|
||||
12次请求等待合计 **101.983秒**,冷首摘要 **29.735秒**,已由12条元数据复算。固定执行顺序、少样本与热缓存影响保留,不从部分更快推出持续性能或缓存收益。请求usage合计prompt254203、completion1445、total255648与协调相符;这是来源报告用量,不补造真实计费。
|
||||
|
||||
## 审计、清理与费用保持
|
||||
|
||||
原审计104项的结束时间早于本轮清理开始;12任务各DELETE一次、清理53项,独立GET复核41项开始于清理结束后,逐任务DELETE均为0。原call ID、云task、node job、本机task、request key及两份受管清理manifest对应一致,五类身份各12唯一。临时身份已撤销,使用12次;本次没有发模型/工程/DELETE请求。
|
||||
|
||||
来源保护累计 **351任务/26身份→363任务/27身份**,与本轮12新增任务分开。12旧表、香港schema6单权威/本机schema7、月50元、2957 micro CNY及4个unknown保持。新增费用0不将unknown结算为0;旧v9末项 `actual_micro=null`与较早wire `local_task_id=null`仍按原字节保留。
|
||||
|
||||
受管清理不证明供应商、外部副本或备份普遍销毁,不证明Purge收益。原v9传输/v10准入/v11质量、PR-01两候选失败未采用、BC-03/AB-04/NL-03/NL-04、v6完整wire及v10原409正文缺证全部保留。
|
||||
|
||||
## 检查、留档与在途依赖
|
||||
|
||||
来源桥接43项、严格评分13项、状态保护8项、保护24项及checkpoint24项按报告登记,本次未重跑。桥接预检的provider未注册、toolCallId误取、两路径夹具ID不一致、system缺ID重开拒绝和路径前缀重复等原失败均保留,不把修复前预检算作模型请求。
|
||||
|
||||
最终Desktop九项/八配置只在来源 `2026-11-10T12:04:15.743679+00:00` 时点有效;最初idle租约未释放失败、旧RC/RT idle失败及Windows符号链接EPERM不覆盖,不认证持续或当前idle。本次不查询正式4318 health,也不读取已安装bundle或配置文件。
|
||||
|
||||
新checkpoint manifest长度987474字节、SHA256 `3e2b7b1c29ffbec206424441ca77b4fc41331115321e24ec49b9ae0d3b3046f2`匹配。4812成员/125610932字节及24项独立核验按来源报告登记;本次未逐件重哈希归档或复用197份RA导出,未重做凭据扫描。来源合并记录所述SC/Krea合法并行更新保持,未追读或改写动态state。
|
||||
|
||||
RD-01同Session、来源驱动的更新/撤销后安全续办当前仅见设计、事务骨架及恢复检查在途,不登记为完成或新质量通过,也不额外唤醒。已验V4必需事件、部分落盘拒绝恢复与auto=false调用方显式捕获要求仍依[前轮格式/恢复登记](MNEMO-NATURAL-TRANSACTION-ADMISSION-READONLY-2026-11-10.md),没有新发布接口资格。
|
||||
|
||||
## Workbench工作区与后续
|
||||
|
||||
上轮20份捕获、22清单成员与18保护摘要复核一致,15条绑定记录原样保留;加入其说明、清单及完成回执后本次21项保护链保持。原65个待提交文件(M3+B4b及既有登记)状态/摘要不变,HEAD仍 `f7705d96347cb19ea4ceab745bb2408061ef7621`,未暂存、提交或推送。
|
||||
|
||||
协调快照中的Mnemo C5g完成范围仅为简单配对sandbox历史语义预检,31partial语义表/15其他已填充域阻塞不等于C3生产adapter计数;C5B-RED-001、C5D-DEP-001、C5E-RED-001、C5G-MATRIX-001仍开放。C5h为来源ready计划,未据此执行。三产品身份/数据库/权限/偏好/费用独立,Core SQLite、J2封存/147冻结及各自质量/发行门槛保持。
|
||||
|
||||
下一Workbench步骤仍为M3最终产品身份/版本、源码包/程序包/Inno Setup及独立安装,之后另冻正式升级与新真实样本。本批不改变B4b/M3顺序;开放透传及持久ask/auto/full保持。RQ有界记录续办收益与普遍重复读/缓存、R/W真机边界、Purge、长期及物理iPhone未验分别记录,历史W/AD/AE/AG成功/网络/unknown不改。
|
||||
|
||||
本轮仅新增本说明与独立`.qa`证据:零模型/工程/DELETE请求,不运行来源脚本或产品测试、不读真实Key/私有数据、不查正式health、不改产品/配置或其他项目、不重放unknown,不迁移/部署/恢复/安装/重启/发行、提交/推送、委派或跨线程回信。依赖接收不构成新增执行授权。
|
||||
@@ -0,0 +1,79 @@
|
||||
# 三项目协调:同会话上下文退休与恢复兼容依赖登记
|
||||
|
||||
2026-11-10(Asia/Shanghai)。按协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 转来的恢复兼容依赖,只读核对公开资料并登记。依赖接收不增加实现、实验、模型调用、迁移、发布或跨线程转发授权;[B4b 默认关闭记忆桥](../QA-B4B-MNEMO-NATIVE.md)和 [M3 统一发行顺序](../M3-RELEASE-PLAN.md)保持。
|
||||
|
||||
来源 **RD-01/02/03 隔离工程通过,RD-04 真实模型 pending**;整体 `review_pending`、`modelQualityPassed=false`,正常 Desktop 默认关闭、未发布。工程机制不能覆盖[上一轮 RQ 的 failed_availability](MNEMO-RECORD-QUALITY-READONLY-2026-11-10.md),也不能代替更新、撤销后的真实续办收益证明。
|
||||
|
||||
## 固定版本与追加差异
|
||||
|
||||
| 来源 | 来信固定字节 / SHA256 | 本次捕获 |
|
||||
| --- | --- | --- |
|
||||
| [RD 工程回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-RECORD-RETIREMENT-prototype.md) | 35224 / `5e21de6958ef56c3db60fdb9be558f635082f0b9d5d689ac34ff626365cd473d` | 全文相符 |
|
||||
| [恢复契约](D:/Ai-project/models/docs/platform/integrations/HARNESS_RECORD_RETIREMENT.md) | 4526 / `b3d87b66f5c5c3a8c42923cb9d5b8987544b31a8fa1cc74bce178e2990528ff7` | 当前5398字节;原4526字节前缀精确相符,末尾追加872字节 |
|
||||
| [RD04 准备回顾](D:/Ai-project/models/docs/platform/reviews/HARNESS-RECORD-RETIREMENT-AB-prepared.md) | 8666 / `e961db642bed5179d8137ab80f78a47dc06be8754b10ce0ecdc3621a3368c031` | 当前9004字节;原8666字节前缀精确相符,末尾追加338字节 |
|
||||
|
||||
当前两份完整文档SHA分别为 `98f9500f20fda9676563b3eb9098197072fefe920560e3624bd7fee471998fdb`、`57bffa83f2fe9a4cda1800850cd01d2897f549fefffbbb77a0022ae376df3a07`。第一轮整文件相等断言失败的[原日志](../../.qa/coordination-readonly-capture-2026-11-10T2306-attempt1.log)保留,没有改成通过或覆盖捕获。随后仅从同一次捕获提取固定长度前缀并验证来信SHA;两个 `requested-prefix.md` 明确是派生文件,不冒充独立捕获的历史整文件。
|
||||
|
||||
追加内容涉及只读中断观察工具。本轮未沿新增链接读取恢复回顾、执行观察器或重验其14项/3项报告,不把追加文字升级为本轮新增恢复通过;来信所述恢复补验仍按在途边界登记。
|
||||
|
||||
[协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-10T1452.json)按单次原字节固定:54746字节,SHA256 `f5ebefa21901266e23b16a7f5845e8a7ec19cb18b87a0d96c4cfb102bb676464`,`completedAtUtc=2026-11-10T15:05:34.531Z`。捕获版本中投递状态为 `tool_accepted_recipient_acknowledgment_pending`、`toolAccepted=true`、`recipientAcknowledged=false`;最终字节确认字段缺省,登记为null,不推断后续确认。没有再次读取动态文件替换身份。
|
||||
|
||||
## V4 必需事件、读取器与部分落盘
|
||||
|
||||
`enableContextRetirement` 是另行启用的原型选项,要求 `integrateCompaction`。原 Session 追加顺序固定为:
|
||||
|
||||
1. required `compaction/records-retirement-prepared`。
|
||||
2. 原 record decision。
|
||||
3. 空 user replacement。
|
||||
4. required `compaction/records-retired`。
|
||||
|
||||
writer 仍V4,云schema6、本机7与原 `/v1` 不变。公开持久化回执的6处 root/union 变化与协调记录逐项相符,已有 `same-version` acknowledgment。host、browser/detached和SDK必须使用匹配的reader;旧14模块的两次探测只分别证明对新discriminator拒绝,不代表完整新Session的旧读者语义验收。
|
||||
|
||||
`appendBatch` 由纯解释器整批预验、重建来源闭包并计价,通过后才发布内存;JSONL仍需flush,不是文件系统原子事务。三种部分落盘前缀会使host拒绝恢复与派发,独立部分投影隐藏选定旧生成正文。不得删prepared、补造closure、强制降级reader或重派unknown;恢复必须围绕原持久结果和匹配解释器处理。本登记没有执行恢复。
|
||||
|
||||
## 保留的任务语义与容量边界
|
||||
|
||||
来源闭包保留完整原single-text直接用户请求、当前记录及原工具完成状态;工具按原turn/step/callId配对,保留原call/result seq、名称与成功/失败,不保留参数或正文。成功工具不能因正文退休而重做,也不等于整个用户任务已经完成。
|
||||
|
||||
原用户文本、系统头和追加日志仍保留,用户原文可能含旧值,不能声称隐私或历史删除。来源、继承或身份不完整、opaque/purged替换、额外高优先级、注入/多模态、未结工具、进行中压缩或容量不足时拒绝整项更新,不截断完整用户请求。
|
||||
|
||||
完整请求、记录、包装及消息开销仍按来源TokenMeter启发式≤1024核对;不是provider精确tokenizer,也不证明完整24×6000窗口可装入。32K/2048、58工具和最多7请求均为来源实验的冻结条件,**不是Workbench新增限制**;本产品开放透传、完整历史及持久ask/auto/full保持。
|
||||
|
||||
## RD04-v2 准备与整计划防重放
|
||||
|
||||
RD04-v2固定计划4993字节,SHA256 `ee244221614a7299ffd049301e4ed781d6f58583854b565bd21987320c954f69`;17份执行依赖和复用源码manifest均有原SHA绑定。来源报告44项built/JSONL、17项评分/原键/清理/整计划保护、3项状态检查通过;本次仅复核公开回执、绑定与数量,没有重跑这些检查。
|
||||
|
||||
固定计划路径的exclusive-create claim和flush/fsync先于身份创建及模型POST;换输出名不能重新运行同一计划。恢复仅查询原输出、原身份和原键,不能删除占用文件。旧v1和118成员准备档案保留,当前源码使旧计划SHA校验正确拒绝;首版只约束单输出、可通过换名绕过的缺陷和修复记录保留,未发生真实重发。
|
||||
|
||||
最多3次摘要、4次继续,共7请求,单请求180秒、总窗1500秒、零价上限属于来源方计划,本协调不请求执行。更新/撤销比较的是保守拒绝与退休后的新输入,不能称同输入A/B;普通未完任务两臂输入相同,共用一次新raw。旧摘要是明确标识的合成夹具,不是实际模型前史;通用继续不补答案,评分与输入构造分离。
|
||||
|
||||
实际模型、三端审计与合成清理仍pending。协调捕获中的恢复补验在途、原资源等待和原acceptance预约只代表来源时点;本轮没有探测当前资源、抢占、预约或唤醒任务,也没有运行计划、模型、审计或清理工具。当前追加文档内容不改变这份固定接收范围。
|
||||
|
||||
## 证据复核、旧失败与费用
|
||||
|
||||
独立核对15份公开元数据、27条来源SHA绑定(23个不同路径)、两份checkpoint manifest及另两份测试日志摘要。27条明确为两份保护报告各4条、Desktop报告1条,以及计划17依赖加1份复用manifest;已与报告原字段对应,脚本只做摘要读取,没有执行。持久化回执先验证唯一已知命令回显前缀,再严格解析单个JSON,没有把任意诊断当JSON忽略。
|
||||
|
||||
163源码/SDK、7构建/2类型、43文档、built12、Python6直接断言及浏览器正负仍为来源报告。Python不是pytest或真实server。本轮未打开私有Session、raw请求响应、评分答案、数据库、凭据、备份或归档正文,也没有从计划的assessment-only字段构造输入或重评分。
|
||||
|
||||
| 来源检查点 | manifest SHA256 | 成员 / 总字节(仅来源报告) |
|
||||
| --- | --- | --- |
|
||||
| RD | `b6547e20c8e5855289de8edc46f6c1815c6a1e259b5f7c90253ddea47c51ddd9` | 4969 / 130128021 |
|
||||
| RD04-v2 | `d5098d858167e765ee402b30dbca8c5e80ce8980e6399ffd41b809c29aacb4a1` | 122 / 6807878 |
|
||||
|
||||
两份manifest本身已按原长度/SHA核对;24项与15项独立检查按来源报告登记,未逐成员重哈希或重做凭据扫描。最初RD独立候选因命令回显被当成纯JSON而失败的原回执保留,后续通过不改写首次失败。
|
||||
|
||||
正常Desktop在来源 `2026-11-10T13:24:30.902378+00:00` 时点的8配置、前8检查通过,第9全局idle失败;准备最终readiness在 `2026-11-10T14:12:01.375347+00:00` 仍 `canStartOwnModelAcceptance=false`。独立GPU工作和预约阻塞保留,不据此声明当前idle或完整Desktop通过。本轮不查正式4318 health,不读取实际安装bundle或配置。
|
||||
|
||||
363原任务、27身份、12旧表、香港schema6单权威、本机7、月50元、2957 micro CNY与4unknown保持;本增量新增模型/DELETE/费用0,零新增不把旧unknown或nullable费用结算成0。原RQ failed_availability、v9/v10/v11、PR-01两候选失败、旧完整窗/质量失败、EPERM/idle、v6完整wire/v10原409正文缺证和BC-03/AB-04/NL-03/NL-04沿前轮登记保留。P8六partial/八阻断、原14字段类型值、原生SSE、手机、持续/P95、长期业务、普遍重复读/缓存、R/W真机边界与Purge收益不代签。
|
||||
|
||||
## Workbench保护与后续
|
||||
|
||||
上一轮13捕获/15清单成员、22条绑定及21保护摘要已核对;旧37258字节协调版本及原65文件保护继续保持。本轮再纳入上一轮说明、清单与完成回执,共24项保护摘要相符;工作区本轮开始时66个待提交文件的原状态、字节和SHA不变,HEAD仍 `f7705d96347cb19ea4ceab745bb2408061ef7621`。
|
||||
|
||||
新独立证据根为 `.qa/coordination-readonly-2026-11-10T2306/`,21份来源捕获、2份明确派生的固定前缀及登记工具/元数据,共30清单成员,见[登记](../../.qa/coordination-readonly-2026-11-10T2306/registration.json)和[清单](../../.qa/coordination-readonly-2026-11-10T2306/MANIFEST.json)。清单SHA256 `fe2e31fc19e5c111b7cc45e7399216138a82188e4232952bc1cfea3087149926`。本轮只读预览曾因本地const重名失败,修正后读取通过,原工具输出保留;它不是来源工程或产品失败。
|
||||
|
||||
协调快照所述Mnemo C5h/C5i仅为有界合成评测收尾/捕获历史准备,`autoIndex=false`/`capture.receive`、36partial表预检/10其他阻塞,不代表完整C3或生产adapter。C5j是来源下一包,J0在另行授权开发中,本轮不执行;C5B-RED-001、C5D-DEP-001、C5E-RED-001、C5G-MATRIX-001仍开放。三产品身份、数据库、权限、偏好及费用独立;Core SQLite、J2封存/147冻结及各自发行门槛保持。
|
||||
|
||||
下一Workbench步骤仍为M3最终产品身份/版本、源码包/程序包/Inno Setup及独立安装,之后另冻正式升级和新真实样本;本次不改变B4b/M3顺序,也不据依赖来信启动下一实施阶段。历史W/AD/AE/AG交付/网络/unknown及未验证事项完整保留。
|
||||
|
||||
本轮仅新增本说明与独立`.qa`证据:零模型/工程/DELETE请求,不运行来源脚本或产品测试、不读真实Key/私有数据、不查正式health、不改产品或配置、不重放unknown,不迁移/部署/恢复/安装/重启/发行、提交/推送、委派或跨线程回信。
|
||||
@@ -0,0 +1,63 @@
|
||||
# 三项目协调:RS 续办质量与跨摘要来源保留
|
||||
|
||||
2026-11-11(Asia/Shanghai)。按协调入口线程 `01a1ae3d-66a7-78b1-945e-b0c1d42c28ae` 的合并来信,只读核对公开来源并登记,供后续协作规划参考。当前 Workbench 交付保持 [1.25.13](../QA-MNEMO-RELEASE-1.25.13.md),下一仍是 M4 新真实兼容/自有工程基线;本记录不启动新开发、模型请求、实验、迁移、发布或跨线程转发,不唤醒来源方在途实现。
|
||||
|
||||
**RS 的 7 次请求已交付,四个续办严格质量为 0/4,`modelAcceptance=failed_quality`。** 修复后另冻结的新合成计划不等同于重跑旧 RV/RD04;不覆盖 [RD04 的 failed_admission](MNEMO-RD04-ADMISSION-INTERRUPTED-READONLY-2026-11-11.md) 或 [RQ 的 failed_availability](MNEMO-RECORD-QUALITY-READONLY-2026-11-10.md),也不授予旧键再次提交资格。
|
||||
|
||||
## 冻结来源及核对范围
|
||||
|
||||
| 公开来源 | 字节 | SHA256 |
|
||||
| --- | ---: | --- |
|
||||
| [RS 验收回顾](D:/Ai-project/models/docs/platform/reviews/RESUME-SUCCESSOR-acceptance.md) | 13409 | `9a072679ddee1bb6632a2694550433c4be34b796613f5fce68f6394bbae278e5` |
|
||||
| [原任务来源诊断](D:/Ai-project/models/logs/rs-task-loss-diagnostic-20261111.json) | 8829 | `c3c757d8e535c2d96e220ebd5efed73fe06e1f7fdaca6a3d13da635c3356cec7` |
|
||||
| [协调记录](D:/Ai-project/Mnemo/docs/integration/coordination-2026-11-10T2101.json)的本地捕获版本 | 62773 | `fc9db228df3cb34b7d014b0e1dd46965dd73facb60b79fa5eda69dd291cd44e9` |
|
||||
|
||||
前两份原字节与来信长度/SHA 相符。协调记录仍在来源方收尾:首次摘要观察为 59906 字节/`cd22efb42d07bb317435d73e6062cf3ffea2f51d1bea321b2ec4ba26f1855e5c`,随后一次完整捕获为表中版本;前者未保存全文,不重建或冒称两个不可变全文。后续核对只使用本地捕获。捕获版本记录工具已接受投递、接收方确认仍 pending,`replyRequired=false`;本轮不向源线程发送确认,也不改来源协调状态。
|
||||
|
||||
本轮直接核对三份公开捕获、诊断所绑定的三份现行实现 SHA,以及 RVP/RS 两份 manifest 原字节。现行 `records-retirement.ts`、`records-checkpoint.ts`、`summary-frame.ts` 的长度和 SHA 均与公开诊断一致,仅做摘要读取,没有执行或修改源码。
|
||||
|
||||
公开协调记录中的九份元数据/八条来源与公开文件绑定、七组请求/清理映射和独立审计结果,按来源报告登记。本轮只在已捕获的公开材料内交叉核对,不重新读取原 report、私有 Session、native 原文、评分答案、凭据、来源数据库或备份,也不沿其中的路径追读这些材料。不把元数据相符写成重新完成原始审计或模型评分。
|
||||
|
||||
## 交付、质量与输入可比性
|
||||
|
||||
| 事实 | 来源结果与含义 |
|
||||
| --- | --- |
|
||||
| 物理提交/任务交付 | 7 次单 POST,7 个原任务均 attempt1;3 个摘要、4 个续办 |
|
||||
| 续办质量 | 0/4;三个分支返回工具,另一分支未给要求的 JSON;`actualToolsExecuted=0` |
|
||||
| 输入关系 | 普通两臂共用同一 raw 摘要;更新、撤销输入不同,不称四臂完全同输入 |
|
||||
| 四个完整检查点 | 545/302/298/298 启发式 token,均不超过冻结 1024 |
|
||||
| 实际请求输入用量 | 约 19.8K~23.2K;通用预检约 103K 是估值,不是实际 tokenizer 或 32K 越界证据 |
|
||||
| 来源状态 | `review_pending`/`failed_quality`;原型未发布,未证明整体语义或工程收益 |
|
||||
|
||||
返回工具只是模型输出,不能写成工具已执行、工程已完成或整个样本通过。完整原输入、严格字段/类型/值评分和原质量失败保持;有限一次交付与时延观察不认证持续可用性、完整窗或生产资格。
|
||||
|
||||
## 来源覆盖缺口与隔离方案
|
||||
|
||||
公开诊断记录:完整原任务进入实际摘要输入,而四个后续持久检查点均未逐字保留完整原任务。`records-retirement` 重建请求与配对工具回执,后续 `records-checkpoint` 保留当前记录和模型摘要,`summary-frame` 从检查点之后的新输入解释当前任务。该证据说明来源覆盖缺口,不能断言每项语义全部丢失,也不能倒推旧 RD04 准入失败的根因。
|
||||
|
||||
来源方正在隔离离线研究的方案是:让来源绑定的原任务与已配对成功工具回执跨摘要持久保留,明确任务完成权威,保留更新/撤销及原优先级;模型生成摘要不能自行宣布任务完成。构造输入与评估分离,不注入 oracle 或评分答案,冻结来源容量不足时明确拒绝。该方案在捕获时仍在实施,未证明完成、发布、回退资格或质量收益,本轮 Workbench 不立即适配。
|
||||
|
||||
后续如进入相关记忆/长会话集成,应先核对新方案的来源身份、完成权威、多次摘要后的任务与工具配对保持、更新/撤销和失败/unknown 行为,再按独立冻结样本评估收益。这是未来集成的证据要求,不恢复 Workbench 的本地预算或自动压缩,也不改变默认开放透传、完整历史和持久 ask/auto/full。
|
||||
|
||||
## 原请求、清理、档案与费用保持
|
||||
|
||||
来源报告先完成 57 项审计,再对七个明确自有合成任务各单次 DELETE,33 项清理和独立 26 项、零 DELETE 复核通过。公开协调中的七组 call/task/job/local/原键具唯一映射,attempt1、input/native/modelVersion、零费用和同一删除回执对应;本轮仅核公开捕获中的这些关系,没有重做 DELETE 或读取原任务内容。
|
||||
|
||||
非流式调用与三端零 chunks 不认证 SSE。备份按原保留期处理,恢复须核对当前删除账本;provider 副本不由平台管理,不能宣称所有外部副本已删除。371 累计任务/30 身份、十二旧表、月 50 元、2957 micro CNY 和 4 unknown 均按来源报告保持,临时身份 used7/quota7 单次撤销。RS/RV/RD04/三探测 claim 永久消耗,旧键、unknown 和已确认 DELETE 均不重发。
|
||||
|
||||
| 来源档案 | 本轮仅核 manifest SHA256 | 成员/独立检查(来源报告) |
|
||||
| --- | --- | --- |
|
||||
| RVP | `eb7d879f7bd0ed9f1d298ee4c484b090e41d8905a15a173f63b03d0d460a972e` | 81/23 |
|
||||
| RS | `8f01f11016b27dc81b51e31e1bdf14aa657bff2592034b5bb1d84e818658741f` | 205/26 |
|
||||
|
||||
两份 manifest 原字节已独立核对,成员正文未重哈希、凭据扫描未重做,原独立检查只按来源结果登记。RS 首 198 成员档案的读取器误将 38 副本视为 38 独立来源,修正为 38 副本/34 来源、四重复一致;首失败、旧 candidate 与后续有效档案保持,不改变质量 0/4。自然卸载后只读检查误把 loaded=null 当对象的 TypeError 和修正也保持。
|
||||
|
||||
## Workbench 当前身份与独立边界
|
||||
|
||||
开始时本地 HEAD 为 `2f2b459c1860b4b3ba6670ce6c5a80dab3b4cb4e`,工作树干净,origin 同分支跟踪引用一致;本轮没有联网查远端。既有 1.25.13 实现 e306f40c、发行35a6d6d9、验收1e309700及正式 final 结果保持,不能将首正式核验与后两项旧断言 code1 写成多次安装:Setup 只有一次成功。186 相关测试不等于本轮全仓或真实模型质量通过。
|
||||
|
||||
既有选择性 782 份证据清单 SHA `611c4c40d9c28b09649d8a041505d1684242f436b7c8b972b7cfbb0c9c65260c` 已重核其 manifest 字节,本轮没有重验782成员、读取正式数据库/会话快照、程序或凭据,也没有新查4318 health。记忆桥继续默认关闭、单项目授权、凭据仅当前宿主内存;真实生产 Mnemo 绑定、语义收益、长期、物理 iPhone、普遍重复读取/缓存、R/W 真机边界与 Purge 收益保持未验证。
|
||||
|
||||
三产品身份、数据库、权限、偏好和费用独立;Mnemo CoreSQLite/U2-B、J2封存/147冻结、P8六 partial/八阻断及各自发行门槛保持。24×6000、14字段精确类型值、58工具及32K/2048/1024属于来源冻结验证条件,不设置成 Workbench 本地限制。来源方在途方案不等于已共享接口变更,本轮不启用绑定或改动程序。
|
||||
|
||||
本轮独立证据根 `.qa/coordination-readonly-2026-11-11-rs/`:九个清单成员、157611 字节,包含五份来源捕获及本轮登记工具/元数据;[登记](../../.qa/coordination-readonly-2026-11-11-rs/registration.json)、[清单](../../.qa/coordination-readonly-2026-11-11-rs/MANIFEST.json),清单 SHA256 `92cae3f275b9d8790c086b528637d83bbcc7262c05c18e26e84ba6084486e839`,逐件重读核验通过。旧证据不追加;本轮只有文档登记与限定本地提交,不推送、发布或改变当前开发顺序,零模型/工程/DELETE请求、零产品测试/来源脚本执行,不委派或跨线程回信。
|
||||
@@ -0,0 +1,503 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"kind": "workbench-readonly-engineering-source-index",
|
||||
"generatedAt": "2026-11-11T14:48:16.130Z",
|
||||
"sourceName": "agent-workbench/ad/node/6be88bee-1995-424a-b15a-2e0e54207d1a/2ff20f93-7a85-4117-9615-b71c976624f8",
|
||||
"ids": {
|
||||
"projectId": "674354be-ffb2-4f2f-b2ce-8b2c4960b008",
|
||||
"sessionId": "af80ce72-1289-4323-8be2-fafbaa210e01",
|
||||
"taskId": "6be88bee-1995-424a-b15a-2e0e54207d1a",
|
||||
"runId": "2ff20f93-7a85-4117-9615-b71c976624f8"
|
||||
},
|
||||
"originalVersion": "1.25.8",
|
||||
"originalDriverCommit": "ed78f02c032f2968e326fb8f696eba1281e0785b",
|
||||
"purpose": "Read-only acceptance of an existing independent Node task. No incoming restricted delegation or new M4 task is represented.",
|
||||
"goalPath": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\goal.txt",
|
||||
"predeclaredPlanPath": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\coordination-readonly-2026-11-11-engineering-ad-r2\\original-predeclared-plan.md",
|
||||
"originalTaskContract": {
|
||||
"schemaVersion": 1,
|
||||
"goal": "请独立修复这个Node费用报表工程,使原九项测试和build通过,并完成本地Git交付。只修改src/amount.mjs、src/summary.mjs、src/report.mjs;按原测试理解金额、币种汇总和报表要求,不修改原测试、package配置、构建脚本、README或owner文件,保留原有暂存/未暂存/未跟踪状态。请自行读取、修改、执行检查并提交,不要只解释方案或口头宣布完成。当前为完全访问权限,可以使用工作台工程工具或终端;不需要等待独立审查,不推送远端。",
|
||||
"kind": "general",
|
||||
"checks": "unspecified"
|
||||
},
|
||||
"originalStructuredAcceptance": [],
|
||||
"originalReportArtifact": {
|
||||
"id": "21a5483c-77f2-445b-838c-7a0b77d2d897",
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\coordination-readonly-2026-11-11-engineering-ad-r2\\21a5483c-77f2-445b-838c-7a0b77d2d897.json",
|
||||
"sha256": "e7a67a742bf4d0a0252a3cb92375ae8c23e40949d1279880fc617ba2c20e80a7",
|
||||
"bytes": 61179,
|
||||
"status": "in_progress"
|
||||
},
|
||||
"reportMetadata": [
|
||||
{
|
||||
"id": "21a5483c-77f2-445b-838c-7a0b77d2d897",
|
||||
"sessionId": "af80ce72-1289-4323-8be2-fafbaa210e01",
|
||||
"projectId": "674354be-ffb2-4f2f-b2ce-8b2c4960b008",
|
||||
"taskId": "6be88bee-1995-424a-b15a-2e0e54207d1a",
|
||||
"runId": "2ff20f93-7a85-4117-9615-b71c976624f8",
|
||||
"kind": "delivery-report",
|
||||
"mediaType": "application/json",
|
||||
"sha256": "e7a67a742bf4d0a0252a3cb92375ae8c23e40949d1279880fc617ba2c20e80a7",
|
||||
"bytes": 61179,
|
||||
"relative": "e7/e7a67a742bf4d0a0252a3cb92375ae8c23e40949d1279880fc617ba2c20e80a7.blob",
|
||||
"createdAt": "2026-11-06T04:52:58.386Z",
|
||||
"status": "stored",
|
||||
"sharedPath": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\coordination-readonly-2026-11-11-engineering-ad-r2\\21a5483c-77f2-445b-838c-7a0b77d2d897.json",
|
||||
"originalPath": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench\\artifacts\\e7\\e7a67a742bf4d0a0252a3cb92375ae8c23e40949d1279880fc617ba2c20e80a7.blob",
|
||||
"copiedBytesExact": true
|
||||
},
|
||||
{
|
||||
"id": "2de05f2f-4c49-4fbc-83f4-dab28c0b02d2",
|
||||
"sessionId": "af80ce72-1289-4323-8be2-fafbaa210e01",
|
||||
"projectId": "674354be-ffb2-4f2f-b2ce-8b2c4960b008",
|
||||
"taskId": "6be88bee-1995-424a-b15a-2e0e54207d1a",
|
||||
"runId": "2ff20f93-7a85-4117-9615-b71c976624f8",
|
||||
"kind": "delivery-report-markdown",
|
||||
"mediaType": "text/markdown",
|
||||
"sha256": "bdb282fc27b8826810bf64511e6b0d4304c3b905de17194ea79e966cbc47780a",
|
||||
"bytes": 9300,
|
||||
"relative": "bd/bdb282fc27b8826810bf64511e6b0d4304c3b905de17194ea79e966cbc47780a.blob",
|
||||
"createdAt": "2026-11-06T04:52:58.389Z",
|
||||
"status": "stored",
|
||||
"sharedPath": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\coordination-readonly-2026-11-11-engineering-ad-r2\\2de05f2f-4c49-4fbc-83f4-dab28c0b02d2.txt",
|
||||
"originalPath": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench\\artifacts\\bd\\bdb282fc27b8826810bf64511e6b0d4304c3b905de17194ea79e966cbc47780a.blob",
|
||||
"copiedBytesExact": true
|
||||
}
|
||||
],
|
||||
"checkLogMetadata": [],
|
||||
"engineering": {
|
||||
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\expense-project",
|
||||
"initialHead": "3a3f545337ac8eee849c2b1b224727019f2824a9",
|
||||
"commit": "88bb25cbf37739359c991dddac26e296a4d29d18",
|
||||
"committedPaths": [
|
||||
"src/amount.mjs",
|
||||
"src/report.mjs",
|
||||
"src/summary.mjs"
|
||||
],
|
||||
"committedCurrent": true,
|
||||
"actualIndependentChecks": {
|
||||
"test": {
|
||||
"status": 0,
|
||||
"signal": null,
|
||||
"error": null,
|
||||
"stats": {
|
||||
"tests": 9,
|
||||
"pass": 9,
|
||||
"fail": 0,
|
||||
"cancelled": 0,
|
||||
"skipped": 0,
|
||||
"todo": 0
|
||||
},
|
||||
"stdoutSha256": "c3bbf67e62ae04805face4ea745574b8d64a243ac421553271950215c9ba140c",
|
||||
"stderrSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
},
|
||||
"build": {
|
||||
"status": 0,
|
||||
"signal": null,
|
||||
"error": null,
|
||||
"stats": null,
|
||||
"stdoutSha256": "9307ee4e4f72392dd2f29f26c8e9fd3c8e55353bbe21f93627f260f108be9c2c",
|
||||
"stderrSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
}
|
||||
},
|
||||
"ownerBytesAndIndexPreserved": true,
|
||||
"supervisorContinuations": 0
|
||||
},
|
||||
"originalOutcome": {
|
||||
"independentEngineeringSuccess": true,
|
||||
"sessionStatus": "completed",
|
||||
"taskStatus": "completed",
|
||||
"runStatus": "completed",
|
||||
"terminalGitObserved": true,
|
||||
"nativeDeliveryRows": 0,
|
||||
"internalDeliveryReportStatus": "in_progress",
|
||||
"taskVerification": {
|
||||
"status": "passed",
|
||||
"checkJobId": "8b536d95-4a02-4806-af8f-f5955984c17f",
|
||||
"fingerprint": "150123dd461a2e7ffdda2531208de32b5b25ad6999d93ec388b1783bd05b9cea",
|
||||
"sourceFingerprint": "312aacaf54d8c95183fb3bfce09090a9be0f565c1ff4a452abc20f65895dcbcd",
|
||||
"checkedAt": "2026-11-06T04:52:41.161Z"
|
||||
},
|
||||
"productStopReason": null,
|
||||
"failure": null,
|
||||
"usage": {
|
||||
"requests": 11,
|
||||
"inputTokens": 171841,
|
||||
"outputTokens": 8713,
|
||||
"totalTokens": 180554,
|
||||
"unknownRequests": 0,
|
||||
"estimatedUnknownTokens": 0
|
||||
},
|
||||
"missing": [
|
||||
{
|
||||
"artifactId": "841b3dd5-e766-42f3-b372-0afdb7e77c82",
|
||||
"reason": "Check log descriptor absent from the sealed session artifact preview; independent-final logs remain in the read scope."
|
||||
},
|
||||
{
|
||||
"artifactId": "e96c4509-32ba-419b-8248-59cc5bb23342",
|
||||
"reason": "Check log descriptor absent from the sealed session artifact preview; independent-final logs remain in the read scope."
|
||||
}
|
||||
],
|
||||
"batchNetworkAndClaudeErrors": "Preserved separately in QA-LIVE-VALIDATION-AD.md; the Node subset does not replace the whole-batch outcome."
|
||||
},
|
||||
"readiness": {
|
||||
"originalTaskAndRunAvailable": true,
|
||||
"originalReportsCopied": true,
|
||||
"nativeEngineeringEvidenceV2Envelope": false,
|
||||
"restrictedDelegationReceiver": "not_established_by_this_source",
|
||||
"originalKeyLookup": "not_established_by_this_source",
|
||||
"currentM4Engineering": "not_dispatched_at_source_index_creation",
|
||||
"allConsumerContractFieldsSatisfied": false,
|
||||
"gaps": [
|
||||
"No producer-native mnemo.engineering-evidence.v2 envelope or receipt is claimed.",
|
||||
"The original task structured acceptance array is empty; the raw goal, predeclared plan and original tests carry its conditions.",
|
||||
"Original terminal Git success and internal in_progress report remain distinct."
|
||||
]
|
||||
},
|
||||
"readScope": {
|
||||
"mode": "read-only-existing-evidence",
|
||||
"allowedFiles": [
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\goal.txt",
|
||||
"bytes": 567,
|
||||
"sha256": "c8dd74010d05bbe7fec2faea7a58a2f8e681c05f05888612a3ab6cd4460b59a5",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/goal.txt"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\contract.json",
|
||||
"bytes": 2965,
|
||||
"sha256": "0a0acf7eae12b436527f119b14aa0c7c7a6bde320e8c315b248b30c20a2025a4",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/contract.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\source-before.json",
|
||||
"bytes": 1203,
|
||||
"sha256": "c134c4b3d20eb04a8690af68f9c503ec4f11bddaac6509dcc71cef2b5c0f3595",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/source-before.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\result.json",
|
||||
"bytes": 15501,
|
||||
"sha256": "086ff018544895884f94ff41d5bbf69e6bf7c07d0b09aef60e04a6f6cba80e64",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/result.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\session.json",
|
||||
"bytes": 774765,
|
||||
"sha256": "90bf4e9c5182fe9dd35b5c890b69945bb5dda8ded2e264b26d93bf06ceee318d",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/session.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\export.json",
|
||||
"bytes": 839293,
|
||||
"sha256": "921e3071fb095993a2719df05453d9ac4e3a5d891e949cbccf0280074a3348a6",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/export.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\attempts.json",
|
||||
"bytes": 41598,
|
||||
"sha256": "8324f8f606b893e15aeb3ed2cca2ce71cbb24945bb412d12b0625130b19b9d60",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/attempts.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\git-log.txt",
|
||||
"bytes": 524,
|
||||
"sha256": "9b13b3b2c56e2141ee63525deedd580d18ab100f0f528a249cabbfa4c2a0ce5e",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/git-log.txt"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\git-status.txt",
|
||||
"bytes": 65,
|
||||
"sha256": "07eb775fd33c1ab9ec904570086bbfeef138795cf6811d5ceacc3b4ba211a7d7",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/git-status.txt"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\baseline\\test.txt",
|
||||
"bytes": 10429,
|
||||
"sha256": "1a1024812314619dca7b682e9848370b9cb8037dc949066c4fedbaca67fc29f9",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/baseline/test.txt"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\baseline\\build.txt",
|
||||
"bytes": 36,
|
||||
"sha256": "9307ee4e4f72392dd2f29f26c8e9fd3c8e55353bbe21f93627f260f108be9c2c",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/baseline/build.txt"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\baseline\\result.json",
|
||||
"bytes": 651,
|
||||
"sha256": "abd44c095636f7844af9e97c85c2d9d24f8b667033dfd5bdc4c153803339c93c",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/baseline/result.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\independent-final\\test.txt",
|
||||
"bytes": 1807,
|
||||
"sha256": "c3bbf67e62ae04805face4ea745574b8d64a243ac421553271950215c9ba140c",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/independent-final/test.txt"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\independent-final\\build.txt",
|
||||
"bytes": 36,
|
||||
"sha256": "9307ee4e4f72392dd2f29f26c8e9fd3c8e55353bbe21f93627f260f108be9c2c",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/independent-final/build.txt"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\independent-final\\result.json",
|
||||
"bytes": 651,
|
||||
"sha256": "e8375595788c7ddf6d9c54df07190be07e2f1c6a7df318bdadfffd4a09afef48",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/independent-final/result.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\expense-project\\package.json",
|
||||
"bytes": 188,
|
||||
"sha256": "6ad63026b2fe9de7513a4f6704c372a86108ee3d018a1f04883dc0a409ea4c24",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/expense-project/package.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\expense-project\\test\\expense.test.mjs",
|
||||
"bytes": 2584,
|
||||
"sha256": "0bcda551dbaeafad9a6e3df8024492ab67d7c09956f4676c393b78249ab44839",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/expense-project/test/expense.test.mjs"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\expense-project\\scripts\\build.mjs",
|
||||
"bytes": 475,
|
||||
"sha256": "7279bc7c707c0f3ccbae63b14f0abed248ad3755be14f9e37128124ae9267526",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/expense-project/scripts/build.mjs"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\expense-project\\src\\amount.mjs",
|
||||
"bytes": 1933,
|
||||
"sha256": "0971180b3bc02b02780fe15d2531164f78020a9184df462a1eac2e4bac7e2d26",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/expense-project/src/amount.mjs"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\expense-project\\src\\report.mjs",
|
||||
"bytes": 1006,
|
||||
"sha256": "4162d18734b176edc39ebcb05de9324ef28cedf8b31e2887fba5bc5c53d8cb7a",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/expense-project/src/report.mjs"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\expense-project\\src\\summary.mjs",
|
||||
"bytes": 1354,
|
||||
"sha256": "c9800f24209adf00a9974b4f32c83b636be066529455bc87e6529ebdf03b408a",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/expense-project/src/summary.mjs"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\d5da6c54-4537-4203-9077-65dcd10edb8b-request.json",
|
||||
"bytes": 22361,
|
||||
"sha256": "8fcc91ffe2fd161412903f5c536f05824d3dee90de209d354548453999a64c16",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/d5da6c54-4537-4203-9077-65dcd10edb8b-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\d5da6c54-4537-4203-9077-65dcd10edb8b-response.sse",
|
||||
"bytes": 8526,
|
||||
"sha256": "34f781d11363b97babf64cc6df9ab9c25649f3b7763f5d2b638864e84836a8a2",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/d5da6c54-4537-4203-9077-65dcd10edb8b-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\e4d9710f-3007-499c-8678-4f1a67d07068-request.json",
|
||||
"bytes": 28655,
|
||||
"sha256": "ebaa79474016a1dd0a0be205da079354ba9bc09c1c9891f84edd7ea9c24f46a9",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/e4d9710f-3007-499c-8678-4f1a67d07068-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\e4d9710f-3007-499c-8678-4f1a67d07068-response.sse",
|
||||
"bytes": 20159,
|
||||
"sha256": "7309b290b096eaf9c3598eb52d63bb7f5f793a59a0ce9c36ce8d2a429ce86bd6",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/e4d9710f-3007-499c-8678-4f1a67d07068-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\c21c8b76-ee68-4b33-9ca3-6555014b2b18-request.json",
|
||||
"bytes": 32752,
|
||||
"sha256": "21e3f65a8f214e2deccdfbc3f3387487f6d789f8ca0974451aad5b05e18767da",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/c21c8b76-ee68-4b33-9ca3-6555014b2b18-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\c21c8b76-ee68-4b33-9ca3-6555014b2b18-response.sse",
|
||||
"bytes": 1183378,
|
||||
"sha256": "145b207478e2e0169bba2fd4282788b675554993e2f3b59436efeb829e5259a5",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/c21c8b76-ee68-4b33-9ca3-6555014b2b18-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\597a345a-4b39-4f38-a03c-61188ef2b426-request.json",
|
||||
"bytes": 49167,
|
||||
"sha256": "2034a2ade9782568a8bc98d8fa2b2511583060483082cd78d92c9684821b5b1a",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/597a345a-4b39-4f38-a03c-61188ef2b426-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\597a345a-4b39-4f38-a03c-61188ef2b426-response.sse",
|
||||
"bytes": 484422,
|
||||
"sha256": "578b3ccdde9276432362a6174c44d1c4d19e3ef9dd86dc827b113555e96b0edd",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/597a345a-4b39-4f38-a03c-61188ef2b426-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\5dfd0750-fc44-41d7-b82f-5e1e31a56d4b-request.json",
|
||||
"bytes": 59863,
|
||||
"sha256": "6fade8726b0d7ff235fddb6ebd00677e63744a9c33675138d0773b0f5a18145c",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/5dfd0750-fc44-41d7-b82f-5e1e31a56d4b-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\5dfd0750-fc44-41d7-b82f-5e1e31a56d4b-response.sse",
|
||||
"bytes": 15614,
|
||||
"sha256": "3a709fce2e840cac4fe800c5cb012fda5ac6685192ca14ce7bd9b659ae530414",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/5dfd0750-fc44-41d7-b82f-5e1e31a56d4b-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\b1e39d27-4eee-4664-b3e2-25da202a4b05-request.json",
|
||||
"bytes": 64525,
|
||||
"sha256": "91bcca633cc66c676e34578b90d929d17d0663f01d6527f8f4d6ef4009d1e321",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/b1e39d27-4eee-4664-b3e2-25da202a4b05-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\b1e39d27-4eee-4664-b3e2-25da202a4b05-response.sse",
|
||||
"bytes": 18404,
|
||||
"sha256": "67fcef56b2cd52095382c62f82f5da8b0c6567b9bc753edb1a271fbaecac2264",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/b1e39d27-4eee-4664-b3e2-25da202a4b05-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\fab4161c-0390-4f8e-8fb8-6ac3dbe5d3b8-request.json",
|
||||
"bytes": 66818,
|
||||
"sha256": "da66bd011c2e6bd7d6b26b271841e9b08029b78588c457d5e6311d2d0729ac9a",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/fab4161c-0390-4f8e-8fb8-6ac3dbe5d3b8-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\fab4161c-0390-4f8e-8fb8-6ac3dbe5d3b8-response.sse",
|
||||
"bytes": 67990,
|
||||
"sha256": "b72fe7b592f1a38a6b6508d403388c2190d85bd89ea7d78bac05e80c579105b3",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/fab4161c-0390-4f8e-8fb8-6ac3dbe5d3b8-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\f1134982-11e9-476f-9f85-c85640acd501-request.json",
|
||||
"bytes": 68155,
|
||||
"sha256": "1c09de1e3a594da2080b728498e91793df2bcc63eecac309a8a1e83526727a0b",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/f1134982-11e9-476f-9f85-c85640acd501-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\f1134982-11e9-476f-9f85-c85640acd501-response.sse",
|
||||
"bytes": 144597,
|
||||
"sha256": "eafa8e79dc8ea65b8792c4055e71807985cde8d3456201493a24667cba8baf98",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/f1134982-11e9-476f-9f85-c85640acd501-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\7b49e6a2-0e9c-4390-a8d7-1cbe4b2df083-request.json",
|
||||
"bytes": 71733,
|
||||
"sha256": "ca0714d6b248a38c6fb4b9eeffe7c7b87430cf5a4ed67250255967ce8b10de6f",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/7b49e6a2-0e9c-4390-a8d7-1cbe4b2df083-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\7b49e6a2-0e9c-4390-a8d7-1cbe4b2df083-response.sse",
|
||||
"bytes": 242957,
|
||||
"sha256": "adf70113ae9933bcdfe9bafaf35a8202694a47b19fd3742519a8fe1307c52e76",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/7b49e6a2-0e9c-4390-a8d7-1cbe4b2df083-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\68bad6fe-e9dc-4543-ad22-618800c07355-request.json",
|
||||
"bytes": 77812,
|
||||
"sha256": "bdecf73374975e083eacb022099aef8771503f50e251a01a10fd33f618fe0a8e",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/68bad6fe-e9dc-4543-ad22-618800c07355-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\68bad6fe-e9dc-4543-ad22-618800c07355-response.sse",
|
||||
"bytes": 126196,
|
||||
"sha256": "c59ed4d9e469c8f1283ec706c5ae8e80b47675cdedb58abce0e771c0fcf3f886",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/68bad6fe-e9dc-4543-ad22-618800c07355-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\edbb488b-1524-4fd3-8196-281abd17840c-request.json",
|
||||
"bytes": 80815,
|
||||
"sha256": "75dd798b522033cf410e2f07be95a7d973cd36869a586fd52443a73ee0752189",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/edbb488b-1524-4fd3-8196-281abd17840c-request.json"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\node\\captures\\edbb488b-1524-4fd3-8196-281abd17840c-response.sse",
|
||||
"bytes": 296931,
|
||||
"sha256": "aa68a4e16b7cae4862ceccf771272d22c9a5f127c158283ba50ccbfb266fe757",
|
||||
"origin": "original-ad-seal",
|
||||
"originalRelative": "node/captures/edbb488b-1524-4fd3-8196-281abd17840c-response.sse"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\coordination-readonly-2026-11-11-engineering-ad-r2\\original-predeclared-plan.md",
|
||||
"bytes": 4745,
|
||||
"sha256": "815ceae01ac5224d6ab3f3b35e7512024f0d9f50c5743173797be7ae6caffbe1",
|
||||
"origin": "original-git-blob",
|
||||
"commit": "ed78f02c032f2968e326fb8f696eba1281e0785b",
|
||||
"repositoryPath": "docs/LIVE-VALIDATION-AD-PLAN.md"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\coordination-readonly-2026-11-11-engineering-ad-r2\\21a5483c-77f2-445b-838c-7a0b77d2d897.json",
|
||||
"bytes": 61179,
|
||||
"sha256": "e7a67a742bf4d0a0252a3cb92375ae8c23e40949d1279880fc617ba2c20e80a7",
|
||||
"origin": "byte-exact-copy-of-original-owned-artifact",
|
||||
"artifactId": "21a5483c-77f2-445b-838c-7a0b77d2d897",
|
||||
"kind": "delivery-report"
|
||||
},
|
||||
{
|
||||
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\coordination-readonly-2026-11-11-engineering-ad-r2\\2de05f2f-4c49-4fbc-83f4-dab28c0b02d2.txt",
|
||||
"bytes": 9300,
|
||||
"sha256": "bdb282fc27b8826810bf64511e6b0d4304c3b905de17194ea79e966cbc47780a",
|
||||
"origin": "byte-exact-copy-of-original-owned-artifact",
|
||||
"artifactId": "2de05f2f-4c49-4fbc-83f4-dab28c0b02d2",
|
||||
"kind": "delivery-report-markdown"
|
||||
}
|
||||
],
|
||||
"originalManifestPath": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\live-validation-ad\\FINAL-MANIFEST.json",
|
||||
"originalManifestSha256": "1c21f6b82bd88c1860b77b783d2556a449a76f0b5e21751ea710ab40f9419440",
|
||||
"excludes": [
|
||||
"Any credential value or credential store",
|
||||
"Whole formal data directory or SQLite database",
|
||||
"Other user projects or unrelated sessions",
|
||||
"Replay, retry, execution, write or native-v2 certification"
|
||||
]
|
||||
},
|
||||
"actions": {
|
||||
"realModelRequests": 0,
|
||||
"engineeringDispatches": 0,
|
||||
"gitWrites": 0,
|
||||
"originalArtifactsChanged": false,
|
||||
"keysReadOrResaved": false,
|
||||
"messagesSentToOtherThreads": 0
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,11 @@
|
||||
# Agent Workbench 已封存 Node 工程的只读验收入口
|
||||
|
||||
本入口响应 2026-11-11 的 W2 既有工程产物核对请求,只提供已结束的 AD 原 Task/Run。它不是新的 M4 实验,也不是 Mnemo 受限入站委托;没有启动、重发模型/工程、执行批准或 Git 写入,没有向源线程发送消息。
|
||||
|
||||
显式元数据:[WORKBENCH-AD-NODE-READONLY-SOURCE-2026-11-11.json](WORKBENCH-AD-NODE-READONLY-SOURCE-2026-11-11.json)。仅允许读取该文件 `readScope.allowedFiles` 中列出的已有封存文件及原字节副本;每项包含绝对路径、bytes、SHA256 和原来源。原封存根 `.qa/live-validation-ad/` 及正式数据不改写。新副本/索引在 `.qa/coordination-readonly-2026-11-11-engineering-ad-r2/`,首组装失败目录保持,不授予读取整个正式数据目录或凭据的范围。
|
||||
|
||||
稳定来源名为 `agent-workbench/ad/node/6be88bee-1995-424a-b15a-2e0e54207d1a/2ff20f93-7a85-4117-9615-b71c976624f8`。原 Project `674354be-ffb2-4f2f-b2ce-8b2c4960b008`、Session `af80ce72-1289-4323-8be2-fafbaa210e01`,正式来源是当时的 1.25.8。原需求、调用前 Git 版本的计划、原九项测试、限定三源码、owner/index 保护、实际检查日志、模型请求原文和 Git 记录均由元数据指向原件。
|
||||
|
||||
原 Node 独立验收成功:11 请求、180554 报告 tokens、0 unknown,原九测试 9/9、build 退出 0,实际限定 Git 提交 `88bb25cbf37739359c991dddac26e296a4d29d18`。这不能覆盖 AD 全批网络/Claude 错误,也不能代签 1.25.13、新 M4、通用智能或 Purge 收益。
|
||||
|
||||
原 Task 的结构化 `acceptance` 是空数组,原需求/预先计划及测试仍存在。原终端 Git 交付与内置报告 `in_progress`、内置 deliveries 空列表必须分别保留,不能把复制报告重新签成 completed。原检查结果引用的两个日志 ID 不在当时的 session 工件预览中,按缺少原描述符登记,不推断原物理日志丢失;独立最终检查的封存日志仍可读取。未生成或声称 `mnemo.engineering-evidence.v2` 原生包、受限接收、原键查询或消费者完整契约通过。消费者应据原字段登记实际缺件/unknown,不以普通会话或本入口替代受限授权。
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,980 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"checkedAt": "2026-11-07T12:35:16.944Z",
|
||||
"milestone": "M2 I09/I11 source acceptance",
|
||||
"baseline": "bdfeb29c85b3231bfd7bc24b8b5f49d40a3beebf",
|
||||
"sourceCommits": {
|
||||
"sources": "71f1805561e6082011ed4f0c70bb38f4546413c5",
|
||||
"core": "9e1feeebb3eca5c82bfa9efb4f32e68a73628df4",
|
||||
"ui": "ee97aaae7389e0aff7f89eced8d06ecce7d4f982"
|
||||
},
|
||||
"productVersion": "1.25.12",
|
||||
"node": {
|
||||
"executable": "D:\\Ai-project\\REA逆向\\tools\\node-v24.22.1-win-x64\\node.exe",
|
||||
"version": "v24.22.1"
|
||||
},
|
||||
"sourceAcceptanceComplete": true,
|
||||
"sourceReview": {
|
||||
"schemaVersion": 1,
|
||||
"baseline": "bdfeb29c85b3231bfd7bc24b8b5f49d40a3beebf",
|
||||
"sources": [
|
||||
{
|
||||
"id": "openai-conversation-state",
|
||||
"provider": "openai",
|
||||
"url": "https://developers.openai.com/api/docs/guides/conversation-state",
|
||||
"checkedAt": "2026-11-07",
|
||||
"file": "sources/openai-conversation-state.md",
|
||||
"sha256": "8a62162c2ef6ff7e94dfdad5ba06f417f5a6c00ea6521e6466d462ff2fcd9b96",
|
||||
"bytes": 24788,
|
||||
"captureSha256": "df7776f8ac54a84b55cbc0d42bc2b80192428921d96b2e272d21754e2f3d5f8b",
|
||||
"representation": "official-docs-mcp-markdown-utf8",
|
||||
"scope": "complete-captured-page",
|
||||
"reusedReadOnly": true
|
||||
},
|
||||
{
|
||||
"id": "openai-prompt-caching",
|
||||
"provider": "openai",
|
||||
"url": "https://developers.openai.com/api/docs/guides/prompt-caching",
|
||||
"checkedAt": "2026-11-07",
|
||||
"file": "sources/openai-prompt-caching.md",
|
||||
"sha256": "01e57dacc2e5c0fb881b2d1909649983e9e327a5d6c17a98bb1183ee00e9d334",
|
||||
"bytes": 56425,
|
||||
"captureSha256": "ed312ad7e34d4de7f5861317481ae82b1bc3b9a1ae78d525c916a0d94231d522",
|
||||
"representation": "official-docs-mcp-markdown-utf8",
|
||||
"scope": "complete-captured-page",
|
||||
"reusedReadOnly": true
|
||||
},
|
||||
{
|
||||
"id": "claude-build-with-claude__prompt-caching",
|
||||
"provider": "anthropic",
|
||||
"url": "https://platform.claude.com/docs/en/build-with-claude/prompt-caching",
|
||||
"checkedAt": "2026-11-06",
|
||||
"sha256": "ff35232788f3f7fbd0f9aae002351567a9ea8e68e5b749069d96dab52882747d",
|
||||
"file": "../official-model-registry/sources/claude-build-with-claude-prompt-caching.md",
|
||||
"originalSha256": "ff35232788f3f7fbd0f9aae002351567a9ea8e68e5b749069d96dab52882747d",
|
||||
"representation": "extracted-official-markdown",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 164465,
|
||||
"reusedReadOnly": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-create",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/api/create-chat-completion",
|
||||
"checkedAt": "2026-11-06",
|
||||
"sha256": "b43ca950bd04ae7f6f537b42e6cb729f543d942dcd848ec43ad4e9bfe059fe1f",
|
||||
"file": "../official-model-registry/sources/deepseek-create.body",
|
||||
"originalSha256": "b43ca950bd04ae7f6f537b42e6cb729f543d942dcd848ec43ad4e9bfe059fe1f",
|
||||
"representation": "received-html-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 129498,
|
||||
"reusedReadOnly": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-thinking",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/guides/thinking_mode",
|
||||
"checkedAt": "2026-11-06",
|
||||
"sha256": "42151b5f86827bff663a61693080a61cdbf99074dee94af756c9fce855baa7cd",
|
||||
"file": "../official-model-registry/sources/deepseek-thinking.body",
|
||||
"originalSha256": "42151b5f86827bff663a61693080a61cdbf99074dee94af756c9fce855baa7cd",
|
||||
"representation": "received-html-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 117778,
|
||||
"reusedReadOnly": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-cache",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/guides/kv_cache",
|
||||
"checkedAt": "2026-11-06",
|
||||
"sha256": "061c0c7055ceb96d0a522ae1829ff5f7126ff08f26af217643e371abb740a081",
|
||||
"file": "../official-model-registry/sources/deepseek-cache.body",
|
||||
"originalSha256": "061c0c7055ceb96d0a522ae1829ff5f7126ff08f26af217643e371abb740a081",
|
||||
"representation": "received-html-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 35850,
|
||||
"reusedReadOnly": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-response",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/api/create-response.md",
|
||||
"checkedAt": "2026-11-07",
|
||||
"sha256": "5e56e45d25db47f2b6966bbb0b2471d324035c33f0c8f81eaaf8fc444e6869da",
|
||||
"file": "../official-model-registry/sources/deepseek-response.md",
|
||||
"originalSha256": "5e56e45d25db47f2b6966bbb0b2471d324035c33f0c8f81eaaf8fc444e6869da",
|
||||
"representation": "received-document-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 33870,
|
||||
"reusedReadOnly": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-message",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/api/create-message.md",
|
||||
"checkedAt": "2026-11-07",
|
||||
"sha256": "46b31d10f429910bf93ea2166537fb803cd2f49f7c5b4ad2ee4f08257b2e17e8",
|
||||
"file": "../official-model-registry/sources/deepseek-message.md",
|
||||
"originalSha256": "46b31d10f429910bf93ea2166537fb803cd2f49f7c5b4ad2ee4f08257b2e17e8",
|
||||
"representation": "received-document-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 51134,
|
||||
"reusedReadOnly": true
|
||||
}
|
||||
],
|
||||
"newRealModelRequests": 0,
|
||||
"previousEvidenceRootsWritten": 0,
|
||||
"scope": "request-block-fidelity-provider-state-and-affinity-contracts-not-live-cache-benefit",
|
||||
"retainedEarlierRounds": [
|
||||
"source-1",
|
||||
"source-2",
|
||||
"source-3"
|
||||
],
|
||||
"sourceReview": [
|
||||
"Messages cache_control and TTL stay on their original system/user/tool blocks.",
|
||||
"OpenAI explicit cache breakpoints and independent retention controls are passed only when supplied.",
|
||||
"Responses complete output items, encrypted reasoning and phase must survive applicable manual history.",
|
||||
"DeepSeek assistant reasoning_content remains with every applicable historical assistant turn.",
|
||||
"Affinity and scope fingerprints are local continuity observations, never supplier cache-hit evidence."
|
||||
]
|
||||
},
|
||||
"inputIdentity": {
|
||||
"schemaVersion": 1,
|
||||
"checkedAt": "2026-11-07T12:24:22.125Z",
|
||||
"finalBuild": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-native-cache-state-2026-11-07\\build-5",
|
||||
"app": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-native-cache-state-2026-11-07\\candidate-app-5",
|
||||
"inputFiles": 880,
|
||||
"compiledFiles": 29,
|
||||
"runtimeFiles": 74,
|
||||
"inputsExact": true,
|
||||
"compiledCopiesExact": true,
|
||||
"runtimeCopiesExact": true,
|
||||
"productFiles": 300,
|
||||
"productMatchesLatestBackendCandidate": true,
|
||||
"backendFiles": 178,
|
||||
"backendMatchesFirstCandidate": true,
|
||||
"testInputs": 213,
|
||||
"unchangedTestInputs": 211,
|
||||
"correctedTestFiles": [
|
||||
"tests/model-center.test.mjs",
|
||||
"tests/model-routing-permission.test.mjs"
|
||||
],
|
||||
"coverage": {
|
||||
"firstCandidate": {
|
||||
"build": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-native-cache-state-2026-11-07\\build-1",
|
||||
"app": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-native-cache-state-2026-11-07\\candidate-app-1",
|
||||
"node": "D:\\Ai-project\\REA逆向\\tools\\node-v24.22.1-win-x64\\node.exe",
|
||||
"files": [
|
||||
"tests/native-cache-state.test.mjs",
|
||||
"tests/model-protocols.test.mjs",
|
||||
"tests/model-stream.test.mjs",
|
||||
"tests/model-parameter-fidelity.test.mjs",
|
||||
"tests/model-request-contract.test.mjs",
|
||||
"tests/model-center.test.mjs",
|
||||
"tests/model-routing-capabilities.test.mjs",
|
||||
"tests/model-routing-permission.test.mjs",
|
||||
"tests/task-context.test.mjs",
|
||||
"tests/model-identity.test.mjs",
|
||||
"tests/model-capacity.test.mjs"
|
||||
],
|
||||
"startedAt": "2026-11-07T12:06:32.105Z",
|
||||
"finishedAt": "2026-11-07T12:09:47.462Z",
|
||||
"durationMs": 195356,
|
||||
"code": 1,
|
||||
"signal": null,
|
||||
"totals": {
|
||||
"tests": 129,
|
||||
"pass": 127,
|
||||
"fail": 2,
|
||||
"cancelled": 0,
|
||||
"skipped": 0,
|
||||
"todo": 0,
|
||||
"duration_ms": 195312.8855
|
||||
},
|
||||
"success": false
|
||||
},
|
||||
"correctedFiles": {
|
||||
"build": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-native-cache-state-2026-11-07\\build-3",
|
||||
"app": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-native-cache-state-2026-11-07\\candidate-app-3",
|
||||
"node": "D:\\Ai-project\\REA逆向\\tools\\node-v24.22.1-win-x64\\node.exe",
|
||||
"files": [
|
||||
"tests/model-center.test.mjs",
|
||||
"tests/model-routing-permission.test.mjs"
|
||||
],
|
||||
"startedAt": "2026-11-07T12:17:59.645Z",
|
||||
"finishedAt": "2026-11-07T12:18:49.887Z",
|
||||
"durationMs": 50242,
|
||||
"code": 0,
|
||||
"signal": null,
|
||||
"totals": {
|
||||
"tests": 26,
|
||||
"pass": 26,
|
||||
"fail": 0,
|
||||
"cancelled": 0,
|
||||
"skipped": 0,
|
||||
"todo": 0,
|
||||
"duration_ms": 50205.9558
|
||||
},
|
||||
"success": true
|
||||
},
|
||||
"unchangedPassedTestsReused": 103,
|
||||
"compositeCoveredTests": 129,
|
||||
"singleFinalFullSuiteClaim": false
|
||||
},
|
||||
"sourceManifestSha256": "ced3703c4afb174c5b939ffb3dd8572f22ff8e5f3fb8489f627f45880315b799",
|
||||
"officialSources": [
|
||||
{
|
||||
"id": "openai-conversation-state",
|
||||
"provider": "openai",
|
||||
"url": "https://developers.openai.com/api/docs/guides/conversation-state",
|
||||
"checkedAt": "2026-11-07",
|
||||
"file": "sources/openai-conversation-state.md",
|
||||
"sha256": "8a62162c2ef6ff7e94dfdad5ba06f417f5a6c00ea6521e6466d462ff2fcd9b96",
|
||||
"bytes": 24788,
|
||||
"captureSha256": "df7776f8ac54a84b55cbc0d42bc2b80192428921d96b2e272d21754e2f3d5f8b",
|
||||
"representation": "official-docs-mcp-markdown-utf8",
|
||||
"scope": "complete-captured-page",
|
||||
"reusedReadOnly": true,
|
||||
"repositoryPath": "tests/fixtures/native-cache-state/sources/openai-conversation-state.md",
|
||||
"bytesExact": true,
|
||||
"committedBytesExact": true
|
||||
},
|
||||
{
|
||||
"id": "openai-prompt-caching",
|
||||
"provider": "openai",
|
||||
"url": "https://developers.openai.com/api/docs/guides/prompt-caching",
|
||||
"checkedAt": "2026-11-07",
|
||||
"file": "sources/openai-prompt-caching.md",
|
||||
"sha256": "01e57dacc2e5c0fb881b2d1909649983e9e327a5d6c17a98bb1183ee00e9d334",
|
||||
"bytes": 56425,
|
||||
"captureSha256": "ed312ad7e34d4de7f5861317481ae82b1bc3b9a1ae78d525c916a0d94231d522",
|
||||
"representation": "official-docs-mcp-markdown-utf8",
|
||||
"scope": "complete-captured-page",
|
||||
"reusedReadOnly": true,
|
||||
"repositoryPath": "tests/fixtures/native-cache-state/sources/openai-prompt-caching.md",
|
||||
"bytesExact": true,
|
||||
"committedBytesExact": true
|
||||
},
|
||||
{
|
||||
"id": "claude-build-with-claude__prompt-caching",
|
||||
"provider": "anthropic",
|
||||
"url": "https://platform.claude.com/docs/en/build-with-claude/prompt-caching",
|
||||
"checkedAt": "2026-11-06",
|
||||
"sha256": "ff35232788f3f7fbd0f9aae002351567a9ea8e68e5b749069d96dab52882747d",
|
||||
"file": "../official-model-registry/sources/claude-build-with-claude-prompt-caching.md",
|
||||
"originalSha256": "ff35232788f3f7fbd0f9aae002351567a9ea8e68e5b749069d96dab52882747d",
|
||||
"representation": "extracted-official-markdown",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 164465,
|
||||
"reusedReadOnly": true,
|
||||
"repositoryPath": "tests/fixtures/official-model-registry/sources/claude-build-with-claude-prompt-caching.md",
|
||||
"bytesExact": true,
|
||||
"committedBytesExact": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-create",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/api/create-chat-completion",
|
||||
"checkedAt": "2026-11-06",
|
||||
"sha256": "b43ca950bd04ae7f6f537b42e6cb729f543d942dcd848ec43ad4e9bfe059fe1f",
|
||||
"file": "../official-model-registry/sources/deepseek-create.body",
|
||||
"originalSha256": "b43ca950bd04ae7f6f537b42e6cb729f543d942dcd848ec43ad4e9bfe059fe1f",
|
||||
"representation": "received-html-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 129498,
|
||||
"reusedReadOnly": true,
|
||||
"repositoryPath": "tests/fixtures/official-model-registry/sources/deepseek-create.body",
|
||||
"bytesExact": true,
|
||||
"committedBytesExact": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-thinking",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/guides/thinking_mode",
|
||||
"checkedAt": "2026-11-06",
|
||||
"sha256": "42151b5f86827bff663a61693080a61cdbf99074dee94af756c9fce855baa7cd",
|
||||
"file": "../official-model-registry/sources/deepseek-thinking.body",
|
||||
"originalSha256": "42151b5f86827bff663a61693080a61cdbf99074dee94af756c9fce855baa7cd",
|
||||
"representation": "received-html-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 117778,
|
||||
"reusedReadOnly": true,
|
||||
"repositoryPath": "tests/fixtures/official-model-registry/sources/deepseek-thinking.body",
|
||||
"bytesExact": true,
|
||||
"committedBytesExact": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-cache",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/guides/kv_cache",
|
||||
"checkedAt": "2026-11-06",
|
||||
"sha256": "061c0c7055ceb96d0a522ae1829ff5f7126ff08f26af217643e371abb740a081",
|
||||
"file": "../official-model-registry/sources/deepseek-cache.body",
|
||||
"originalSha256": "061c0c7055ceb96d0a522ae1829ff5f7126ff08f26af217643e371abb740a081",
|
||||
"representation": "received-html-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 35850,
|
||||
"reusedReadOnly": true,
|
||||
"repositoryPath": "tests/fixtures/official-model-registry/sources/deepseek-cache.body",
|
||||
"bytesExact": true,
|
||||
"committedBytesExact": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-response",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/api/create-response.md",
|
||||
"checkedAt": "2026-11-07",
|
||||
"sha256": "5e56e45d25db47f2b6966bbb0b2471d324035c33f0c8f81eaaf8fc444e6869da",
|
||||
"file": "../official-model-registry/sources/deepseek-response.md",
|
||||
"originalSha256": "5e56e45d25db47f2b6966bbb0b2471d324035c33f0c8f81eaaf8fc444e6869da",
|
||||
"representation": "received-document-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 33870,
|
||||
"reusedReadOnly": true,
|
||||
"repositoryPath": "tests/fixtures/official-model-registry/sources/deepseek-response.md",
|
||||
"bytesExact": true,
|
||||
"committedBytesExact": true
|
||||
},
|
||||
{
|
||||
"id": "deepseek-message",
|
||||
"provider": "deepseek",
|
||||
"url": "https://api-docs.deepseek.com/api/create-message.md",
|
||||
"checkedAt": "2026-11-07",
|
||||
"sha256": "46b31d10f429910bf93ea2166537fb803cd2f49f7c5b4ad2ee4f08257b2e17e8",
|
||||
"file": "../official-model-registry/sources/deepseek-message.md",
|
||||
"originalSha256": "46b31d10f429910bf93ea2166537fb803cd2f49f7c5b4ad2ee4f08257b2e17e8",
|
||||
"representation": "received-document-bytes",
|
||||
"scope": "complete-captured-source",
|
||||
"bytes": 51134,
|
||||
"reusedReadOnly": true,
|
||||
"repositoryPath": "tests/fixtures/official-model-registry/sources/deepseek-message.md",
|
||||
"bytesExact": true,
|
||||
"committedBytesExact": true
|
||||
}
|
||||
],
|
||||
"realModelRequests": 0,
|
||||
"productionWrites": 0
|
||||
},
|
||||
"committedInputs": {
|
||||
"checkedAt": "2026-11-07T12:32:35.873Z",
|
||||
"base": "bdfeb29c85b3231bfd7bc24b8b5f49d40a3beebf",
|
||||
"head": "ee97aaae7389e0aff7f89eced8d06ecce7d4f982",
|
||||
"sourceCommits": [
|
||||
"71f1805561e6082011ed4f0c70bb38f4546413c5",
|
||||
"9e1feeebb3eca5c82bfa9efb4f32e68a73628df4",
|
||||
"ee97aaae7389e0aff7f89eced8d06ecce7d4f982"
|
||||
],
|
||||
"rows": [
|
||||
{
|
||||
"path": "scripts/verify-native-cache-state-ui.mjs",
|
||||
"bytes": 26680,
|
||||
"sha256": "5a40f1a284174e2d7e3394441e6f091d23fc75eb5c495fcd5353d54b9624596d",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/client/ModelCenter.css",
|
||||
"bytes": 18118,
|
||||
"sha256": "8e10c42d67393805ed2a3b64fa10303381d90d9e08eba8f6c2c0d5074a478fc1",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/client/ModelCenter.tsx",
|
||||
"bytes": 50355,
|
||||
"sha256": "7f01e7e59b50a009d6a32dfdb573e6a83637020649a4b4921a017afbe75c150e",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/client/ModelProviderStateDetails.css",
|
||||
"bytes": 1404,
|
||||
"sha256": "1c1f5c2c124b004b68d20affd4c9bad9594d0bc00ab63a0e569b922f49ce1417",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/client/ModelProviderStateDetails.tsx",
|
||||
"bytes": 3118,
|
||||
"sha256": "4172ab798b7afe9dbb2ecb50c6eb0e399b8fe9c7a759c3f78f7d0bbb2ae8e1f8",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/client/ModelRequestLedger.tsx",
|
||||
"bytes": 11836,
|
||||
"sha256": "e35dab71c350b758e2c0878f439665190bce6ff422ad91b395f89171e65bc626",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/client/ModelRoutingDetails.tsx",
|
||||
"bytes": 4801,
|
||||
"sha256": "a3b93157e305954076fa78e62a285a01eb58a307a8c6cf27a2338e89a653e338",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/client/RequestContextPanel.tsx",
|
||||
"bytes": 13414,
|
||||
"sha256": "b2c6152198eba74aa1c4d0cbfb29d9ca354b71bab2cf34309b2929beb2723e1e",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/client/types.ts",
|
||||
"bytes": 49501,
|
||||
"sha256": "b45fc5851ae0d8620f503cb00b338cbde9dc76c8e9d2b3b3a374bebb1c10e95e",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/core/context/request-plan.mjs",
|
||||
"bytes": 56816,
|
||||
"sha256": "3f56d3cc7c0757c62133a086f6b2743398a16be4cf87ecff1bce472fbadab6e1",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/core/models/model-affinity.mjs",
|
||||
"bytes": 3384,
|
||||
"sha256": "fd642ecfb4c8a2e085812a1a03f5fc63784103758d4482b842555e2dada7a717",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/core/models/model-hub.mjs",
|
||||
"bytes": 47089,
|
||||
"sha256": "e9c8fdc19e1076ef993c23ddf3d101c2a1a9a3bc2824c1733e6d0785a38f11cc",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/core/models/protocols.mjs",
|
||||
"bytes": 12242,
|
||||
"sha256": "2f071a2e0c8308d55fde1913ee4980b6a181f935cd4836951a196af7ab165ce2",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/shared/contracts/anthropic-messages.mjs",
|
||||
"bytes": 6732,
|
||||
"sha256": "7e245ba13e98fef31f5aef5f430e7b9691c46ca4c91ad1c90c1d4864f3a1afa9",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/shared/contracts/model-provider-state.mjs",
|
||||
"bytes": 3660,
|
||||
"sha256": "8de752e559e32cb6cab6820a54001affde3996c07fb9e1636a2a1a189012c471",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "src/shared/contracts/model-request.mjs",
|
||||
"bytes": 9247,
|
||||
"sha256": "b0a2b6cdb7939785eee61fb188e536a0886b8281ae9e9e403dc990ec12c2f0c8",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "tests/fixtures/native-cache-state/SOURCE.json",
|
||||
"bytes": 4883,
|
||||
"sha256": "ced3703c4afb174c5b939ffb3dd8572f22ff8e5f3fb8489f627f45880315b799",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "tests/fixtures/native-cache-state/sources/openai-conversation-state.md",
|
||||
"bytes": 24788,
|
||||
"sha256": "8a62162c2ef6ff7e94dfdad5ba06f417f5a6c00ea6521e6466d462ff2fcd9b96",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "tests/fixtures/native-cache-state/sources/openai-prompt-caching.md",
|
||||
"bytes": 56425,
|
||||
"sha256": "01e57dacc2e5c0fb881b2d1909649983e9e327a5d6c17a98bb1183ee00e9d334",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "tests/model-center.test.mjs",
|
||||
"bytes": 16904,
|
||||
"sha256": "c3aa558bdc79a9dc4181c3df85a598c94c3345081a12b7cca0a638cc9bb43af9",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "tests/model-routing-permission.test.mjs",
|
||||
"bytes": 21169,
|
||||
"sha256": "b9339274c946a33e60928c9aa90fce2c894e32b5dd176ea982da0b355a735fde",
|
||||
"committedWorkingFrozenExact": true
|
||||
},
|
||||
{
|
||||
"path": "tests/native-cache-state.test.mjs",
|
||||
"bytes": 29644,
|
||||
"sha256": "ce063484e154f072aa024dd20a49fb0c9ea5846ed8b78f45baa12fb40e93567c",
|
||||
"committedWorkingFrozenExact": true
|
||||
}
|
||||
],
|
||||
"allowedChangedInputs": true,
|
||||
"byteExact": true,
|
||||
"productWorktreeClean": true
|
||||
},
|
||||
"finalUI": {
|
||||
"checkedAt": "2026-11-07T12:32:35.904Z",
|
||||
"ends": [
|
||||
{
|
||||
"stage": "web-ui-5",
|
||||
"resultFile": "web-ui-5/web-d6b3ea8a-032b-4eef-a718-0efd492e3803/result.json",
|
||||
"success": true,
|
||||
"appRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-native-cache-state-2026-11-07\\candidate-app-5",
|
||||
"actualElectron": false,
|
||||
"hostUrl": "http://127.0.0.1:51514",
|
||||
"checks": [
|
||||
"Existing pools retain their original Run/disabled scope without saving new defaults or issuing model requests",
|
||||
"All four pool scopes save through the real form and survive reload; new pools start with session affinity and explicit selection persists",
|
||||
"The request ledger shows completed-response affinity and applicable native states separately from the retained original unknown state and cache usage",
|
||||
"The actual task inspector exposes the frozen state projection and upstream-owned context limits at 390px",
|
||||
"Changed provider scope reselects the next eligible pool member, retains old states and one original upstream 400, and fabricates no old metadata",
|
||||
"Manual model selection takes precedence, transmits its changed header and preserves the original 400 once without applying old opaque history or replaying tools",
|
||||
"State details and scope settings remain readable and usable in the three existing palettes without narrow-screen document overflow",
|
||||
"Candidate host restart retains saved scope, original pools, native cache settings, original metadata/cipher/snapshot and history; no tool or request is replayed"
|
||||
],
|
||||
"screenshots": [
|
||||
{
|
||||
"file": "existing-pools-1280.png",
|
||||
"sha256": "76421e822001fcd91efc5ba380723b8aa928a4ead63a2ae90f88ea68d2b68445",
|
||||
"method": "playwright-page"
|
||||
},
|
||||
{
|
||||
"file": "new-session-pool-1280.png",
|
||||
"sha256": "7026f014adc9301bfc94f76d36d891b1d64528accd7f39a8035168592efbb80d",
|
||||
"method": "playwright-page"
|
||||
},
|
||||
{
|
||||
"file": "task-pool-390.png",
|
||||
"sha256": "80c91f796599e7db989792a8ea93172b366f33973d1e46f2a2b827eeded643f7",
|
||||
"method": "playwright-page"
|
||||
},
|
||||
{
|
||||
"file": "continued-history-1280.png",
|
||||
"sha256": "36a122de9da3a7dea868cde6e20198f127c2044addd89cb18b946e4a517f6333",
|
||||
"method": "playwright-page"
|
||||
},
|
||||
{
|
||||
"file": "context-state-390.png",
|
||||
"sha256": "9212ea44c965134340d596528511ce4057131ccbf7524b3446fa852daf5097ec",
|
||||
"method": "playwright-page"
|
||||
},
|
||||
{
|
||||
"file": "changed-state-original-error-390.png",
|
||||
"sha256": "df74718897986f6f911ad0b4620048e97e48ca85650cbf4f525edb91b278f6a9",
|
||||
"method": "playwright-page"
|
||||
},
|
||||
{
|
||||
"file": "manual-selection-original-error-390.png",
|
||||
"sha256": "664827ca950d238a6512056d903c6f0b3b1a0ef61b39edc63fa70aa0fe69c61c",
|
||||
"method": "playwright-page"
|
||||
},
|
||||
{
|
||||
"file": "state-green-390.png",
|
||||
"sha256": "6c682a50caf511a41d2468ad955b8b91c38d34f1ed7a0d40419fe12d11f1d930",
|
||||
"method": "playwright-page"
|
||||
},
|
||||
{
|
||||
"file": "state-blue-390.png",
|
||||
"sha256": "d93ee1edecff973a47c059a1ae8a70f8b6762e426d2bc1add72149e874790570",
|
||||
"method": "playwright-page"
|
||||
},
|
||||
{
|
||||
"file": "state-purple-390.png",
|
||||
"sha256": "49c700c76dc3164c2e97738e53e6958ed14f1c17713c0ac215474b700dee9794",
|
||||
"method": "playwright-page"
|
||||
}
|
||||
],
|
||||
"contrast": [
|
||||
{
|
||||
"label": "green summary",
|
||||
"color": "rgb(40, 56, 47)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 12.375218151996188
|
||||
},
|
||||
{
|
||||
"label": "green note",
|
||||
"color": "rgb(85, 107, 135)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 5.469902501657171
|
||||
},
|
||||
{
|
||||
"label": "blue summary",
|
||||
"color": "rgb(36, 51, 75)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 12.7261609258878
|
||||
},
|
||||
{
|
||||
"label": "blue note",
|
||||
"color": "rgb(85, 107, 135)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 5.469902501657171
|
||||
},
|
||||
{
|
||||
"label": "purple summary",
|
||||
"color": "rgb(55, 51, 63)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 12.296868538533582
|
||||
},
|
||||
{
|
||||
"label": "purple note",
|
||||
"color": "rgb(85, 107, 135)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 5.469902501657171
|
||||
}
|
||||
],
|
||||
"errors": [],
|
||||
"externalBlocked": [],
|
||||
"syntheticRequests": 5,
|
||||
"originalUnknownAttempts": 1,
|
||||
"originalMetadataExact": true,
|
||||
"originalCipherExact": true,
|
||||
"originalSnapshotExact": true,
|
||||
"originalHistoryExact": true,
|
||||
"realModelRequests": 0,
|
||||
"realCredentialsRead": 0,
|
||||
"productionWrites": 0,
|
||||
"viewportWidths": [
|
||||
1280,
|
||||
390
|
||||
]
|
||||
},
|
||||
{
|
||||
"stage": "electron-ui-1",
|
||||
"resultFile": "electron-ui-1/electron-8288962e-dccf-4b68-af88-c1ab414a6e0d/result.json",
|
||||
"success": true,
|
||||
"appRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m2-native-cache-state-2026-11-07\\candidate-app-5",
|
||||
"actualElectron": true,
|
||||
"hostUrl": "http://127.0.0.1:49470",
|
||||
"checks": [
|
||||
"Existing pools retain their original Run/disabled scope without saving new defaults or issuing model requests",
|
||||
"All four pool scopes save through the real form and survive reload; new pools start with session affinity and explicit selection persists",
|
||||
"The request ledger shows completed-response affinity and applicable native states separately from the retained original unknown state and cache usage",
|
||||
"The actual task inspector exposes the frozen state projection and upstream-owned context limits at 390px",
|
||||
"Changed provider scope reselects the next eligible pool member, retains old states and one original upstream 400, and fabricates no old metadata",
|
||||
"Manual model selection takes precedence, transmits its changed header and preserves the original 400 once without applying old opaque history or replaying tools",
|
||||
"State details and scope settings remain readable and usable in the three existing palettes without narrow-screen document overflow",
|
||||
"Candidate host restart retains saved scope, original pools, native cache settings, original metadata/cipher/snapshot and history; no tool or request is replayed"
|
||||
],
|
||||
"screenshots": [
|
||||
{
|
||||
"file": "existing-pools-1280.png",
|
||||
"sha256": "28cf8b907206581c360332aa4b20c6e7b035a985a1d2ac6b9ee67e8c0871bb19",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 1280,
|
||||
"height": 960
|
||||
}
|
||||
},
|
||||
{
|
||||
"file": "new-session-pool-1280.png",
|
||||
"sha256": "c0db064b9079b100149d71d62e63b41a30a025aeb0551f593531224c2ee08e71",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 1280,
|
||||
"height": 960
|
||||
}
|
||||
},
|
||||
{
|
||||
"file": "task-pool-390.png",
|
||||
"sha256": "dc80d6680a3c95e1c3008dc555e77250cb69475e7ac7c6f5af604c9f474409a5",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 390,
|
||||
"height": 844
|
||||
}
|
||||
},
|
||||
{
|
||||
"file": "continued-history-1280.png",
|
||||
"sha256": "ac66c9a0f476b46ac10302c310cf1e1328839d7408033297a1ef64a79aefe820",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 1280,
|
||||
"height": 960
|
||||
}
|
||||
},
|
||||
{
|
||||
"file": "context-state-390.png",
|
||||
"sha256": "8a59a8897f23201cc93ef633ad4cdfd7febd6aeb1d02723be1fcd26bf21995f8",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 390,
|
||||
"height": 844
|
||||
}
|
||||
},
|
||||
{
|
||||
"file": "changed-state-original-error-390.png",
|
||||
"sha256": "a74278d90550e334bcf34492bf9887268a936a36c0a3e36b05366e499e33a52a",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 390,
|
||||
"height": 844
|
||||
}
|
||||
},
|
||||
{
|
||||
"file": "manual-selection-original-error-390.png",
|
||||
"sha256": "faf88e5559ba552d749fd8df559eb7f6ae45a73482b426a68d5133153c96367f",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 390,
|
||||
"height": 844
|
||||
}
|
||||
},
|
||||
{
|
||||
"file": "state-green-390.png",
|
||||
"sha256": "beeb74a2e0aa01b5751523addf4ee9d98361b8e540885f4a5877479b540fd702",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 390,
|
||||
"height": 844
|
||||
}
|
||||
},
|
||||
{
|
||||
"file": "state-blue-390.png",
|
||||
"sha256": "b5fef93757654577f56bfbd058192d9d3e94f506ff605aeecf9501669f2d88e0",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 390,
|
||||
"height": 844
|
||||
}
|
||||
},
|
||||
{
|
||||
"file": "state-purple-390.png",
|
||||
"sha256": "fa9f551ad43bd5e8f38f8fb3eaf4a51c28a5af04b988c48ff455daa1024d8a0b",
|
||||
"method": "electron-native-stay-hidden",
|
||||
"size": {
|
||||
"width": 390,
|
||||
"height": 844
|
||||
}
|
||||
}
|
||||
],
|
||||
"contrast": [
|
||||
{
|
||||
"label": "green summary",
|
||||
"color": "rgb(40, 56, 47)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 12.375218151996188
|
||||
},
|
||||
{
|
||||
"label": "green note",
|
||||
"color": "rgb(85, 107, 135)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 5.469902501657171
|
||||
},
|
||||
{
|
||||
"label": "blue summary",
|
||||
"color": "rgb(36, 51, 75)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 12.7261609258878
|
||||
},
|
||||
{
|
||||
"label": "blue note",
|
||||
"color": "rgb(85, 107, 135)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 5.469902501657171
|
||||
},
|
||||
{
|
||||
"label": "purple summary",
|
||||
"color": "rgb(55, 51, 63)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 12.296868538533582
|
||||
},
|
||||
{
|
||||
"label": "purple note",
|
||||
"color": "rgb(85, 107, 135)",
|
||||
"background": "rgb(255, 255, 255)",
|
||||
"fontSize": "12px",
|
||||
"ratio": 5.469902501657171
|
||||
}
|
||||
],
|
||||
"errors": [],
|
||||
"externalBlocked": [],
|
||||
"syntheticRequests": 5,
|
||||
"originalUnknownAttempts": 1,
|
||||
"originalMetadataExact": true,
|
||||
"originalCipherExact": true,
|
||||
"originalSnapshotExact": true,
|
||||
"originalHistoryExact": true,
|
||||
"realModelRequests": 0,
|
||||
"realCredentialsRead": 0,
|
||||
"productionWrites": 0,
|
||||
"viewportWidths": [
|
||||
1280,
|
||||
390
|
||||
]
|
||||
}
|
||||
],
|
||||
"realModelRequests": 0,
|
||||
"realCredentialsRead": 0,
|
||||
"productionWrites": 0,
|
||||
"totalFinalEndToEndRequests": 10,
|
||||
"allEndToEndRequestsNotWholeStageTotal": true,
|
||||
"success": true
|
||||
},
|
||||
"review": {
|
||||
"schemaVersion": 1,
|
||||
"milestone": "M2 I09/I11 native cache blocks and cross-Run provider state",
|
||||
"sourceCommits": {
|
||||
"sources": "71f1805561e6082011ed4f0c70bb38f4546413c5",
|
||||
"core": "9e1feeebb3eca5c82bfa9efb4f32e68a73628df4",
|
||||
"ui": "ee97aaae7389e0aff7f89eced8d06ecce7d4f982"
|
||||
},
|
||||
"implementationReview": [
|
||||
"Messages system/user/tool/tool_result block order, cache_control, independent TTLs and native extensions retain their original positions.",
|
||||
"Responses complete output items, encrypted reasoning, phase, native tool metadata and explicit cache controls retain supplied upstream fields.",
|
||||
"Schema2 native assistant state is bound to protocol, actual request URL, exact model, connection, credential revision and request header template. Unknown or incompatible original state remains stored; the request uses ordinary text and original tool calls/results.",
|
||||
"Plain Chat pools now project history for every member when provider state exists, including same-model members with different credentials; switching only the model field no longer sends the first member's opaque reasoning to another route.",
|
||||
"New pools default to session affinity, with persisted task/run/none choices; original pools keep their original run/disabled settings. Optional affinity persistence failure does not turn a completed upstream response into an error or cause replay.",
|
||||
"Model center saves actual scope; ledger shows selected-route state and affinity provenance; context shows candidate connection count without adding the same state across projections.",
|
||||
"Default unrestricted/passthrough/preserve, persisted ask/auto/full, user native parameters and complete history remain unchanged. No cache controls or optional output-token fields are automatically added.",
|
||||
"Only completed response affinity is stored across Runs; manual selection wins and independent diagnostics do not overwrite main-task continuity. Completed/interrupted tools are not replayed by these metadata paths."
|
||||
],
|
||||
"validation": {
|
||||
"sourceFinal": "source-5.log: 16/16, original source-1..4 preserved",
|
||||
"firstCandidate": "candidate-backend-1: 127/129, two outdated affinity assumptions failed",
|
||||
"correctedCandidateFiles": "candidate-backend-2: 26/26 for both affected test files",
|
||||
"coverage": "103 unchanged passed tests reused by exact backend/test SHA plus 26 corrected-file tests; no single final full-suite claim",
|
||||
"finalCandidate": "build-5/candidate-app-5: 880 inputs, 300 product sources, 178 backend inputs, 213 test inputs, 29 compiled outputs, 74 runtime files",
|
||||
"finalWeb": "web-ui-5: 8 groups, 10 screenshots, 5 owned HTTP requests",
|
||||
"finalElectron": "electron-ui-1: 8 groups, 10 native hidden-window screenshots, 5 owned HTTP requests",
|
||||
"retention": "Original metadata, DPAPI ciphertext, stored model snapshot and original history exact on both ends; saved synthetic credentials/native parameters survive independent host restart; one executed list_files tool, no replay",
|
||||
"sourceIdentity": "22 changed frozen inputs match current Git blobs and working bytes; 8 official captures and source manifest match their original SHA",
|
||||
"oldSeals": "16171 original entries across I10/I13/I15/I14 roots checked read-only, original manifest SHA exact",
|
||||
"cleanup": "No current candidate node/Electron processes or listeners at recorded nonformal ports; formal 4318 excluded"
|
||||
},
|
||||
"visualReview": {
|
||||
"workflow": "Existing Orven React model center and task inspector, Chinese programming workflow, persistent scope and request provenance; no new console",
|
||||
"reviewedScreenshots": [
|
||||
"web-ui-5/web-d6b3ea8a-032b-4eef-a718-0efd492e3803/new-session-pool-1280.png",
|
||||
"web-ui-5/web-d6b3ea8a-032b-4eef-a718-0efd492e3803/state-green-390.png",
|
||||
"web-ui-5/web-d6b3ea8a-032b-4eef-a718-0efd492e3803/state-blue-390.png",
|
||||
"web-ui-5/web-d6b3ea8a-032b-4eef-a718-0efd492e3803/state-purple-390.png",
|
||||
"electron-ui-1/electron-8288962e-dccf-4b68-af88-c1ab414a6e0d/new-session-pool-1280.png",
|
||||
"electron-ui-1/electron-8288962e-dccf-4b68-af88-c1ab414a6e0d/context-state-390.png",
|
||||
"electron-ui-1/electron-8288962e-dccf-4b68-af88-c1ab414a6e0d/manual-selection-original-error-390.png",
|
||||
"electron-ui-1/electron-8288962e-dccf-4b68-af88-c1ab414a6e0d/state-purple-390.png"
|
||||
],
|
||||
"observations": "Real Chinese text, saved fields, long request URLs/state SHA and original 400 remain readable at 1280/390. Existing horizontal tab scrolling is separate from document overflow. All 20 final screenshots SHA-checked, 8 opened for visual review; measured summary/note contrasts are at least 5.46.",
|
||||
"failuresPreserved": "web-ui-1 label matching; web-ui-2 viewport transition hides existing narrow inspector; web-ui-3 expected A after automatic reselection chose B; web-ui-4 attempted background tools while narrow inspector was open. Drivers corrected from observed UI/request evidence; no force clicks."
|
||||
},
|
||||
"historical": [
|
||||
"W/AD/AE/AG network failures and unknown usage retain original classifications and receipts.",
|
||||
"Repeated-reading general improvement is unproven; R/W reserve-v2 live exclusion boundary remains untested.",
|
||||
"Purge benefit remains unknown; this batch sends no real model requests and does not resume old experiments or user projects.",
|
||||
"Mobile metadata performance, original host identity observation failure/late lifecycle cause and physical-device paths remain open."
|
||||
],
|
||||
"remaining": [
|
||||
"I27/I28 CI with separate runnable checks and artifact/source binding.",
|
||||
"M3 final independent source/program release, actual schema2 migration deltas and rollback after the new handler has produced records; do not remove new ledger entries or the existing per-table equality guard.",
|
||||
"M4 new official/compatible-provider live compatibility and autonomous Node/React/Python samples, each frozen separately after M3.",
|
||||
"M5 remaining native provider/media/server-tool protocols; M6 real engineering/computer/interactive-host paths; M7 performance/physical-phone/ecosystem; M8 comparable actual benefit and Purge AB/BA."
|
||||
],
|
||||
"productionDeployed": false,
|
||||
"officialServiceHealthRechecked": false,
|
||||
"realModelRequests": 0,
|
||||
"realCredentialsRead": 0,
|
||||
"productionWrites": 0,
|
||||
"installed": false,
|
||||
"restartedProduction": false,
|
||||
"delegated": false,
|
||||
"sourceAcceptanceComplete": true,
|
||||
"wholeSystemOrLiveBenefitComplete": false
|
||||
},
|
||||
"priorSeals": {
|
||||
"checkedAt": "2026-11-07T12:28:30.311Z",
|
||||
"records": [
|
||||
{
|
||||
"root": ".qa/m2-cache-usage-2026-11-07",
|
||||
"entries": 2948,
|
||||
"bytes": 1018919970,
|
||||
"manifestSha256": "b3438146114a6c0f0f303f5fa025ec11c647d560188beb14ae0405aaa9a202f7",
|
||||
"entriesExact": true,
|
||||
"dependencyLinksNotTraversed": 3,
|
||||
"writes": 0
|
||||
},
|
||||
{
|
||||
"root": ".qa/m2-request-ledger-2026-11-07",
|
||||
"entries": 4328,
|
||||
"bytes": 1620432307,
|
||||
"manifestSha256": "6ed883ec1fb3e2147795f696802b575e4508e15775c4b764bc8aea166aa34035",
|
||||
"entriesExact": true,
|
||||
"dependencyLinksNotTraversed": 4,
|
||||
"writes": 0
|
||||
},
|
||||
{
|
||||
"root": ".qa/m2-official-registry-2026-11-07",
|
||||
"entries": 5688,
|
||||
"bytes": 1037272654,
|
||||
"manifestSha256": "fbe078468917d79b280aecd4038e537b87ec5de212a772449ba000d7e507d763",
|
||||
"entriesExact": true,
|
||||
"dependencyLinksNotTraversed": 6,
|
||||
"writes": 0
|
||||
},
|
||||
{
|
||||
"root": ".qa/m2-request-contract-2026-11-07",
|
||||
"entries": 3207,
|
||||
"bytes": 1498866416,
|
||||
"manifestSha256": "3963ad9e02d3f62c1ae4d5fefcb59fdc0bcdc63682be1c72109542a60a2b408c",
|
||||
"entriesExact": true,
|
||||
"dependencyLinksNotTraversed": 3,
|
||||
"writes": 0
|
||||
}
|
||||
],
|
||||
"realCredentialsRead": 0,
|
||||
"productionWrites": 0,
|
||||
"success": true
|
||||
},
|
||||
"cleanup": {
|
||||
"checkedAt": "2026-11-07T12:33:10.4367296Z",
|
||||
"root": "m2-native-cache-state-2026-11-07",
|
||||
"count": 0,
|
||||
"listenCount": 0,
|
||||
"processes": [],
|
||||
"listeners": [],
|
||||
"portsChecked": [
|
||||
11434,
|
||||
49470,
|
||||
49869,
|
||||
51514,
|
||||
56527,
|
||||
56542,
|
||||
56544,
|
||||
62945,
|
||||
63879
|
||||
],
|
||||
"formalPortExcluded": 4318,
|
||||
"realCredentialsRead": 0,
|
||||
"productionWrites": 0
|
||||
},
|
||||
"evidence": {
|
||||
"path": ".qa/m2-native-cache-state-2026-11-07/MANIFEST.json",
|
||||
"files": 5126,
|
||||
"bytes": 2008555583,
|
||||
"manifestSha256": "c887364e54fe3f68e5f432155db87517fb8413966f5786156bedceddadbf2217",
|
||||
"dependencyLinksRecordedNotTraversed": 5,
|
||||
"sealedReadOnlyFromNow": true,
|
||||
"subsequentPushReceiptsSeparate": true
|
||||
},
|
||||
"productionDeployed": false,
|
||||
"officialServiceHealthRechecked": false,
|
||||
"realModelRequests": 0,
|
||||
"realCredentialsRead": 0,
|
||||
"productionWrites": 0,
|
||||
"installed": false,
|
||||
"restartedProduction": false,
|
||||
"delegated": false,
|
||||
"scope": "Source and independent deterministic candidate only; no live-provider, delivery or cache/Purge benefit claim",
|
||||
"historicalUnknown": [
|
||||
"W/AD/AE/AG network failures and unknown usage retain original classifications and receipts.",
|
||||
"Repeated-reading general improvement is unproven; R/W reserve-v2 live exclusion boundary remains untested.",
|
||||
"Purge benefit remains unknown; this batch sends no real model requests and does not resume old experiments or user projects.",
|
||||
"Mobile metadata performance, original host identity observation failure/late lifecycle cause and physical-device paths remain open."
|
||||
],
|
||||
"remaining": [
|
||||
"I27/I28 CI with separate runnable checks and artifact/source binding.",
|
||||
"M3 final independent source/program release, actual schema2 migration deltas and rollback after the new handler has produced records; do not remove new ledger entries or the existing per-table equality guard.",
|
||||
"M4 new official/compatible-provider live compatibility and autonomous Node/React/Python samples, each frozen separately after M3.",
|
||||
"M5 remaining native provider/media/server-tool protocols; M6 real engineering/computer/interactive-host paths; M7 performance/physical-phone/ecosystem; M8 comparable actual benefit and Purge AB/BA."
|
||||
]
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,429 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"verifiedAt": "2026-11-11T14:57:36.924Z",
|
||||
"finalPass": true,
|
||||
"finalBoundaryCases": 11,
|
||||
"finalFixtureModelRequests": 15,
|
||||
"originalNodeTests": 9,
|
||||
"originalNodeTestsPassed": 9,
|
||||
"fixedImplementationSuppliedOnlyToFixture": true,
|
||||
"realModelRequests": 0,
|
||||
"installedProgram": {
|
||||
"version": "1.25.13",
|
||||
"root": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\versions\\Agent-Workbench-1.25.13-win-x64-20261110190745270",
|
||||
"serverBundlePath": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\versions\\Agent-Workbench-1.25.13-win-x64-20261110190745270\\dist-server\\server.mjs",
|
||||
"serverBundleSha256": "35b3d1502da0121d12a2e1ceae9ca96f1d03391cd60d9e4d4d0e501d1d3320b3"
|
||||
},
|
||||
"runs": [
|
||||
{
|
||||
"root": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m4-driver-a01e6b39-4119-4069-b188-7f66ed83978d",
|
||||
"resultSha256": "6fa66bb14568e324289b884ca19061f9d481cd80398919e04d311d5b12837f93",
|
||||
"success": false,
|
||||
"fixtureRequests": 10,
|
||||
"engineeringRequests": 0,
|
||||
"boundaryCases": [
|
||||
{
|
||||
"prefix": "preflight/chat-completions-text",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/chat-completions-tools",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/responses-text",
|
||||
"protocol": "responses",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/responses-tools",
|
||||
"protocol": "responses",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/anthropic-messages-text",
|
||||
"protocol": "anthropic-messages",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/anthropic-messages-tools",
|
||||
"protocol": "anthropic-messages",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/json",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/streaming-override",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "streaming",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/m4-error",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "error",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
}
|
||||
],
|
||||
"firstError": "AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:\n+ actual - expected\n\n+ 'interrupted'\n- 'failed'\n\n at runProbe (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/m4-live/verify-driver.mjs:114:19)\n at process.processTicksAndRejections (node:internal/process/task_queues:104:5)\n at async file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/m4-live/verify-driver.mjs:129:5",
|
||||
"realModelRequests": 0
|
||||
},
|
||||
{
|
||||
"root": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m4-driver-16e746a7-2ba0-41ac-9496-e19fdacd04dc",
|
||||
"resultSha256": "3e10b15415aa8dfb318be6191dbcc697d06927763bcbed91235df9fad43c28cd",
|
||||
"success": false,
|
||||
"fixtureRequests": 15,
|
||||
"engineeringRequests": 5,
|
||||
"boundaryCases": [
|
||||
{
|
||||
"prefix": "preflight/chat-completions-text",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/chat-completions-tools",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/responses-text",
|
||||
"protocol": "responses",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/responses-tools",
|
||||
"protocol": "responses",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/anthropic-messages-text",
|
||||
"protocol": "anthropic-messages",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/anthropic-messages-tools",
|
||||
"protocol": "anthropic-messages",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/json",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/streaming-override",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "streaming",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/m4-error",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "error",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/m4-unknown",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "unknown",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
}
|
||||
],
|
||||
"firstError": "AssertionError [ERR_ASSERTION]: {\"failure\":null,\"final\":{\"test\":{\"status\":0,\"signal\":null,\"error\":null,\"stats\":{\"tests\":9,\"pass\":9,\"fail\":0,\"cancelled\":0,\"skipped\":0,\"todo\":0},\"stdoutSha256\":\"0f5fc1b31bd42ae405bf71ef06ef9f352cd2eaf63cf9f8f15fe79009003d3a5c\",\"stderrSha256\":\"e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855\"},\"build\":{\"status\":0,\"signal\":null,\"error\":null,\"stats\":null,\"stdoutSha256\":\"9307ee4e4f72392dd2f29f26c8e9fd3c8e55353bbe21f93627f260f108be9c2c\",\"stderrSha256\":\"e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855\"}},\"cleanupErrors\":[{\"label\":\"verify-config-and-definitions\",\"error\":\"AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:\\n+ actual - expected\\n... Skipped lines\\n\\n [\\n {\\n baseUrl: 'http://127.0.0.1:11434/v1',\\n brand: '自定义',\\n capabilities: {\\n...\\n requiresKey: false,\\n+ revision: 4,\\n- revision: 2,\\n source: 'legacy',\\n streaming: false\\n },\\n {\\n baseUrl: 'http://127.0.0.1:65032/v1',\\n\\n at file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/m4-live/run.mjs:91:12\\n at process.processTicksAndRejections (node:internal/process/task_queues:104:5)\\n at async guard (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/m4-live/run.mjs:67:83)\\n at async file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/m4-live/run.mjs:87:3\"}],\"requests\":5}\n\nfalse !== true\n\n at file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/m4-live/verify-driver.mjs:139:97\n at process.processTicksAndRejections (node:internal/process/task_queues:104:5)",
|
||||
"realModelRequests": 0
|
||||
},
|
||||
{
|
||||
"root": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m4-driver-490968ae-ff86-4217-8e33-98ce56af5f5d",
|
||||
"resultSha256": "25af3ee86e1ee9dd0fd97b0d1799fbdcd013f08ddd37aab11a61bd38eb6eaa45",
|
||||
"success": true,
|
||||
"fixtureRequests": 15,
|
||||
"engineeringRequests": 5,
|
||||
"boundaryCases": [
|
||||
{
|
||||
"prefix": "preflight/chat-completions-text",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/chat-completions-tools",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/responses-text",
|
||||
"protocol": "responses",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/responses-tools",
|
||||
"protocol": "responses",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/anthropic-messages-text",
|
||||
"protocol": "anthropic-messages",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/anthropic-messages-tools",
|
||||
"protocol": "anthropic-messages",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/json",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/streaming-override",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "streaming",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/m4-error",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "error",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/m4-unknown",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "unknown",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
}
|
||||
],
|
||||
"firstError": null,
|
||||
"realModelRequests": 0
|
||||
},
|
||||
{
|
||||
"root": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\m4-driver-40fb93c8-1de0-4590-a611-bc6f51406b60",
|
||||
"resultSha256": "24c6db8cd81f7831669cb07a85e6005f99714cdd831e10ce23fc2825c400979f",
|
||||
"success": true,
|
||||
"fixtureRequests": 15,
|
||||
"engineeringRequests": 5,
|
||||
"boundaryCases": [
|
||||
{
|
||||
"prefix": "byte-preservation",
|
||||
"expected": "exact-invalid-utf8-bytes",
|
||||
"requests": 0,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/chat-completions-text",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/chat-completions-tools",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/responses-text",
|
||||
"protocol": "responses",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/responses-tools",
|
||||
"protocol": "responses",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/anthropic-messages-text",
|
||||
"protocol": "anthropic-messages",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/anthropic-messages-tools",
|
||||
"protocol": "anthropic-messages",
|
||||
"scope": "tools",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/json",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/streaming-override",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "streaming",
|
||||
"expected": "success",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/m4-error",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "error",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
},
|
||||
{
|
||||
"prefix": "preflight/m4-unknown",
|
||||
"protocol": "chat-completions",
|
||||
"scope": "text",
|
||||
"expected": "unknown",
|
||||
"requests": 1,
|
||||
"passed": true
|
||||
}
|
||||
],
|
||||
"firstError": null,
|
||||
"realModelRequests": 0
|
||||
}
|
||||
],
|
||||
"finalSourceFiles": [
|
||||
{
|
||||
"path": "scripts/m4-live/common.mjs",
|
||||
"bytes": 8155,
|
||||
"sha256": "df14cc482dcac914ac70d57d7ee8ed4df4be8c06281c2ead49963977b33e0cb3"
|
||||
},
|
||||
{
|
||||
"path": "scripts/m4-live/compatibility.mjs",
|
||||
"bytes": 7897,
|
||||
"sha256": "fac6e3edd6068e7369ce60e311480adf282b95f1a5a61ad3bd4ce3d5babb1ce6"
|
||||
},
|
||||
{
|
||||
"path": "scripts/m4-live/initial.mjs",
|
||||
"bytes": 2419,
|
||||
"sha256": "8e1207dc8d483418efae6ded248ef689857ed300c2db43a413e5b169728dafcd"
|
||||
},
|
||||
{
|
||||
"path": "scripts/m4-live/inspector.mjs",
|
||||
"bytes": 3432,
|
||||
"sha256": "c179f4e0236d5b03d61965e491eb5e4f1de45f881efa2885008c474eb3c36b8b"
|
||||
},
|
||||
{
|
||||
"path": "scripts/m4-live/payload.mjs",
|
||||
"bytes": 7371,
|
||||
"sha256": "914e27ae815fd79f8ed70604297c154b8874cb1a72e5e1ecd11286f19c5b21ff"
|
||||
},
|
||||
{
|
||||
"path": "scripts/m4-live/project.mjs",
|
||||
"bytes": 4497,
|
||||
"sha256": "16264b327e73a82476a5ae8ebada6b96189e0197d8a9f4423a6eb807b9f3f3da"
|
||||
},
|
||||
{
|
||||
"path": "scripts/m4-live/run.mjs",
|
||||
"bytes": 11666,
|
||||
"sha256": "0657f9e7489389f7991f8ba9646bed0d7289e22acbb0afe949b9443ea09a789f"
|
||||
},
|
||||
{
|
||||
"path": "scripts/m4-live/share-existing-ad.mjs",
|
||||
"bytes": 8732,
|
||||
"sha256": "71b0fdbd12dcdbbeae1d274c06d557a407c92ff37ff61d499a53e160aeccf204"
|
||||
},
|
||||
{
|
||||
"path": "scripts/m4-live/verify-driver.mjs",
|
||||
"bytes": 15450,
|
||||
"sha256": "066e7a9277186b9c56203d54e72ea9608a16b1b7833787e8926993c9d31a4af3"
|
||||
}
|
||||
],
|
||||
"scope": "Final byte-download path and all three protocols, source captures, exact cancel/error status, original Node checks, owner/index, Git and config restoration. First failures retained; no provider capability or autonomous-engineering claim."
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,20 @@
|
||||
{
|
||||
"files": 541,
|
||||
"bytes": 134884464,
|
||||
"manifestSha256": "558c9b548559798279d7f8e61deb3b7344920753a53dbf4e84eec374c6f3a3df",
|
||||
"manifestBytes": 102840,
|
||||
"excluded": [
|
||||
{
|
||||
"path": "react/project/node_modules",
|
||||
"scope": "Reconstructible installed dependencies only; original package-lock and npm-ci log included.",
|
||||
"lockSha256": "48c138a05408667bbd332000731598415c9ae281d5072daa18544025ad89f7ff"
|
||||
}
|
||||
],
|
||||
"modelRequests": 40,
|
||||
"reportedTotalTokens": 629106,
|
||||
"unknownRequests": 0,
|
||||
"externalRequests": 1,
|
||||
"keysReadOrResaved": false,
|
||||
"oldSamplesReplayed": false,
|
||||
"purgeExecuted": false
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
+58
@@ -1,5 +1,45 @@
|
||||
# 汇总发现
|
||||
|
||||
2026-11-12 重启丢Key的根因是三个正常配置入口默认临时内存,已有DPAPI没有成为默认。1.25.14已修复默认/原子保存和清除;54相关、真实新进程及当前Key保存、实际两端和隔离安装通过。正式七条Key原已有有效副本,保留SHA后真正重启为PID29884全部自动恢复,原数据/选择/修订/420请求保持;不是供应商错误。首失败、基线虚拟目录缺陷、原生验收入口错误及封存依赖引用拒绝均保留;715成员封存,见docs/QA-CREDENTIAL-PERSISTENCE-1.25.14.md。Claude403/历史unknown/Purge未验不改签。
|
||||
|
||||
2026-11-12 Mnemo来源指定旧AG的同Goal工件连接已有限完成;113=19+94、独立69/31隔离HTTP分别核回执,原9/build/57及8请求109888tokens不重复计。本轮只读九副本/五发布回执SHA,4321的404是来源旧观察,不称当前生产接通;生产者/完整指纹/派发接收/父交付/语义未验。Claude403两请求/usage未知和原541/41清单保持,见docs/integration/MNEMO-GOAL-WORKER-READONLY-2026-11-12.md;不重派或补签AD/M4。
|
||||
|
||||
2026-11-12 修后Claude验收5ca65a45和驱动97f132d3已普通推送、远端/fetch逐SHA一致,见docs/gitatlas-claude-proxy-correction-push.json;原M4收尾c8c36c27也已完成。541旧封存/41新封存保持,Git不含原件/Key/程序包。目录成功与403客户端权限失败并列,真实usage/费用null和全部历史收益unknown保持。
|
||||
|
||||
2026-11-12 修后Claude的目录已完整14型号且含选定,但两模型请求均403 permission_error,原文明确分组仅允许Claude Code客户端。实际/v1/messages、x-api-key占位、stream及max_tokens128000字段完整,响应188字节JSON完整,不能因.sse后缀当流式通过;无型号/usage,实际tokens/费用null。规划requestId/物理wireAttemptId从原账本登记,可选requestKey空字段保持,不重派。原状态与当前七连接一审通过,41新封存及旧541保持;见docs/QA-CLAUDE-PROXY-CORRECTION-2026-11-12.md。需要供应商普通API客户端权限,不以目录成功认证权重或收益。
|
||||
|
||||
2026-11-12 M4普通推送e132601b成功,前六条连同验收七条逐SHA与远端/fetch一致,37待推送文件无私钥/真实Key模式或归档/依赖路径,见docs/gitatlas-m4-live-push.json。源码包和证据原件未入Git,runner结果仍unknown;修后Claude另记录,原M4事实不改签。
|
||||
|
||||
2026-11-12 M4最终原状态审计通过,原53527行/127导出/5244工件/377请求与本批九捕获导出保持;三工程共34请求,含预检总40请求629106报告tokens/0unknown。三工程同SHA同范围重复读取0、440960cacheRead仅为观察,cacheWrite未知不补0。541成员134884464字节封存SHA558c9b54及十三驱动已复核;见[最终验收](docs/QA-M4-LIVE-BASELINE.md)。人类在封存后报告Claude代理Key此前写错并更正,原因尚无独立证明;旧两次自报GPT/目录缺项完整保持,修后另冻至多两新预检,原审计不能用于回退人类配置。NRO与AD/另选AG分别只读登记,不代签生产语义或Purge。
|
||||
|
||||
I27/I28推送完成:源码与验收44b6a09b已普通增量推送至既有GitAtlas分支,远端HEAD和重新fetch的14个提交逐SHA一致,见[回执](docs/gitatlas-m2-ci-push.json)。回执/完成接续最后另提交并再次核对;最终4a03ec52干净1222/1222及7595封存原件保持,正式未部署/重启,runner/job/真实供应商及历史收益unknown不代签。下一按docs/M3-RELEASE-PLAN.md推进实际schema2迁移与新结果后的兼容回退;下方进行中说明按原时点理解。
|
||||
|
||||
I27/I28最终干净Windows自然1222/1222、146文件,3392886ms;实际两端7组4图及字体/两RAF/PNG尺寸/原DOM全部复核,1376导出输入/29编译不变。新7595原件150586505字节已封存,五旧M2根21297原件7184046930字节及原清单SHA一致;首1211/17失败和第二轮原生PNG滞后完整保持。2001 companion记录测试872353.5061ms,不把正确保留写成性能优化;远端runner/job/上传unknown。后续doc HEAD与4a03ec52程序输入一致,正式未部署,下一M3需精确迁移/新结果后兼容回退,不能删账本或降schema。
|
||||
|
||||
I27/I28本次接续:开放模式广告曾绕过taskToolAllowed,执行端仍遵守原诊断/无需测试约定;本轮只对齐广告/排除原因,未新增执行守卫。研究首失败来自旧16K受控夹具Evidence元数据预算,不是外部模型故障。原生截图存在DOM已经模型中心而PNG仍加载/工具箱的真实证据差异;第二轮不作视觉通过,第三轮隐藏QA窗口关闭background throttling并等加载/字体/有限动画/两RAF后四图实看通过。最终干净已提交源码全套仍待,历史失败/unknown完整保留。
|
||||
|
||||
I27/I28公开只读调查:GitAtlas实返Gitea1.26.2、仓库has_actions=true;当前API公开private=false,不擅自修改可见性。Actions运行与runner接口匿名均401,不能据此断言有可用执行器。已保存1.26版quickstart/act-runner/comparison三份HTTP200原HTML及SHA;最新入口的308/301亦保持,不当正文成功。本地固定Node24.22.1已核,planning catchup退出0无补遗。
|
||||
|
||||
I27/I28初核:现有构建依赖Node24与dist-server,Python测试需独立pytest8.4.2,Playwright/Electron版本锁在package-lock;没有既有.gitea/.github工作流,GitAtlas Actions版本/启用/runner未知。源码包按目录白名单导出,暂不含.gitea,需要实际Git ZIP核对。历史manual/restricted fixture保留,不能为CI删除旧断言或以原依赖junction宣称干净安装。
|
||||
|
||||
I09/I11封存收口:5126原件、2008555583字节、5依赖junction不递归;原四M2根16171原件/manifest只读一致,22源码Git字节一致,实际候选进程/记录非正式端口监听0。源码/隔离验收完成与正式部署/真实缓存/Purge收益分别记录;下一CI后M3必须验证迁移增量与新结果回退。封存后误用Tianshu文档标题的失败patch只记录在新close/当前进展,不回写原证据根。
|
||||
|
||||
I09/I11最终Electron通过且实际复看1280新池编辑与390任务上下文:现有Orven布局/中文/长SHA/下拉与真实保存保留,输出“本次不限/不附加上限/由上游处理”与冻结状态分别呈现。网页绿/紫状态图亦复看,三配色和窄屏保持一致。两个端各5本机POST、3成功/2原400,6账本含原unknown;完成工具1、不重放,原metadata/cipher/snapshot/history全部exact。本批不证明真实缓存命中/上游普遍兼容。
|
||||
|
||||
最终网页两图实看:web-ui-5 new-session-pool-1280的连接保持下拉、真实成员及保存动作布局完整;state-blue-390原400/状态数量/地址可读且无文档横溢,顶部当前A与历史B请求分别保留,没有倒改旧请求型号。手机横向导航为既有必要滚动,不误记为文档溢出。实际Electron仍在独立运行,桌面及紫/绿色最终图另验。
|
||||
|
||||
I09/I11最终输入首核对通过:880输入/300产品源码/178后端/213测试/29编译/74runtime。178后端全与首候选相同;213测试输入中211未改,两改测试文件与26/26候选一致,复用103原通过项。8官方原件/source清单与Git原字节一致,来源正文6尾空白不修剪。web-ui-5实际8组/10图、5本机请求+1原unknown账本、原metadata/cipher/snapshot/history exact、三配色被测文字对比≥5.46;实际Electron与清理仍待,不把网页通过当桌面或真实供应商通过。
|
||||
|
||||
I09/I11 web-ui-3四组通过后仍失败:实际第4次选中owned-b,未发送A修改的X-Owned-Scope;原驱动误以为失效亲和仍应选A。原requests.json核对前三次A及相符reasoning,第四次B且无旧reasoning。按实际自动重选保留400,并新增实际模型中心手选A一次核对修改头、原错与不重放;最终应5本机请求/6账本而非旧4/5。候选3产品不变,另冻最后验收输入,避免重复编译与runtime复制。实际已查看web-ui-3的1280账本与390上下文:中文/长地址与状态SHA可读,手机没有文档溢出;最终三配色和Electron仍需验。
|
||||
|
||||
I09/I11接续:candidate-backend-1为127/129,两项旧假设已核对。跨Run权重轮询测试必须显式选run,默认新池应session;关闭保持时不再暗存连接,重新启用run后应先按权重选择,再沿用本次记录。web-ui-2先完成3组/3本机请求,在宽屏打开检视后直接缩到手机触发App已有关闭wide行为,失败现场只显示会话;改为先设390再由真实任务入口打开,不修改已有布局行为。上下文候选状态不能把同一原状态跨连接相加,改为候选连接数,各连接单独列状态数量。
|
||||
|
||||
I09/I11 source-4为14/15:收窄断言后确认真实纯Chat混池缺陷,不是整正文误命中。completionBody在所有成员均无native/streaming时只构造首成员body,actualRequest只换model,旧reasoning_content进入后备连接。现有状态时必须逐成员投影,连同型号相同但凭据不同的池验证。原source-3/4失败保留;原生Messages/Responses SSE、停止续同Task与可选亲和存储故障均在source-4通过。未知亲和记录保留不重写,optional存储失败不改变已完成上游响应的结果。
|
||||
|
||||
I09/I11接续决策:新池可默认会话保持,但旧池必须保留原Run范围;亲和只来自完成的真实响应且诊断路径独立。状态身份要核对实际请求路径/认证模板/版本头/型号/连接与凭据修订,改变原生普通生成参数不应误认凭据变化。状态无法适用时保留原件和原因、以常规文本/原调用构造新历史,不丢用户数据或自动重放工具。沿既有Orven池编辑与上下文/请求账本补真实控制和反馈,不另立页面。
|
||||
|
||||
I09/I11具体缺口:messagesBody重建文本和工具时丢非结构元数据,messagesHistory反向只取text并改变tool_result顺序;Messages/Responses原生助手状态只有endpoint/model绑定,缺实际请求版本/头/凭据范围;ModelHub sticky仅runId且重启不持久。需要按来源区分声明原生输入与不透明输出状态,当前状态不能跨供应商或凭据误用,原件不能因此丢弃。
|
||||
|
||||
原生状态核对新增源方参考:冻结预设/用户设置不等于actual standard会话采用,创建失败也不代表发过模型请求;输入估算/额外预留、压缩请求值与当次fetch数据必须分别核对。手机客户端期限/确认、服务端原任务成功、容量样本和被清理的分段证据不能互相补数或推定延迟段。新同步仅按源方报告登记,自己的透传/预算/身份和原未知结论不改变。
|
||||
|
||||
I14收尾复核:实际请求头观察必须绑定prepared fetch参数,原Run只能表示冻结设置;认证模板/独立Key引用、HTTP传输层补头和真实缓存命中证据分别说明。最终两端各8组验证通知正常滚动、原配置/密文/快照保持;871输入与29输出一致,176后端/209测试未变可复用原143/143。首身份验证误用TAP计数不是产品失败,原件保留后仅修验证器。源码交付不升级为真实供应商/自主工程/Purge收益通过;迁移增量与激活新处理器后的回退仍必须在M3证明。
|
||||
@@ -1415,3 +1455,21 @@ New API学习:305正文/201嵌入定义已分别审读;来源清单与逐篇
|
||||
现有官方导出明确 Claude Opus 5.5 的 thinking 不能使用旧 enabled/disabled/between_tools;模型目录还含 disabled/between_tools/server_tool 子项而当前解析不全。DeepSeek 工具历史要求保留所有 reasoning_content;供应商缓存是 best effort,不能保证不掉缓存。各官方容量/推理/缓存/停止原因需带精确 ID、版本、协议、来源和观测时间,兼容网关信息不能升级成官方承诺。文档头部 D/E 已落后于正式 F,应增加当前入口并保留历史段落。
|
||||
|
||||
收尾新增O21:CredentialVault.reload对合成101条登记正常、102条整体blocked,原metadata不删除;不调用构造器/加解密/真实存储,0真实Key/请求/进程/正式写入。作为I38固定加载总数门槛加入首批修复,最终38事项/21编号观察+O18-C。源715输入SHA一致,24可取官方来源记录与17来源读取失败单列;30能力区域和M0–M8全部映射完成。七当前文档入口、AGENTS及规划指针已统一;历史报告不回写。
|
||||
|
||||
2026-11-11:已恢复会话、核对HEAD f7705d96及原68待提交文件;M3/B4b实现已隔离通过但未发行。选1.25.13交付,不重新做只读协调。
|
||||
2026-11-11:Mnemo原生合成宿主源码SHA与旧v2一致,可在本轮新目录新建合成绑定重验;不使用旧connection/private或旧运行目录,不改Mnemo源码。186项联合回归和构建通过,不将M2历史全量结果借称当前全量。
|
||||
2026-11-11:M3+B4b最终1.25.13已经正式激活,不是仅源码交付。来源35a6d6d9,1399输入/30编译重建exact、186相关/4打包、实际Inno11/11/F跨版7/7/桥24/24与两端/基础文件全部按本轮证据通过。实际Setup351789ms,旧102项目/127会话/8741记录/22085事件/outbox/5244工件保持,374请求迁移schema2;原PID63892/instance/Provider和Key存在状态保留。
|
||||
2026-11-11:跨版本正式验收不能把新增公开字段算成数据损失,也不能直接忽略。实际新增官方modelSpecification/requestContract/能力来源、账本count/分页/存储说明以及execution的mcpJobCount/mcpJobsTruncated,分别按持久registry原声明/固定resolver与原MCP计数核验后,旧公开配置/所有其他字段及127导出原摘要一致。原精确迁移核对保留,sqlite_sequence必须核其声明新增而非强求整体旧SHA。QA误用旧归一默认声明的初次诊断、所有旧断言失败、正常关闭窗口日期类型差异均保留;正式仅一次安装、0请求重放。
|
||||
2026-11-11:正式宽窄四图实看和final后正常恢复新版桌面,后台未重启;选择性782份811582693字节清单611c4c40逐件核对,旧封存清单exact。Mnemo桥默认关闭/单项目/只共享明确来源修订/凭据宿主内存范围不变;真实Mnemo生产绑定和真实模型语义/长期/物理iPhone/Purge收益保持unknown。
|
||||
|
||||
2026-11-11 1.25.13正式交付推送完成:验收1e309700及打包修复35a6d6d9已普通增量推送至既有GitAtlas分支,远端与重新fetch逐SHA一致;docs/gitatlas-mnemo-1.25.13-push.json随完成接续另提交并核最终HEAD。正式后台仍63892/1.25.13,新版桌面已恢复;782封存原件保持、0真实模型/Key提取重存,下一M4新真实基线,Mnemo生产绑定与真实收益未知保持。
|
||||
|
||||
2026-11-11 M4:首次公开API核对正式1.25.13/63892仍空闲,权限/透传/累计0保持;六启用连接当前都hasKey=false,不能套用旧发行时“凭据保留”作为当前可请求证明。已留新observation原件并异步请求界面恢复。旧AD驱动固定1.25.8、旧metadata账本、未允许Messages官方必填输出字段,需要新驱动,不能直接续旧claim或改旧报告。原基线9/9、React9、Python10仅用作未改输入,不拿历史成功代签当前模型。
|
||||
|
||||
2026-11-11 M4:恢复三连接后,Claude官方必填max_tokens在隔离真实wire为128000;Chat/Responses默认无主动上限。新版公开账本取消状态为interrupted,legacy-default revision只是configRevision投影,原字段必须精确核对后单列增量。捕获原件应使用现有download=1原字节接口,JSON字符分页会无法保存非法UTF8;最终边界与工程驱动通过且自有宿主正常关闭,0真实模型。AD只读原报告e7a67a74/bdb282fc逐SHA复制,46文件和两个旧描述符缺项、结构化acceptance空数组/in_progress/终端Git分列,未生成原生v2或扩大授权。
|
||||
|
||||
2026-11-11 M4真实阶段:Claude必填max_tokens=128000已真实发送,missing-field未复现;上游文本/工具都自报gpt-6.1-sol,完整目录11型号未列claude-opus-5-5,因此不能说真实Claude兼容通过。DeepSeek Node自主三文件/9测试/build/Git和内置最终报告闭环,七不同文件读取无同SHA范围重复;cacheRead122112、uncached48911、reasoning6113,cacheWrite十请求unknown,不将单样本观察升级为因果收益。外部新你好会话不是M4请求;53527原行和127原导出/5244原工件核对,model_attempts的项目归属来自sessions关系,无project_id列,驱动误查原失败保持。
|
||||
|
||||
2026-11-11 M4 React:空delivery_job数组不代表未提交,此例模型用run_command终端Git,原terminal-git-evidence verified/currentVerified与最后verified_committed对应c55644a3四文件;processAttribution仍unproven,不补签进程认证。13次read_file同SHA/范围重复0,改后版本读取是新事实;cacheRead196096/uncached68493/reasoning3262、cacheWrite十四unknown,仅描述观察。模型两浏览器作业无非文本工件描述符,未采用额外采集器;只主张独立页面实际两PNG,原0结果和采集源码另留ancillary。
|
||||
|
||||
2026-11-11 M4 Python:原十pytest与compileall、仅normalize/render/summary三文件Git和owner全部index实际通过;早报告in_progress、最终verified_committed分列,终端verified但进程认证unproven保持。六次同SHA/范围重复0,cacheRead122752/uncached46738/reasoning2847,cacheWrite十unknown。新增审计全部九个M4完成会话的捕获导出也必须持续exact,防止只核历史127而忽略本批后续漂移;当前全部通过,不改变产品/Key或旧样本。
|
||||
|
||||
@@ -0,0 +1,13 @@
|
||||
Copyright Eemeli Aro <eemeli@gmail.com>
|
||||
|
||||
Permission to use, copy, modify, and/or distribute this software for any purpose
|
||||
with or without fee is hereby granted, provided that the above copyright notice
|
||||
and this permission notice appear in all copies.
|
||||
|
||||
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH
|
||||
REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
|
||||
FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT,
|
||||
INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS
|
||||
OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
|
||||
TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF
|
||||
THIS SOFTWARE.
|
||||
Generated
+17
-3
@@ -1,12 +1,12 @@
|
||||
{
|
||||
"name": "local-agent-workbench",
|
||||
"version": "1.25.12",
|
||||
"version": "1.25.14",
|
||||
"lockfileVersion": 3,
|
||||
"requires": true,
|
||||
"packages": {
|
||||
"": {
|
||||
"name": "local-agent-workbench",
|
||||
"version": "1.25.12",
|
||||
"version": "1.25.14",
|
||||
"dependencies": {
|
||||
"@babel/parser": "7.29.11",
|
||||
"@peculiar/x509": "3.1.0",
|
||||
@@ -31,7 +31,8 @@
|
||||
"esbuild": "^0.25.10",
|
||||
"innosetup-compiler": "6.3.1",
|
||||
"typescript": "^5.9.3",
|
||||
"vite": "^7.1.12"
|
||||
"vite": "^7.1.12",
|
||||
"yaml": "2.8.1"
|
||||
}
|
||||
},
|
||||
"node_modules/@babel/code-frame": {
|
||||
@@ -3120,6 +3121,19 @@
|
||||
"dev": true,
|
||||
"license": "ISC"
|
||||
},
|
||||
"node_modules/yaml": {
|
||||
"version": "2.8.1",
|
||||
"resolved": "https://registry.npmjs.org/yaml/-/yaml-2.8.1.tgz",
|
||||
"integrity": "sha512-lcYcMxX2PO9XMGvAJkJ3OsNMw+/7FKes7/hgerGUYWIoWu5j/+YQqcZr5JnPZWzOsEBgMbSbiSTn/dv/69Mkpw==",
|
||||
"dev": true,
|
||||
"license": "ISC",
|
||||
"bin": {
|
||||
"yaml": "bin.mjs"
|
||||
},
|
||||
"engines": {
|
||||
"node": ">= 14.6"
|
||||
}
|
||||
},
|
||||
"node_modules/yargs": {
|
||||
"version": "15.4.1",
|
||||
"resolved": "https://registry.npmjs.org/yargs/-/yargs-15.4.1.tgz",
|
||||
|
||||
+6
-2
@@ -1,6 +1,6 @@
|
||||
{
|
||||
"name": "local-agent-workbench",
|
||||
"version": "1.25.12",
|
||||
"version": "1.25.14",
|
||||
"private": true,
|
||||
"type": "module",
|
||||
"description": "Orven — an AI development workspace. From an idea to a running result.",
|
||||
@@ -11,6 +11,9 @@
|
||||
"desktop": "powershell.exe -NoProfile -ExecutionPolicy Bypass -File scripts/Start-Desktop.ps1",
|
||||
"verify:desktop": "node scripts/verify-desktop-window.mjs",
|
||||
"test": "node --test --test-concurrency=2 tests/*.test.mjs",
|
||||
"ci:contract": "node scripts/ci.mjs --profile contract",
|
||||
"ci:windows": "node scripts/ci.mjs --profile windows",
|
||||
"verify:official-contracts": "node scripts/verify-official-contracts.mjs",
|
||||
"verify:ui": "node scripts/verify-ui.mjs",
|
||||
"verify:orven": "node scripts/verify-orven-ui.mjs",
|
||||
"verify:appearance": "node scripts/verify-orven-appearance.mjs",
|
||||
@@ -78,6 +81,7 @@
|
||||
"esbuild": "^0.25.10",
|
||||
"innosetup-compiler": "6.3.1",
|
||||
"typescript": "^5.9.3",
|
||||
"vite": "^7.1.12"
|
||||
"vite": "^7.1.12",
|
||||
"yaml": "2.8.1"
|
||||
}
|
||||
}
|
||||
|
||||
+106
@@ -1,5 +1,81 @@
|
||||
# 统一执行进度
|
||||
|
||||
2026-11-12 1.25.14修复与验收958f31cd及前四条普通增量推送通过,远端/重新fetch及五条逐SHA一致,docs/gitatlas-credential-persistence-1.25.14-push.json保留回执。回执/完成接续最后一起提交推送核最终HEAD;程序未再变更,PID29884与七条自动恢复Key、715封存和原数据保持。下方推送收尾为原时点,M8与全部历史未验另冻。
|
||||
|
||||
2026-11-12 1.25.14凭据重启修复开发、发行和正式验收完成:源码5b51151f/41e59768、验收修正0415e9d5,54相关/1000输入30编译/实际两端10步5图/隔离Inno四组通过。正式Setup234945ms保留PID14388/Provider后,正常退出并从原数据根启动PID29884;七条Key恢复、全部原行/139导出/5583工件和选择/密文/登记193/420请求exact,最终空闲/9229关闭。715原件178252120字节封存;当前限定收尾提交和普通GitAtlas增量推送。真实Key未读取/复制/重加密,真实供应商0,不续旧任务或Purge,历史unknown保持。
|
||||
|
||||
2026-11-12 三项目新增只读来源核对并登记:九公开副本、五已发布回执bytes/SHA、19/94/69计数/类型退出码和旧AG原身份通过;docs/integration/MNEMO-GOAL-WORKER-READONLY-2026-11-12.md保留有限同Goal/显式complete与尚未载入/认证/父交付/语义边界。0模型、0工程、0HTTP或正式服务操作,不回复来源、不改原封存。按既有本地小提交纪律收尾,待后续开发阶段增量推送;不因本协调新增Claude/Purge执行。
|
||||
|
||||
2026-11-12 本轮实际执行/验收/封存/推送结束:原M4 c8c36c27、修后Claude97f132d3/5ca65a45已与远端/fetch一致;docs/gitatlas-claude-proxy-correction-push.json与完成接续最后限定提交后核最终HEAD。541/41封存保持,Claude403未通过且两请求不追加;下一M8另冻,不重跑已过工程或改变旧异常。
|
||||
|
||||
2026-11-12 Claude独立修后范围结束:97f132d3驱动离线10项通过后一次冻结,目录14完整/选定在内,两新例各一请求均上游403“仅Claude Code客户端”,2unknown、费用null、0工具执行。audit-44ea125a一次保持原58298行/137导出/418请求/5576工件及当前七连接/池/选择/配置/凭据SHA;41成员36284086字节SHA9b8a1cfc封存复核,原M4 541清单保持。docs/QA-CLAUDE-PROXY-CORRECTION-2026-11-12.md记录原错误和完整身份,当前限定验收提交/普通GitAtlas推送收尾,不追加或重跑工程/Purge/重启。
|
||||
|
||||
2026-11-12 M4验收e132601b及前六条普通推送成功,七条与远端和重新fetch逐SHA一致;docs/gitatlas-m4-live-push.json回执及完成接续另提交后核最终HEAD。541原件/40请求保持,产品不重发行。下一另冻修后Claude目录及两项预检,原失败和Purge收益unknown保留。
|
||||
|
||||
2026-11-12 M4阶段6原批次最终审计、541成员134884464字节封存和逐SHA复核通过;40请求629106报告tokens/0unknown,三新工程原验收与Git已实际完成。本次仅收尾文档/驱动提交与普通增量GitAtlas推送,不重复产品测试、发行或重启。封存后人类更正Claude代理Key,接着另冻当前修订、原文本/工具两预检最多两请求;旧两mismatch/目录缺项及全部历史unknown保留,不回退用户配置、不读Key、不重跑三工程/Purge。NRO七公开副本/原索引只读核对,来源不回信。最终QA在docs/QA-M4-LIVE-BASELINE.md。
|
||||
|
||||
I27/I28推送完成:源码与验收44b6a09b已普通增量推送至既有GitAtlas分支,远端HEAD和重新fetch的14个提交逐SHA一致,见[回执](docs/gitatlas-m2-ci-push.json)。回执/完成接续最后另提交并再次核对;最终4a03ec52干净1222/1222及7595封存原件保持,正式未部署/重启,runner/job/真实供应商及历史收益unknown不代签。下一按docs/M3-RELEASE-PLAN.md推进实际schema2迁移与新结果后的兼容回退;下方进行中说明按原时点理解。
|
||||
|
||||
I27/I28最终干净轮完成:4a03ec52、146文件1222/1222,0失败/取消/跳过/todo;实际Edge/隐藏Electron7组4图全部实看,原宿主/Electron/记录TCP再次结束拒绝,1376源码/29编译不变。原首1194/1211与下载/辅助/视觉失败保持,新7595份150586505字节封存;五旧根21297份7184046930字节只读exact。验收/机器记录/AGENTS等入口正在限定提交和普通GitAtlas增量推送,不将本地通过代签runner/job;M3计划70d1d8c7、Mnemo/国内入口1a9c51b1/3d649b3b仅规划/只读,不操作正式/Key/模型。2001操作规模性能和历史unknown保持,下一M3实际迁移与保留新结果的回退。
|
||||
|
||||
I27/I28接续收尾:三处Python/推理测试修正d5207dc6已提交并专项23/23;运行时工具广告与已有任务执行契约对齐、研究采用当前开放默认,三个实际文件22/22及真实tool-prefix-observation4/4分别有证据。原首Windows全套1211项1194过17败自然结束,原件保持;提醒并发首失败无最后状态,不能补造唯一根因。最新实际Edge/隐藏Electron7组4图逐张实看通过,第二轮原生PNG滞后不接受为视觉通过,另记字体/两个RAF/原尺寸及真实TCP拒绝证据。当前核改动/限定提交后重新真正npm ci/新Python QA/Electron安装/构建/全套,不将现有专项称最终干净CI或远端job;0真实模型/Key/正式写入,Mnemo只读单独登记,M3迁移增量/结果后回退和历史unknown保持。
|
||||
|
||||
880c88c7已提交最小回放,独立contract已真npm ci73.3s/build18.1s、来源/工作流通过且14指定回归运行中;原Windows全套仍在大记录/Python等阶段。复核冒烟清理从health缺失误推端口关闭,已补真实TCP监听观察和同页事件去重,直接工具8/8,最终记录器专项和实际两端尚待;不提前代签全套/远端。Mnemo最终七原件SHA及五组离线摘要匹配,只读反馈补充,未接通服务或联合验收。
|
||||
|
||||
CI最小回放夹具65份2452465字节完成:112原来源/Git blob前后一致,四engine投影和R原证明/六回执/请求与旧输入相同。三原套件加CI漂移/缺失/逃逸边界35/35,源码包仍排除完整历史档案/会话/凭据;新CI导出后先核夹具。原e75a6427全套正在大记录/Python等阶段,首失败照存;最终干净安装仍待,不以专项或本机结果称远端job通过。
|
||||
|
||||
CI首四个旧断言已核对:Messages原生块整组保留且完整往返、Orven标题、input/context容量分开、目录绑定实际请求/凭据修订;三文件专项41/41,原e75a6427全套继续收集。engine-context另因包外历史输入加载失败,model-reservation/original-purge同类依赖将改成最小带来源夹具;不删测试、不复制用户数据、不改原历史或正式服务。Mnemo新增规范先登记只读,不扩大本轮权限。
|
||||
|
||||
CI镜像/UTF8源码36c6be4f,新增URL边界pipeline6/6。新e75a6427已真正npm ci65.9s/独立venv51.7s/官方锁定checksums的显式公共镜像Electron26.6s/build6.6s及来源/工作流通过,全部146测试文件正在单轮运行。旧官方下载主动终止失败原件、未知远端执行器保持;未跳浏览器/测试、未读取真实Key或正式写入。
|
||||
|
||||
0b4f1de4第二干净run止于官方Electron暂存0字节,552652ms后原pid3284/父60820/程序/UTC核对后主动结束,exit1与interrupted=false原字段保持;所有已过npm/venv/来源/工作流不冒称全套通过。公共官方及镜像HEAD/首范围确可读,完整下载仍待。显式公共镜像/官方checksum、Python UTF8及URL凭据拒绝已补,下一新run。Mnemo并发SHA变更被拒绝,本地只读反馈另存/不接入,主任务CI继续。
|
||||
|
||||
CI下载/上传修复0b4f1de4限定提交9输入,更新pipeline5/5。原Chromium实际625374ms/0正文,先两次身份条件安全拒绝,后从原JSON UTC与真实pid/父/参数/0.003436秒创建差核对才结束自有子树;原runner exit1和interrupted=false字段照存,另标外部人为结束,不能归成网络自然退出。新962be523从0b4f1de4再次真正干净npm/venv/显式Electron/Edge/全套运行中,正式和旧封存不动。
|
||||
|
||||
af42b949首干净run止于非产品必需Chromium正文0字节,HEAD成功不代替下载。自有62420/64192身份核对后结束下载子树,原npm/Python通过和Chromium失败分别保存;后续改为产品真实Edge/隐藏Electron实测,缺浏览器仍失败。显式Electron过滤环境/原日志与隐藏证据上传补强一起另提交。产品src与历史fixture未修改,不重复旧实验或正式操作。
|
||||
|
||||
af42b949干净run的隐式Electron实际已成功,正在独立cache下载Chromium197.5MiB;原run按自身冻结输入继续。ROOT显式electron-install改动未传入原run,需另提交/新run验证,不补造原下载PID/日志。仅源码stat脏标.gitattributes无实际diff;刷新索引另核对,未还原内容。下一关注编译/全套实际失败及两端冒烟,历史unknown不动。
|
||||
|
||||
CI源码af42b949已限定提交20输入;来源a43262c1。首提交空行检查拒绝后误启动的preflight失败保持。干净run b499f71d已真正npm ci(38s)/锁依赖树/来源/工作流、Python3.11.15新venv及七wheel/pytest8.4.2/pip25.3全过;Electron路径getter触发隐式未独立记日志下载,已定位,下一显式过滤环境下载并另固定提交/新run。无模型/Key/正式操作,原11/11不代签新干净运行。
|
||||
|
||||
I27/I28源码阶段最终专项11/11、官方四来源清单/26记录/14回归输入与workflow真正YAML检查通过;首10/11和两个来源辅助失败保持。工作流只打包精确ci.yml,不遍历.gitea其它配置。实现从已提交HEAD导出/新依赖/venv/原日志/源码及编译前后核对,缺环境/真实跳过非零退出,实际桌面冒烟已准备。限定提交后执行独立干净Windows全套,正式/真实模型/Key/旧工程不动。
|
||||
|
||||
I27/I28来源阶段本地提交完成:14输入逐Git/原字节与已知凭据签名核对,7官方CI来源/4 action ref/7 PyPI wheel固定。官方契约完整44来源/26记录/14回归文件首离线通过;快照首同级引用、原source-package LF与.gitattributes文本CRLF失败保持。新专项首10/11,9新CI/官方检查通过,源码ZIP fixture补明确LF后单独复验。统一入口和实际桌面/干净环境验证继续,尚未远端执行/发行/正式写入。
|
||||
|
||||
I27/I28阶段1来源固定:7份CI官方正文/action元数据、4 action逐commit的ls-remote、7个PyPI通用wheel精确SHA完成。Python准备器首Colorama标签与第二轮误认为并行响应已齐的失败保留,第三轮只补缺文件,未污染外部Python。Mnemo指定v1 README SHA一致,已写本地只读字段/接入依赖反馈,当前不接入/请求/回复源线程或混入CI提交。下一实现统一入口、官方规格差异与缺环境非零退出。
|
||||
|
||||
I27/I28调查2完成:基线tree cf362e31固定,GitAtlas匿名版本/仓库API成功,1.26.2且Actions启用,runner/runs均401;不提取账户token或安装runner。取得对应版本三官方正文,原重定向/失败完整记录。几次调查误猜测试/浏览器模块名和PowerShell通配符路径已保留为读取错误,后续只用rg实际文件列表;产品未修改。下一固定来源并实现可重复检查入口与官方更新回归。
|
||||
|
||||
I27/I28阶段1开始:470d6948干净,已读AGENTS/采用计划/1.25计划/路线和现有构建、源码打包、Python/官方注册表检查;规划catchup及固定Node身份核对。新证据根独立创建,尚未请求真实模型/Key/正式服务或GitAtlas API。接续输出过大截断、PowerShell未引用tree revision的只读调查错误已记入新计划;不重做I09/I11。
|
||||
|
||||
I09/I11阶段1–5完成:71f18055/9e1feeeb/ee97aaae及验收1a897576已普通增量推送、远端逐SHA一致;5126原件/880输入/29编译/20图与22Git输入保持,原四根16171项只读exact。当前只提交回执和完成接续再普通推送核对;下一I27/I28 CI,然后M3迁移增量/新结果回退。不部署/重启/模型/Key/旧实验或原封存写入,历史unknown保持。
|
||||
|
||||
I09/I11 UI源码ee97aaae已限定提交,22变更冻结输入/Git原字节复核通过;5126份2008555583字节封存,MANIFEST c887364e54fe3f68e5f432155db87517fb8413966f5786156bedceddadbf2217,旧四根16171项只读exact,自有进程及记录非正式端口监听0。当前验收/机器记录/接续入口提交及普通GitAtlas增量推送收尾;新封存根只读,后续写独立close根。一次接续文档patch误用Tianshu标题整体未应用,按实际标题重做,不追加原封存。
|
||||
|
||||
I09/I11实际Electron独立8组/10图完成:5本机请求、原metadata/cipher/snapshot/history exact、合成Claude凭据重载、1工具不重放;窗口保持隐藏,原生capturePage,正常关闭并独立宿主重启,三配色/1280及390和外部/页面错误0。已复看网页绿/紫及Electron宽表单/窄上下文,真实字体与结构可读;最终源/界面审查、限定提交与封存收尾,正式未写入/重启。
|
||||
|
||||
I09/I11 web-ui-5完成8组/10图,5本机请求及6账本、所有原metadata/cipher/snapshot/history exact、1工具未重放;外部页面/错误0,三配色/1280及390通过。最终880输入/300源码/178后端/213测试/29编译/74runtime首身份复核通过,规范化0项;source-5为16/16,原127/129加两文件26/26按SHA覆盖129(103未变复用)。实际Electron独立运行中;随后限定UI/回归提交及最终封存,不作正式部署/缓存/Purge收益声明。
|
||||
|
||||
I09/I11 web-ui-4在手机任务抽屉打开后直接点后方工具箱,被可见检视区截住,4组/4请求的原失败与failure-center保留。open辅助先按用户Escape返回,再打开模型中心;不force点击后方元素,不改产品布局。最终按.gitattributes只规范化本批变更文本CRLF为LF、保留原冻结字节和来源正文,另冻build-5验最终源码/桌面。
|
||||
|
||||
I09/I11 build-3/candidate-3为880输入/29编译/74runtime,相关两文件26/26通过。web-ui-3通过4组,原第4次确实自动重选B,因驱动误期待A修改头而失败;保留4请求/原图/原日志。最新驱动分别验证自动B和UI手选A(新增一请求),不修改产品选线;最后使用build-4冻结验收输入及不变candidate-3,产品编译输入逐SHA复用。实际宽窄两图已复看,最终两端/三配色和原件保持仍待。
|
||||
|
||||
I09/I11接续已读原计划/未提交diff并取web-ui-2结果:3组通过后窄屏截图等待失败(3本机请求),candidate-backend-1为127/129。核源码确认两项为旧亲和假设、检视跨手机边界会关闭已有wide状态;按显式run/关闭不暗存更新旧测试,验收器先设390再打开真实任务入口。候选状态汇总改显示连接数,避免跨连接重复加原状态。两次文档patch使用错误findings/progress标题,均整体未应用,第三次按真实锚点完成;原错误保留。下一另冻build-3,只重验受改测试文件和两端核心路径,复用需逐SHA核对的未变后端结果。
|
||||
|
||||
I09/I11独立build-1通过:880输入、29编译、74runtime逐复制SHA;相关候选回归进行中。web-ui-1在第一个旧池编辑的精确标签定位失败,实看1280截图和原DOM已确认下拉存在且显示原Run范围;嵌套select选项参与label文本,新增与可见文案一致的aria-label后另冻build-2。0模型请求,原失败截图/文本/日志保留。旧metadata、密文和快照尚待最终两端验收,不抢先宣称exact。
|
||||
|
||||
I09/I11阶段2核心完成:source-5最终16/16、typecheck-2退出0。纯Chat状态逐成员投影已通过相同型号/不同凭据、实际停用/改头边界;原生块TTL/工具元数据、JSON/SSE签名/加密reasoning/phase、跨Run/Store与Provider重启、手选、同Task停止恢复和optional保存故障保持。原source-1/2/3/4失败原件不动。核心限定本地提交后独立构建与相关回归/两端UI继续;未发行或触正式服务。
|
||||
|
||||
I09/I11 source-4完整新15项为14过1失败:纯Chat池只换型号、误用首成员推理状态已定位并修复,新增相同型号/不同凭据的独立投影边界,下一source-5不覆盖原轮。原生JSON/SSE及phase/cache块、停止恢复不重放、unknown亲和和可选保存故障通过。额外只读调查误猜两个.d.mts路径以及rg参数放在--后均保留为工具使用错误,无产品或正式数据改变。
|
||||
|
||||
I09/I11来源提交前git diff --check发现官方原文6处原有尾空白,来源字节/SHA保持不修剪;限定排除这两份官方正文后核对其它新增文件,后续另核Git blob原字节。不是产品检查失败,不改原来源来制造通过。
|
||||
|
||||
I09/I11阶段1来源复核完成:8份官方来源固定,SOURCE.json SHA ced3703c4afb174c5b939ffb3dd8572f22ff8e5f3fb8489f627f45880315b799;OpenAI完整Docs原件复用、Claude/DeepSeek六份逐SHA一致。接续再次打开当前两篇页面,原source-1/2/3失败保留,未重跑旧轮次;首次mkdir既有目录拒绝后改为新文件名写入。下一完成核心与用户连接保持范围及确定性边界,限定来源提交先行。
|
||||
|
||||
I09/I11接续复核(2026-11-07):沿bdfeb29c与已有未提交实现继续,planning catchup退出0且无补遗;已核AGENTS最新范围、采用计划、1.25计划与M2第五批计划。工作树含缓存块/身份/亲和核心及新测试,界面/来源封存/验证/提交尚未完成,不把简短接续摘要当验收。初次工作区根AGENTS不存在、过宽清单输出截断和假定CI目录不存在仅属只读调查错误;以后限定Agent-Workbench真实路径和输出范围。当前不委派/真实模型/Key/正式写入或重启,先核未提交输入再完成本批。
|
||||
|
||||
I09/I11阶段1开始:确认Messages转换丢系统/内容/工具缓存注解、亲和仅在Run;先保存当前OpenAI官方缓存/会话原件并复用I15 Claude/DeepSeek来源。planning catchup退出0/无补遗。初读误猜model-settings/model-native-settings文件名及PowerShell rg glob、过量来源结果输出仅属调查错误,后续改为真实文件列表/摘要输出。Mnemo新增准确报告/固定计划SHA与SSE partial只读资料另保留,不替代工作台验收;主任务继续。
|
||||
|
||||
Mnemo新调用方预算/状态同步已仅作只读登记:源方新版手机短3/3、900秒49/60发起/48客户端确认/p95约48.4/125新鲜采样仍失败;后查49服务端原任务成功不补成客户端窗口通过。47无网络响应与8离线/过期、分段事件清理后的延迟unknown分别保留。Harness2048→23→1/额外4096预留、修后2048与默认压缩65536/standard身份缺项分开;最新启动不能创建会话、0模型请求,14公开检查与独立边界失败分别记录,完整读改测/恢复未通过。固定旧计划SHA一致,明确reviews路径未找到。见[只读同步](docs/integration/MNEMO-CALLER-BUDGET-STATE-READONLY-2026-11-07.md)。工作台M2下一I09/I11/CI及M3顺序、开放默认/模型偏好/历史unknown保持,无真实模型/正式写入/委派或向源线程回复。
|
||||
|
||||
I14/I12本批阶段5完成:3207份1498866416字节已封存,manifest 3963ad9e02d3f62c1ae4d5fefcb59fdc0bcdc63682be1c72109542a60a2b408c;封存后全3207原件/871工作输入只读复核,35改变输入Git原字节与17当前文档链接通过,2个文档仅规范化行尾、产品输入未变。来源337d1bc9、核心f95507ab、界面26472d28及验收04c0556c已普通增量推送至既有GitAtlas feat/task-agreement-1.25,远端逐SHA一致。回执和完成接续最后另提交/推送再核对,不递归改回执。下一I09/I11缓存块/跨Run状态与I27/I28 CI,再M3真实schema2迁移增量/新处理器结果后回退;未发行/正式操作/真实模型,W/AD/AE/AG与重复读/R/W边界/Purge收益unknown保持。
|
||||
@@ -2123,3 +2199,33 @@ New API研究阶段一234b76cc已提交;阶段二逐篇305正文/201接口全
|
||||
审查组装首轮因不存在的protocols图像锚点失败,原assemble-round-1.log保留,改用真实input-attachments后第二轮通过;收尾O21合成元数据发现凭据登记101上界,不读取Key,第三轮38事项/30区域/22观察记录(21编号+纠正)/715输入SHA通过。已生成全层报告、官方来源账本/机器矩阵和M0–M8路线,38项验收全部映射;七当前文档/AGENTS/规划入口已统一。正封存.raw及四最终交付输入,验证后限定本地提交和既有GitAtlas分支普通增量推送。
|
||||
|
||||
审查封存和交付核对首轮通过:73份2365463字节原件清单a6119c607bfbbc96a876afd499298356c4a79404e47acfa47acec9bc18b33b9d、4最终报告/矩阵/来源/路线SHA、715输入、38事项与无环依赖/9阶段、89新文档和当前入口链接通过;0产品源码变更。报告统计已在封存前按实际计数8P0/25P1/5P2纠正,未保留错误总计。48c9824474ab7f753b4855c592bcd46e9ad36650已普通增量推送,ls-remote逐SHA一致,远端从4da2550b前进。回执和本完成状态另提交后核对最终HEAD,不重跑模型/产品测试或发行。
|
||||
|
||||
## 2026-11-11 1.25.13 M3+B4b 正式交付
|
||||
当前用户明确要求开发、提交、验收。按 docs/M3-B4B-RELEASE-1.25.13.md 推进:复核→相关验证/构建→产品提交→独立源码/干净重建/程序与安装器→实际安装/迁移/网页和隐藏Electron→正式升级/验收/普通GitAtlas推送。覆盖上一阶段不提交/不发行安排;原历史证据、异常、真实质量和Purge unknown保持。新证据 .qa/m3-b4b-release-2026-11-11/,旧封存只读。
|
||||
|
||||
2026-11-11 初始证据脚本将中文file URL误作路径,首执行ENOENT且未写快照;已改为fileURLToPath,错误保留在工具记录。
|
||||
2026-11-11 构建首轮失败:npm子命令PATH命中Node16;构建驱动现前插固定Node24目录,保留build-first.log后另轮构建。当前正式只读health为1.25.12/PID63892,尚未激活新版。
|
||||
2026-11-11 阶段2完成:Node24 build-second通过;13相关文件固定并发2,186/186、316246.0644ms、0跳过取消。版本1.25.13四入口(package/lock两处/diagnostics)一致,正式未激活。接下来限定提交76文件→已提交源码独立clean npm ci/重建→实际包/安装/UI。
|
||||
2026-11-11 首包生成后登记因打包器回写许可证而失败,未安装;修为输出目录生成,补YAML许可。packaging相关4/4通过,产品186项对应源码未改;首失败保留,最终独立证据另用-r2。
|
||||
2026-11-11 最终发行35a6d6d9,1399输入/30编译独立clean npm ci exact;真实程序/Setup登记通过,打包未改工作树。合成Mnemo后端24/24与安装副本网页4组/10图通过;网页发送回执1280/来源390两图实看。跨版迁移、Electron及Inno生命周期继续。
|
||||
2026-11-11 实际发行7/7跨版迁移和保留新结果恢复、停止fresh restart继续账本通过;Mnemo安装副本网页/Electron各4组/10图,基础文件9/9,实际八图查看。旧基础驱动遗漏Orven文件入口的首失败已保留,改为真实工具入口后通过,仅验收脚本变化无产品编译变化。新合成Mnemo宿主及临时私有文件已正常关闭/撤销;Inno首干净安装已通过,后续生命周期仍在执行。
|
||||
2026-11-11 实际Inno生命周期11/11、550937ms通过。正式首快照因旧窗口存在退出;核对PID/创建时间/路径后正常关闭,日期对象误比较原观察保留。第二快照102项目/127会话/schema1。唯一正式Setup351789ms返回0,原宿主63892和instance保持,1.25.13已激活;QA随后公开视图/导出/序号旧断言三轮失败均留原件,没有重装或重放模型。新增规格/默认契约/能力逐原registry重算,完整账本374条/100预览验证;127会话的两项新MCP计数对原行核验后原导出摘要exact。精确迁移守卫核对marker/序号等新增,原8741记录/22085事件及outbox/5244工件和公开配置全保持。
|
||||
2026-11-11 正式网页/实际Electron宽窄四图均实看,final再次原数据exact/9229关闭;新版桌面正常恢复登记PID71480,后台仍63892。782份811582693字节选择性原件清单611c4c40封存后逐件核验、旧M3/B4b清单原字节exact。报告和身份已生成,限定验收提交/普通GitAtlas增量推送收尾。0真实模型/Key提取重存、无原任务续跑;Mnemo默认关闭,真实生产绑定/长期/语义/物理手机/Purge收益unknown保持。
|
||||
|
||||
2026-11-11 1.25.13正式交付推送完成:验收1e309700及打包修复35a6d6d9已普通增量推送至既有GitAtlas分支,远端与重新fetch逐SHA一致;docs/gitatlas-mnemo-1.25.13-push.json随完成接续另提交并核最终HEAD。正式后台仍63892/1.25.13,新版桌面已恢复;782封存原件保持、0真实模型/Key提取重存,下一M4新真实基线,Mnemo生产绑定与真实收益未知保持。
|
||||
|
||||
2026-11-11 M4开始:读取规范/两主计划/官方路线/规划尾部及AD/AE参考;catchup退出0无未同步报告。首次公开API观察保存.qa/m4-live-baseline-2026-11-11/observation.json,正式63892/1.25.13空闲,六连接无凭据;异步请求界面恢复,0真实请求。读取期间猜测的validate.mjs及顶层model-hub/ledger路径不存在,已改用rg文件清单定位,不修改旧驱动或证据。新M4计划已写,阶段2准备进行中。
|
||||
|
||||
2026-11-11 M4:用户已恢复三模型,公开确认DeepSeek20/Claude8/Responses15及三credentialRevision1;原选择仍Claude单成员池。新驱动改用完整公开账本、Messages精确必填字段和原错误/取消/owner全部index核对。首隔离驱动m4-driver-a01e6b39完成8正常预检及1原HTTP错误,主动取消案例原账本为interrupted/unknown,验证器错误期望failed而退出,0真实模型、Node阶段未执行;保留首原件,仅修准确状态断言后另新根复验。
|
||||
|
||||
2026-11-11 M4第二隔离驱动m4-driver-16e746a7:10预检/错误/取消通过,固定Node实际5请求/840合成tokens、原9/9/build/限定Git与owner全部index完成,但收尾误把legacy-default公开revision(跟随configRevision,由2到4)要求原值而失败,原结果不改绿。已按准确configRevision增量逐字段核对公开legacy投影,真实连接所有定义仍要求exact;不改产品默认、原检查或模型参数,另新根验证。只读AD索引首组装发现两检查日志描述符未列入旧session预览,改按实际缺项登记,未补读整个数据库;r2两个原报告按SHA原字节复制,46允许文件/2缺项、原in_progress和终端Git保持,0新工程/模型/Git写入或回信。
|
||||
|
||||
2026-11-11 M4第三隔离驱动m4-driver-490968ae完整通过:10预检边界、5固定模型工程请求/840合成tokens,原9/9/build/限定Git和owner全部index、准确legacy公开修订恢复通过,正式1.25.13处理器的隔离宿主已正常关闭。随后复核原件捕获发现JSON字符分页不能保存非法UTF8原字节,驱动改用现有download=1原字节通路,追加非法UTF8夹具并验证最终捕获路径;不改产品解析或协议实现,前三原根保持。
|
||||
|
||||
2026-11-11 M4最终驱动m4-driver-40fb93c8:11边界(含非法UTF8原字节)和Node固定模型5请求/840合成tokens通过;原9/9/build/限定Gitdb0840a3、owner全部index和真实配置恢复通过,三协议/原错误/中断unknown原件完整。docs/m4-driver-validation-2026-11-11.json逐四根结果SHA/最终9脚本/正式server bundle记录,0真实模型,未重打包或正式重启。只读来源r2已生成46允许文件/两原报告/两个日志描述符缺项,不补签v2、原in_progress或旧失败。阶段2限定提交后启动真实冻结。
|
||||
|
||||
2026-11-11 M4真实兼容六新请求完成,三DNS/TLS诊断通过且零HTTP/模型请求;两协议passed/matched,Claude两mismatch和目录缺项保留。新Node十请求180463tokens/0unknown,84451ms、原9/9/build/三源码Git3cd994ff22c1658cbe4c26a7e317ce6b7b314745,owner全部原index保持;内置delivery1d2a6e2a completed及最后原报告completed,三早晚报告分别捕获。首审计node-4e580c59遇外部你好一会话/一请求而失败;显式只读记录其ID/每行SHA/七工件/4301tokens/mismatch后,第二node-e3388cf2因不存在的ledger.project_id失败,未改数据。按实际sessions关系修驱动,第三node-2e02c1a0完整原53527行/127导出/5244工件/377请求和公开配置、选择恢复通过,config124→126/registry176→178等已声明增量准确重算,9229关闭。阶段文档生成;未修改产品源码或重启1.25.13,下一独立React。
|
||||
|
||||
2026-11-11 M4 React一次完成:session36f7bbd7/Taskdddcbdf2/Run c41c7af7、14请求271333tokens/0unknown、72040ms、原九测试/typecheck/build/仅四源码Gitc55644a37016c772a9ac8e4d9a98dbc12a6c1223与全部保护字节/index通过。实际Edge157宽窄六交互、2PNG均view_image查看,视觉记录原SHA保存;模型两浏览器作业、终端原Git证据与最终报告verified_committed分别保持。额外只读二进制工件检查0,原采集源码/回执留ancillary且未采纳驱动。react-e49f5eb3审计一次过,原53527行/127导出/5244工件/377请求保持,累计本批30请求与外部1请求单列,原选择/六连接/池恢复和9229关闭。仅摘要驱动增加报告/页面原来源字段,无产品变更或重启。Purge预备只读定位曾误读不存在的live-ag/run与freeze,已由rg清单确认live-ag/initial及live-af通用驱动,未执行旧入口。下一阶段限定提交后Python。
|
||||
|
||||
2026-11-11 M4 Python一次成功:session26b24e0c/Task875afb7b/Run c430c404,10请求176083tokens/0unknown/48767ms、原10pytest/compileall/三源码Gitfa2ce16a2e6bd372b0754637df84979b413a09ad,原owner全部index保持。两组早晚JSON/Markdown原字节、终端verified与最终verified_committed分别保存;机器记录docs/m4-python-2026-11-11.json生成。python-fc99c576一次审计,原53527行/127导出/5244工件/377请求和新九份捕获导出exact,累计40模型请求与外部1分列;六连接/池/选择/PID63892/instance保持,config124→130/registry176→182及前插/诊断/序列明确复算,9229关闭。仅audit新增本批导出保持核对;sealmjs语法检查通过,实际最终封存仍待。无产品改变、发行或重启。
|
||||
|
||||
@@ -0,0 +1,13 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import { createHash } from 'node:crypto';
|
||||
import { REQUEST_LEDGER_MIGRATION } from '../src/core/storage/request-ledger-schema.mjs';
|
||||
|
||||
export function writeDataCompatibility(root) {
|
||||
const describe = name => { const bytes = fs.readFileSync(path.join(root, name)); return { path: name, bytes: bytes.length, sha256: createHash('sha256').update(bytes).digest('hex') }; };
|
||||
const declaration = { schemaVersion: 1, application: 'agent-workbench-database', readWriteSchemas: [1, 2], ledgerMigration: REQUEST_LEDGER_MIGRATION,
|
||||
sourceInputs: ['src/core/models/request-ledger.mjs', 'src/core/storage/state-database.mjs', 'src/core/storage/request-ledger-schema.mjs'].map(describe),
|
||||
bundles: ['dist-server/server.mjs', 'dist-server/hot-upgrade.mjs'].map(describe) };
|
||||
fs.writeFileSync(path.join(root, 'dist-server/data-compatibility.json'), JSON.stringify(declaration, null, 2) + '\n');
|
||||
return declaration;
|
||||
}
|
||||
@@ -1,5 +1,6 @@
|
||||
import { build } from 'esbuild';
|
||||
import fs from 'node:fs';
|
||||
import { writeDataCompatibility } from './build-data-compatibility.mjs';
|
||||
// Bundled CommonJS dependencies (QR PNG encoder) still import Node builtins.
|
||||
const nodeRequire = { js: "import { createRequire as __awbCreateRequire } from 'node:module'; const require = __awbCreateRequire(import.meta.url);" };
|
||||
fs.mkdirSync('dist-server', { recursive: true });
|
||||
@@ -16,4 +17,5 @@ await build({ entryPoints: ['src/desktop/main.mjs'], outfile: 'dist-desktop/main
|
||||
platform: 'node', target: 'node24', format: 'cjs', sourcemap: false, legalComments: 'eof' });
|
||||
await build({ entryPoints: ['src/desktop/preload.cjs'], outfile: 'dist-desktop/preload.cjs', bundle: true, external: ['electron'],
|
||||
platform: 'node', target: 'node24', format: 'cjs', sourcemap: false, legalComments: 'eof' });
|
||||
console.log('Server bundle built.');
|
||||
writeDataCompatibility(process.cwd());
|
||||
console.log('Server bundle and database compatibility declaration built.');
|
||||
|
||||
@@ -0,0 +1,111 @@
|
||||
import assert from 'node:assert/strict';
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import net from 'node:net';
|
||||
import { randomUUID } from 'node:crypto';
|
||||
import { chromium, _electron } from 'playwright';
|
||||
import { HostController } from '../src/core/desktop/host-controller.mjs';
|
||||
import { observeWindowsProcess } from '../src/core/desktop/process-identity.mjs';
|
||||
import { desktopExecutableIdentity } from '../src/core/desktop/app-installation.mjs';
|
||||
import { projectEnvironment } from '../src/core/process-environment.mjs';
|
||||
import { browserInstallation } from '../src/core/browser/browser-driver.mjs';
|
||||
import { openWorkbenchTool } from './ui-navigation.mjs';
|
||||
import { fileDigest, writeNewJson, loopbackPortStatus } from './ci-utils.mjs';
|
||||
|
||||
const args=process.argv.slice(2);
|
||||
if(args.includes('--help')) {console.log('node scripts/ci-desktop-smoke.mjs --evidence NEW_DIRECTORY — actual product Microsoft Edge and hidden Electron through locked Playwright against an owned new host/project. No model, external project, formal configuration or installation.');process.exit(0);}
|
||||
if(args.length!==2 || args[0]!=='--evidence')throw new Error('Use --help');
|
||||
assert.equal(process.platform,'win32');
|
||||
const appRoot=process.cwd(),evidence=path.resolve(args[1]);fs.mkdirSync(evidence);
|
||||
const ownedRoot=path.join(appRoot,'.qa','ci-smoke-'+randomUUID().slice(0,8)),dataRoot=path.join(ownedRoot,'data'),projectRoot=path.join(ownedRoot,'project');fs.mkdirSync(projectRoot,{recursive:true});
|
||||
const original=Buffer.from('# Owned CI project\n\nSource retained while actual web and Electron entry paths are exercised.\n');
|
||||
fs.writeFileSync(path.join(projectRoot,'README.md'),original,{flag:'wx'});
|
||||
const reservation=net.createServer();await new Promise(resolve=>reservation.listen(0,'127.0.0.1',resolve));const port=reservation.address().port;await new Promise(resolve=>reservation.close(resolve));
|
||||
const controller=new HostController({appRoot,dataRoot,port,nodePath:process.execPath,timeoutMs:60000});
|
||||
const report={schemaVersion:1,passed:false,startedAt:new Date().toISOString(),ownedRoot,port,groups:[],screenshots:[],pageErrors:[],externalRequests:[],scope:'Actual isolated product Microsoft Edge/Electron through locked Playwright and owned source/data only; no model or formal installation.'};
|
||||
let browser,application,electronPid,hostIdentity,activePage,stage='host-start';
|
||||
const record=(name,detail={})=>report.groups.push({name,passed:true,...detail});
|
||||
const watched=new WeakSet();
|
||||
const watch=(page)=>{activePage=page;if(watched.has(page))return;watched.add(page);page.on('pageerror',error=>report.pageErrors.push(error.message));page.on('request',request=>{const url=new URL(request.url());if(['http:','https:'].includes(url.protocol) && !(url.hostname==='127.0.0.1' && Number(url.port)===port))report.externalRequests.push(request.url());});};
|
||||
async function capture(page,mode,width,height){
|
||||
const id=mode+'-'+width,file=path.join(evidence,id+'.png');
|
||||
if(mode==='electron'){
|
||||
await application.evaluate(({BrowserWindow},size)=>BrowserWindow.getAllWindows()[0].setContentSize(...size),[width,height]);
|
||||
await page.waitForFunction(size=>innerWidth===size[0] && innerHeight===size[1],[width,height]);
|
||||
}else await page.setViewportSize({width,height});
|
||||
const dialog=page.getByRole('dialog',{name:'模型中心',exact:true});
|
||||
await dialog.getByRole('heading',{name:'新任务默认模型',exact:true}).waitFor({timeout:30000});
|
||||
await dialog.getByRole('combobox',{name:'新任务默认模型',exact:true}).and(page.locator('select:enabled')).waitFor({timeout:30000});
|
||||
const painted=await dialog.evaluate(async element=>{
|
||||
await document.fonts.ready;
|
||||
await Promise.all(element.getAnimations({subtree:true}).filter(animation=>animation.effect?.getTiming().iterations!==Infinity).map(animation=>animation.finished.catch(()=>{})));
|
||||
const frames=[];await new Promise(resolve=>requestAnimationFrame(first=>{frames.push(first);requestAnimationFrame(second=>{frames.push(second);resolve();});}));
|
||||
return{fontStatus:document.fonts.status,frames};
|
||||
});
|
||||
assert.equal(painted.fontStatus,'loaded');assert.ok(painted.frames[1]>painted.frames[0]);
|
||||
if(mode==='electron'){
|
||||
const value=await application.evaluate(async({BrowserWindow})=>{const window=BrowserWindow.getAllWindows()[0];const [width,height]=window.getContentSize();const image=await window.capturePage({x:0,y:0,width,height},{stayHidden:true,stayAwake:true});return{base64:image.toPNG().toString('base64'),visible:window.isVisible(),size:image.getSize()};});
|
||||
assert.equal(value.visible,false);assert.equal(value.size.width,width);assert.equal(value.size.height,height);fs.writeFileSync(file,Buffer.from(value.base64,'base64'),{flag:'wx'});
|
||||
}else await page.screenshot({path:file,fullPage:false});
|
||||
writeNewJson(path.join(evidence,id+'.dom.json'),{title:await page.title(),width,height,painted,html:await page.content(),overflow:await page.evaluate(()=>document.documentElement.scrollWidth>innerWidth)});
|
||||
assert.equal(await page.evaluate(()=>document.documentElement.scrollWidth>innerWidth),false,'Document overflows');
|
||||
const bytes=fs.readFileSync(file);assert.equal(bytes.subarray(1,4).toString(),'PNG');assert.ok(bytes.length>1000);
|
||||
report.screenshots.push({file,width,height,sha256:fileDigest(file),bytes:bytes.length});
|
||||
}
|
||||
async function userPath(page,mode,projectId){
|
||||
watch(page);await page.locator('.connection.connected').waitFor({timeout:30000});assert.match(await page.title(),/Orven/i);
|
||||
await page.getByRole('combobox',{name:'当前项目',exact:true}).selectOption(projectId);
|
||||
record(mode+' project selection and connected build');
|
||||
await openWorkbenchTool(page,'打开模型中心');await page.getByRole('dialog',{name:'模型中心',exact:true}).waitFor();
|
||||
await capture(page,mode,1280,900);await page.keyboard.press('Escape');
|
||||
if(mode==='electron')await application.evaluate(({BrowserWindow})=>BrowserWindow.getAllWindows()[0].setContentSize(390,844));else await page.setViewportSize({width:390,height:844});
|
||||
await openWorkbenchTool(page,'打开模型中心');await page.getByRole('dialog',{name:'模型中心',exact:true}).waitFor();
|
||||
await capture(page,mode,390,844);await page.keyboard.press('Escape');
|
||||
record(mode+' actual model center at 1280 and 390');
|
||||
}
|
||||
try{
|
||||
hostIdentity=(await controller.start()).identity;report.host={pid:hostIdentity.pid,createdAt:hostIdentity.createdAt};
|
||||
const project=await controller.api('/projects',{method:'POST',input:{root:projectRoot}}),session=await controller.api('/sessions',{method:'POST',input:{projectId:project.id}});
|
||||
record('owned new host/project/session',{projectId:project.id,sessionId:session.id});
|
||||
stage='product-edge';const edge=browserInstallation(appRoot);assert.equal(edge.status,'available','Product Microsoft Edge is required');report.edge=edge;browser=await chromium.launch({executablePath:edge.executable.path,headless:true});report.edgeVersion=browser.version();
|
||||
const page=await browser.newPage({viewport:{width:1280,height:900}});watch(page);await page.goto(controller.url);await userPath(page,'web',project.id);await browser.close();browser=null;
|
||||
stage='electron';const executable=desktopExecutableIdentity(appRoot).path;
|
||||
application=await _electron.launch({executablePath:executable,args:[path.join(appRoot,'dist-desktop/main.cjs'),'--app-root',appRoot,'--data-root',dataRoot,'--node',process.execPath,'--port',String(port),'--hidden-window'],cwd:appRoot,env:projectEnvironment(),timeout:60000});
|
||||
electronPid=await application.evaluate(()=>process.pid);report.electron={pid:electronPid,executable,sha256:fileDigest(executable),version:await application.evaluate(()=>process.versions.electron)};
|
||||
const window=await application.firstWindow();
|
||||
// Keep this owned QA window hidden while giving its renderer foreground
|
||||
// frame scheduling. This affects capture only and ends with the process.
|
||||
await application.evaluate(({BrowserWindow})=>BrowserWindow.getAllWindows()[0].webContents.setBackgroundThrottling(false));
|
||||
report.electron.capturePolicy={hiddenWindow:true,backgroundThrottling:false,requireLoadedModelCard:true,waitForFontsAndTwoFrames:true};
|
||||
await userPath(window,'electron',project.id);
|
||||
const status=await controller.status();assert.equal(status.identity.pid,hostIdentity.pid);assert.ok(status.desktopWindows.entries.some(entry=>entry.pid===electronPid));
|
||||
record('same owned host and actual registered Electron');
|
||||
const center=await controller.api('/model-center');assert.equal(center.attempts.length,0,'No model request is permitted in CI smoke');
|
||||
assert.deepEqual(fs.readFileSync(path.join(projectRoot,'README.md')),original);assert.deepEqual(report.pageErrors,[]);assert.deepEqual(report.externalRequests,[]);
|
||||
record('original source retained; zero model attempts or outside page requests');
|
||||
report.passed=true;
|
||||
}catch(error){
|
||||
report.failure={stage,name:error.name,message:error.message,stack:error.stack};process.exitCode=1;
|
||||
if(activePage&&!activePage.isClosed()){
|
||||
try{
|
||||
const domFile=path.join(evidence,'failure-'+stage+'.html'),imageFile=path.join(evidence,'failure-'+stage+'.png');
|
||||
fs.writeFileSync(domFile,await activePage.content(),{flag:'wx'});
|
||||
if(stage==='electron'&&application){
|
||||
const base64=await application.evaluate(async({BrowserWindow})=>{const window=BrowserWindow.getAllWindows()[0];const [width,height]=window.getContentSize();return(await window.capturePage({x:0,y:0,width,height},{stayHidden:true,stayAwake:true})).toPNG().toString('base64');});
|
||||
fs.writeFileSync(imageFile,Buffer.from(base64,'base64'),{flag:'wx'});
|
||||
}else await activePage.screenshot({path:imageFile,fullPage:true,timeout:10000});
|
||||
report.failureCapture={url:activePage.url(),dom:{file:domFile,bytes:fs.statSync(domFile).size,sha256:fileDigest(domFile)},image:{file:imageFile,bytes:fs.statSync(imageFile).size,sha256:fileDigest(imageFile)}};
|
||||
}catch(captureError){report.failureCaptureError={message:captureError.message};}
|
||||
}
|
||||
}
|
||||
finally{
|
||||
try{if(browser)await browser.close();if(application)await application.close();
|
||||
if(await controller.health())report.hostStop=await controller.stop();
|
||||
const deadline=Date.now()+15000;while(electronPid && observeWindowsProcess(electronPid) && Date.now()<deadline)await new Promise(resolve=>setTimeout(resolve,100));
|
||||
assert.equal(electronPid?!!observeWindowsProcess(electronPid):false,false,'Owned Electron did not exit');assert.equal(await controller.health(),null,'Owned port still responds');
|
||||
report.portObservation=await loopbackPortStatus(port);assert.equal(report.portObservation.status,'refused','Recorded CI port still accepts TCP connections');
|
||||
report.cleanup={hostExited:hostIdentity?observeWindowsProcess(hostIdentity.pid)===null:true,electronExited:electronPid?observeWindowsProcess(electronPid)===null:true,portClosed:report.portObservation.status==='refused'};
|
||||
assert.equal(report.cleanup.hostExited,true);
|
||||
}catch(error){report.passed=false;report.cleanupFailure={message:error.message,stack:error.stack};process.exitCode=1;}
|
||||
report.finishedAt=new Date().toISOString();writeNewJson(path.join(evidence,'result.json'),report);console.log(JSON.stringify({passed:report.passed,groups:report.groups.length,screenshots:report.screenshots.length,report:path.join(evidence,'result.json'),failure:report.failure?.message,cleanupFailure:report.cleanupFailure?.message}));
|
||||
}
|
||||
@@ -0,0 +1,21 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"node": "24.22.1",
|
||||
"npm": "11.19.1",
|
||||
"python": "3.11.15",
|
||||
"pytest": "8.4.2",
|
||||
"pip": "25.3",
|
||||
"playwright": "1.64.1",
|
||||
"electron": "45.2.0",
|
||||
"runnerLabel": "awb-windows-x64",
|
||||
"profiles": {
|
||||
"contract": "Pinned official sources, registry/version/deprecation regressions and full application build. No Windows/real-provider claim.",
|
||||
"windows": "Fresh locked dependencies, full source tests without skips, actual product Microsoft Edge/Electron through locked Playwright and owned installation regressions. No formal installation or model request."
|
||||
},
|
||||
"actions": {
|
||||
"checkout": "https://gitea.com/actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683",
|
||||
"setupNode": "https://github.com/actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020",
|
||||
"setupPython": "https://github.com/actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065",
|
||||
"uploadArtifact": "https://gitea.com/actions/upload-artifact@ff15f0306b3f739f7b6fd43fb5d26cd321bd4de5"
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,8 @@
|
||||
# Python 3.11.15 CI QA only. Exact universal wheels from PyPI; use --require-hashes --only-binary=:all:.
|
||||
pytest==8.4.2 --hash=sha256:872f880de3fc3a5bdc88a11b39c9710c3497a547cfa9320bc3c5e62fbf272e79
|
||||
iniconfig==2.3.0 --hash=sha256:f631c04d2c48c52b84d0d0549c99ff3859c98df65b3101406327ecc7d53fbf12
|
||||
packaging==25.0 --hash=sha256:29572ef2b1f17581046b3a2227d5c611fb25ec70ca1ba8554b24b0e69331a484
|
||||
pluggy==1.6.0 --hash=sha256:e920276dd6813095e9377c0bc5566d94c932c33b27a3e3945d8389c374dd4746
|
||||
Pygments==2.19.2 --hash=sha256:86540386c03d588bb81d44bc3928634ff26449851e99741617ecb9037ee5ec0b
|
||||
colorama==0.4.6 --hash=sha256:4f1d9991f5acc0ca119f9d443620b77f9d6b33703e51011c16baf57afb285fc6
|
||||
pip==25.3 --hash=sha256:9655943313a94722b7774661c21049070f6bbb0a1516bf02f7c8d5d9201514cd
|
||||
@@ -0,0 +1,150 @@
|
||||
import assert from 'node:assert/strict';
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import net from 'node:net';
|
||||
import { createHash, randomUUID } from 'node:crypto';
|
||||
import { spawn } from 'node:child_process';
|
||||
import { readSourceZip } from './package-source.mjs';
|
||||
|
||||
export const digest = bytes => createHash('sha256').update(bytes).digest('hex');
|
||||
export const fileDigest = file => digest(fs.readFileSync(file));
|
||||
export const writeNewJson = (file, value) => fs.writeFileSync(file, JSON.stringify(value, null, 2) + '\n', { flag: 'wx' });
|
||||
export function binaryMirror(value) {
|
||||
if (!value) return null;
|
||||
const url = new URL(value);
|
||||
assert.ok(url.protocol === 'https:' && !url.username && !url.password && !url.search && !url.hash && url.pathname.endsWith('/'), 'Binary mirror must be an explicit HTTPS directory without credentials.');
|
||||
return url.href;
|
||||
}
|
||||
|
||||
export async function loopbackPortStatus(port) {
|
||||
assert.ok(Number.isSafeInteger(port) && port > 0 && port <= 65535, 'Recorded CI port required');
|
||||
return new Promise((resolve, reject) => {
|
||||
const socket = net.createConnection({ host: '127.0.0.1', port });
|
||||
let settled = false;
|
||||
const finish = (error, status) => {
|
||||
if (settled) return; settled = true; clearTimeout(timer); socket.destroy();
|
||||
if (error) reject(error); else resolve({ host: '127.0.0.1', port, status });
|
||||
};
|
||||
const timer = setTimeout(() => finish(new Error('Recorded CI port observation timed out')), 5000);
|
||||
socket.once('connect', () => finish(null, 'accepting'));
|
||||
socket.once('error', error => error.code === 'ECONNREFUSED' ? finish(null, 'refused') : finish(error));
|
||||
});
|
||||
}
|
||||
export function createCiRun(outputRoot, profile) {
|
||||
assert.ok(['contract', 'windows'].includes(profile), 'Unknown CI profile');
|
||||
const root = path.resolve(outputRoot), run = path.join(root, `run-${profile}-${Date.now().toString(36)}-${randomUUID().slice(0,8)}`);
|
||||
fs.mkdirSync(root, { recursive: true }); fs.mkdirSync(run);
|
||||
for (const directory of ['evidence', 'source', 'cache', 'tmp']) fs.mkdirSync(path.join(run, directory));
|
||||
return { root: run, evidence: path.join(run, 'evidence'), source: path.join(run, 'source'), cache: path.join(run, 'cache'), tmp: path.join(run, 'tmp') };
|
||||
}
|
||||
|
||||
// Read only named OS variables. Never enumerate credential values or forward
|
||||
// model settings, npm user config, provider tokens or external-engine paths.
|
||||
export function ciEnvironment(run, input = process.env, node = process.execPath) {
|
||||
const allowed = ['PATH','PATHEXT','SystemRoot','SystemDrive','WINDIR','ComSpec','TEMP','TMP','USERPROFILE','HOMEDRIVE','HOMEPATH','APPDATA','LOCALAPPDATA','ProgramFiles','ProgramFiles(x86)','CommonProgramFiles','HOME','LANG','LC_ALL'];
|
||||
const env = {};
|
||||
for (const key of Object.keys(input)) if (allowed.some(name=>name.toLowerCase() === key.toLowerCase())) env[key] = input[key];
|
||||
const inherited = Object.keys(env).find(key=>key.toLowerCase()==='path');
|
||||
const originalPath = inherited ? env[inherited] : '';
|
||||
if (inherited) delete env[inherited];
|
||||
env.PATH = [path.dirname(node), originalPath].filter(Boolean).join(path.delimiter);
|
||||
env.TEMP = run.tmp; env.TMP = run.tmp; env.CI = 'true'; env.PYTEST_DISABLE_PLUGIN_AUTOLOAD = '1';
|
||||
const userConfig = path.join(run.root, 'npm-user.conf'), globalConfig = path.join(run.root, 'npm-global.conf');
|
||||
for (const file of [userConfig, globalConfig]) if (!fs.existsSync(file)) fs.writeFileSync(file, '', { flag: 'wx' });
|
||||
env.npm_config_userconfig = userConfig; env.npm_config_globalconfig = globalConfig;
|
||||
env.npm_config_registry = 'https://registry.npmjs.org/'; env.npm_config_cache = path.join(run.cache, 'npm');
|
||||
env.npm_config_audit = 'false'; env.npm_config_fund = 'false'; env.npm_config_update_notifier = 'false';
|
||||
env.PLAYWRIGHT_BROWSERS_PATH = path.join(run.cache, 'playwright');
|
||||
env.electron_config_cache = path.join(run.cache, 'electron');
|
||||
return env;
|
||||
}
|
||||
|
||||
export async function loggedCommand({ id, executable, args = [], cwd, env, evidence, timeoutMs = 0, check = true }) {
|
||||
assert.match(id, /^[a-z0-9-]+$/);
|
||||
const stdoutFile = path.join(evidence, `${id}.stdout.log`), stderrFile = path.join(evidence, `${id}.stderr.log`);
|
||||
let stdoutFd = fs.openSync(stdoutFile, 'wx'), stderrFd = fs.openSync(stderrFile, 'wx');
|
||||
const startedAt = new Date().toISOString(), started = Date.now();
|
||||
let child, timer, interrupted = false, spawnError;
|
||||
const terminate = () => {
|
||||
interrupted = true;
|
||||
if (!child?.pid) return;
|
||||
if (process.platform === 'win32') {
|
||||
const stop = spawn('taskkill.exe', ['/PID', String(child.pid), '/T', '/F'], { windowsHide: true, stdio: 'ignore' });
|
||||
stop.on('error', () => { try { child.kill(); } catch {} });
|
||||
} else child.kill('SIGTERM');
|
||||
};
|
||||
const onInterrupt = () => terminate();
|
||||
process.once('SIGINT', onInterrupt); process.once('SIGTERM', onInterrupt);
|
||||
try {
|
||||
const result = await new Promise(resolve => {
|
||||
child = spawn(executable, args, { cwd, env, windowsHide: true, stdio: ['ignore', stdoutFd, stderrFd] });
|
||||
child.once('spawn', () => writeNewJson(path.join(evidence, `${id}.started.json`), { id, executable, args, cwd, pid: child.pid, parentPid: process.pid, startedAt, environmentNames: Object.keys(env).sort() }));
|
||||
child.once('error', error => { spawnError = { name: error.name, message: error.message, code: error.code }; });
|
||||
child.once('close', (exitCode, signal) => resolve({ exitCode, signal }));
|
||||
if (timeoutMs > 0) { timer = setTimeout(terminate, timeoutMs); timer.unref(); }
|
||||
});
|
||||
fs.closeSync(stdoutFd); stdoutFd = -1; fs.closeSync(stderrFd); stderrFd = -1;
|
||||
const record = { id, executable, args, cwd, pid: child.pid ?? null, startedAt, finishedAt: new Date().toISOString(), durationMs: Date.now() - started, ...result, interrupted, spawnError: spawnError ?? null,
|
||||
stdout: { file: stdoutFile, bytes: fs.statSync(stdoutFile).size, sha256: fileDigest(stdoutFile) }, stderr: { file: stderrFile, bytes: fs.statSync(stderrFile).size, sha256: fileDigest(stderrFile) } };
|
||||
writeNewJson(path.join(evidence, `${id}.json`), record);
|
||||
console.log(JSON.stringify({ command: id, exitCode: record.exitCode, interrupted, durationMs: record.durationMs, evidence }));
|
||||
if (check && (record.exitCode !== 0 || record.signal || interrupted || spawnError)) throw Object.assign(new Error(`CI command failed: ${id}. Original logs: ${stdoutFile} and ${stderrFile}`), { record });
|
||||
return record;
|
||||
} finally {
|
||||
clearTimeout(timer); process.removeListener('SIGINT', onInterrupt); process.removeListener('SIGTERM', onInterrupt);
|
||||
// Spawn errors still produce original records; handles are closed above.
|
||||
for (const fd of [stdoutFd, stderrFd]) if (fd >= 0) try { fs.closeSync(fd); } catch {}
|
||||
}
|
||||
}
|
||||
|
||||
export function unpackVerifiedSource(archive, destination) {
|
||||
assert.ok(archive.kind === 'source-code' && /^[a-f0-9]{40}$/.test(archive.sourceCommit));
|
||||
assert.equal(fileDigest(archive.path), archive.sha256, 'Source ZIP SHA changed');
|
||||
const entries = readSourceZip(archive.path), prefix = `Agent-Workbench-${archive.version}-source/`;
|
||||
assert.equal(entries.size, archive.files.length, 'Unexpected source archive entries');
|
||||
const seen = new Set();
|
||||
for (const file of archive.files) {
|
||||
assert.ok(!seen.has(file.path), 'Duplicate source path'); seen.add(file.path);
|
||||
assert.ok(!file.path.includes('\\') && !file.path.startsWith('/') && !file.path.split('/').some(segment=>segment==='..' || segment.includes(':')), 'Invalid source archive path');
|
||||
const output = path.resolve(destination, file.path), bytes = entries.get(prefix + file.path);
|
||||
assert.ok(output.startsWith(path.resolve(destination) + path.sep));
|
||||
assert.ok(Buffer.isBuffer(bytes), `Missing archived input ${file.path}`);
|
||||
assert.equal(bytes.length, file.bytes, file.path); assert.equal(digest(bytes), file.sha256, file.path);
|
||||
fs.mkdirSync(path.dirname(output), { recursive: true }); fs.writeFileSync(output, bytes, { flag: 'wx' });
|
||||
}
|
||||
return verifySourceInputs(destination, archive.files);
|
||||
}
|
||||
|
||||
export function verifySourceInputs(root, files) {
|
||||
for (const file of files) {
|
||||
const full = path.join(root, file.path); assert.ok(fs.lstatSync(full).isFile() && !fs.lstatSync(full).isSymbolicLink(), file.path);
|
||||
assert.equal(fileDigest(full), file.sha256, `Source input changed: ${file.path}`);
|
||||
}
|
||||
return { passed: true, files: files.length, identity: 'exported ZIP bytes after committed .gitattributes, before and after actual dependency install/build/tests' };
|
||||
}
|
||||
|
||||
export function builtFiles(root) {
|
||||
const directories = ['dist-client','dist-server','dist-desktop'], files = [];
|
||||
const walk = (relative) => {
|
||||
for (const entry of fs.readdirSync(path.join(root, relative), { withFileTypes: true }).sort((a,b)=>a.name.localeCompare(b.name))) {
|
||||
assert.equal(entry.isSymbolicLink(), false, 'Build output link');
|
||||
const current = path.join(relative, entry.name);
|
||||
if (entry.isDirectory()) walk(current);
|
||||
else { assert.ok(entry.isFile()); const full = path.join(root, current); files.push({ file: current.split(path.sep).join('/'), bytes: fs.statSync(full).size, sha256: fileDigest(full) }); }
|
||||
}
|
||||
};
|
||||
for (const directory of directories) { assert.ok(fs.statSync(path.join(root,directory)).isDirectory(), `Missing ${directory}`); walk(directory); }
|
||||
return files;
|
||||
}
|
||||
|
||||
export function tapTotals(text) {
|
||||
const result = {};
|
||||
for (const match of text.matchAll(/^# (tests|suites|pass|fail|cancelled|skipped|todo|duration_ms) (\d+(?:\.\d+)?)\s*$/gm)) result[match[1]] = Number(match[2]);
|
||||
assert.ok(result.tests > 0 && Number.isInteger(result.pass), 'Missing full Node TAP totals');
|
||||
return result;
|
||||
}
|
||||
|
||||
export function requireCompleteTests(totals) {
|
||||
for (const name of ['fail','cancelled','skipped','todo']) assert.equal(totals[name], 0, `CI ${name} is not a verified pass`);
|
||||
assert.equal(totals.pass, totals.tests, 'CI not every collected test passed');
|
||||
}
|
||||
+125
@@ -0,0 +1,125 @@
|
||||
import assert from 'node:assert/strict';
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import { execFileSync } from 'node:child_process';
|
||||
import { createRequire } from 'node:module';
|
||||
import { fileURLToPath, pathToFileURL } from 'node:url';
|
||||
import { packageSource } from './package-source.mjs';
|
||||
import { verifyOfficialContracts } from './official-contract-check.mjs';
|
||||
import { createCiRun, ciEnvironment, loggedCommand, unpackVerifiedSource, verifySourceInputs, builtFiles, tapTotals, requireCompleteTests, fileDigest, writeNewJson, binaryMirror } from './ci-utils.mjs';
|
||||
|
||||
const root = path.resolve(path.dirname(fileURLToPath(import.meta.url)), '..');
|
||||
const args = process.argv.slice(2), options = { profile: 'contract', outputRoot: path.join(root, '.qa/ci'), python: process.env.AGENT_WORKBENCH_TEST_PYTHON };
|
||||
if (args.includes('--help')) {
|
||||
console.log('node scripts/ci.mjs [--profile contract|windows] [--python ABSOLUTE_BASE_PYTHON] [--output-root DIRECTORY] [--electron-mirror HTTPS_DIRECTORY]\nExports committed Git source to a new directory; performs a real npm ci, official contract/build/test checks and Windows smoke. No current node_modules, data, keys or engine installation are reused. Each invocation preserves its own logs and failures. An explicit binary mirror uses the same locked official checksums.');
|
||||
process.exit(0);
|
||||
}
|
||||
for (let i=0;i<args.length;i++) {
|
||||
const name = { '--profile':'profile', '--python':'python', '--output-root':'outputRoot', '--electron-mirror':'electronMirror' }[args[i]];
|
||||
if (!name || !args[i+1]) throw new Error('Unsupported CI argument; use --help');
|
||||
options[name] = args[++i];
|
||||
}
|
||||
assert.ok(['contract','windows'].includes(options.profile));
|
||||
const run = createCiRun(options.outputRoot, options.profile), env = ciEnvironment(run);
|
||||
const report = { schemaVersion: 1, profile: options.profile, startedAt: new Date().toISOString(), passed: false, root: run.root, evidence: run.evidence,
|
||||
commands: [], scope: { realModelRequests: 0, realKeysRead: 0, formalDataOrConfigWrites: 0, deployment: false, remoteRunClaim: false },
|
||||
limitations: ['Pinned source and controlled local checks do not prove live-provider compatibility, model delivery or Purge/cache gains.', 'GitAtlas runner and remote results must be independently confirmed.', 'Formal schema2 migration and post-activation rollback remain M3 acceptance.'] };
|
||||
let stage = 'preflight';
|
||||
const command = async (id, executable, argv, cwd = run.source, extraEnv = {}) => {
|
||||
try { const result = await loggedCommand({ id, executable, args: argv, cwd, env: { ...env, ...extraEnv }, evidence: run.evidence }); report.commands.push(result); return result; }
|
||||
catch(error) { if (error.record) report.commands.push(error.record); throw error; }
|
||||
};
|
||||
const nodeCommand = (id, argv, cwd, extraEnv) => command(id, process.execPath, argv, cwd, extraEnv);
|
||||
const readOutput = id => fs.readFileSync(path.join(run.evidence, `${id}.stdout.log`), 'utf8').trim();
|
||||
try {
|
||||
const git = argv => execFileSync('git', argv, { cwd: root, encoding: 'utf8', windowsHide: true, env }).trim();
|
||||
const inputPaths = ['src','public','scripts','tests','.gitea','package.json','package-lock.json','vite.config.ts','tsconfig.json','index.html','mobile.html','.gitattributes'];
|
||||
report.source = { commit: git(['rev-parse','HEAD']), tree: git(['rev-parse','HEAD^{tree}']), branch: git(['branch','--show-current']), worktreeStatus: git(['status','--porcelain=v1']) };
|
||||
assert.equal(git(['diff','--name-only','HEAD','--',...inputPaths]), '', 'Commit program and test changes before CI; it verifies committed source only.');
|
||||
assert.equal(git(['ls-files','--others','--exclude-standard','--',...inputPaths]), '', 'Untracked program or test inputs are not part of committed CI.');
|
||||
const lock = JSON.parse(fs.readFileSync(path.join(root, 'scripts/ci-environment.json')));
|
||||
assert.equal(lock.schemaVersion, 1); assert.equal(process.version, 'v'+lock.node, 'Use the pinned Node runtime');
|
||||
const electronMirror = binaryMirror(options.electronMirror);
|
||||
if (options.profile === 'windows') assert.ok(process.platform === 'win32' && process.arch === 'x64', 'Windows profile requires actual Windows x64');
|
||||
const npmCli = path.join(path.dirname(process.execPath), 'node_modules/npm/bin/npm-cli.js');
|
||||
assert.ok(fs.statSync(npmCli).isFile(), 'Use Node with its bundled npm');
|
||||
await nodeCommand('npm-version', [npmCli,'--version'], run.tmp); assert.equal(readOutput('npm-version'), lock.npm, 'Pinned npm version required');
|
||||
report.environment = { node: { version: process.version, executable: process.execPath, sha256: fileDigest(process.execPath) }, npm: { version: lock.npm, cli: npmCli, sha256: fileDigest(npmCli) }, platform: process.platform, arch: process.arch, ciLockSha256: fileDigest(path.join(root,'scripts/ci-environment.json')) };
|
||||
if (options.profile === 'windows') {
|
||||
assert.ok(options.python && path.isAbsolute(options.python) && fs.statSync(options.python).isFile(), '--python or AGENT_WORKBENCH_TEST_PYTHON must name the absolute base QA Python.');
|
||||
await command('python-base', options.python, ['-I','-X','utf8','-c','import sys,json; print(json.dumps({"version":sys.version.split()[0],"executable":sys.executable,"baseExecutable":sys._base_executable,"basePrefix":sys.base_prefix}))'], run.tmp);
|
||||
const python = JSON.parse(readOutput('python-base')); assert.equal(python.version, lock.python, 'Pinned base Python version required');
|
||||
report.environment.pythonBase = { ...python, sha256: fileDigest(options.python) };
|
||||
}
|
||||
stage = 'committed-source-export';
|
||||
const archive = packageSource({ root, outputRoot: run.evidence });
|
||||
writeNewJson(path.join(run.evidence,'source-archive.json'), archive);
|
||||
report.source = { ...report.source, archiveSha256: archive.sha256, archiveFile: archive.path, packageLockSha256: archive.files.find(file=>file.path==='package-lock.json').sha256, files: archive.files.length };
|
||||
report.sourceBefore = unpackVerifiedSource(archive, run.source);
|
||||
assert.equal(fs.existsSync(path.join(run.source,'node_modules')), false, 'Fresh source must not have dependencies');
|
||||
writeNewJson(path.join(run.evidence,'source-before.json'), report.sourceBefore);
|
||||
stage = 'replay-fixtures';
|
||||
await nodeCommand('replay-fixtures', ['scripts/verify-replay-fixtures.mjs',path.join(run.evidence,'replay-fixtures-result.json')]);
|
||||
report.replayFixtures = JSON.parse(readOutput('replay-fixtures'));
|
||||
stage = 'official-contracts';
|
||||
await nodeCommand('official-registry', ['scripts/verify-official-model-registry.mjs',path.join(run.evidence,'official-registry-result.json')]);
|
||||
await nodeCommand('official-contracts', ['scripts/verify-official-contracts.mjs','--output',path.join(run.evidence,'official-contracts-result.json')]);
|
||||
const contracts = verifyOfficialContracts(run.source); report.contracts = { snapshotSha256: contracts.snapshotSha256, registryVersion: contracts.registryVersion, requestDefaultsVersion: contracts.requestDefaultsVersion, sources: contracts.sourceInventory.reduce((sum,item)=>sum+item.sources.length,0), records: contracts.recordsCount };
|
||||
stage = 'fresh-npm-ci';
|
||||
await nodeCommand('npm-ci', [npmCli,'ci','--no-audit','--no-fund']);
|
||||
assert.equal(fs.lstatSync(path.join(run.source,'node_modules')).isSymbolicLink(), false, 'No node_modules junction reuse');
|
||||
const packageLock = JSON.parse(fs.readFileSync(path.join(run.source,'package-lock.json'))), installed = {};
|
||||
for (const name of ['playwright','playwright-core','electron','typescript','vite','esbuild','yaml']) {
|
||||
const file = path.join(run.source,'node_modules',name,'package.json'), actual = JSON.parse(fs.readFileSync(file));
|
||||
assert.equal(actual.version, packageLock.packages['node_modules/'+name].version, name+' differs from lock');
|
||||
installed[name] = { version: actual.version, sha256: fileDigest(file) };
|
||||
}
|
||||
assert.equal(installed.playwright.version,lock.playwright); assert.equal(installed.electron.version,lock.electron);
|
||||
report.environment.packages = installed;
|
||||
await nodeCommand('dependency-tree', [npmCli,'ls','--all','--json']);
|
||||
stage = 'workflow-contract';
|
||||
await nodeCommand('workflow-check', ['scripts/verify-ci-workflow.mjs',path.join(run.evidence,'workflow.json')]);
|
||||
if (options.profile === 'windows') {
|
||||
stage = 'fresh-python-qa';
|
||||
const venv = path.join(run.root,'python-qa');
|
||||
await command('python-venv',options.python,['-I','-X','utf8','-m','venv',venv],run.tmp);
|
||||
const python = path.join(venv,'Scripts/python.exe'); env.AGENT_WORKBENCH_TEST_PYTHON = python;
|
||||
await command('python-install',python,['-I','-X','utf8','-m','pip','--isolated','install','--disable-pip-version-check','--no-input','--require-hashes','--only-binary=:all:','--index-url','https://pypi.org/simple','--cache-dir',path.join(run.cache,'pip'),'-r',path.join(run.source,'scripts/ci-python-requirements.txt')],run.tmp);
|
||||
await command('python-packages',python,['-I','-X','utf8','-c','import sys,pytest,pip,json; print(json.dumps({"python":sys.version.split()[0],"pytest":pytest.__version__,"pip":pip.__version__,"executable":sys.executable,"prefix":sys.prefix,"basePrefix":sys.base_prefix}))'],run.tmp);
|
||||
const pythonInfo = JSON.parse(readOutput('python-packages')); assert.equal(pythonInfo.python,lock.python); assert.equal(pythonInfo.pytest,lock.pytest); assert.equal(pythonInfo.pip,lock.pip); assert.notEqual(pythonInfo.prefix,pythonInfo.basePrefix);
|
||||
report.environment.pythonQa = { ...pythonInfo, sha256: fileDigest(python), requirementsSha256: fileDigest(path.join(run.source,'scripts/ci-python-requirements.txt')) };
|
||||
await command('python-pip-check',python,['-I','-X','utf8','-m','pip','--isolated','check'],run.tmp);
|
||||
stage = 'explicit-electron-install';
|
||||
const electronEnv = {};
|
||||
if (electronMirror) {
|
||||
electronEnv.ELECTRON_MIRROR = electronMirror;
|
||||
electronEnv.ELECTRON_CUSTOM_DIR = lock.electron;
|
||||
report.environment.electronDownload = { mirror: electronMirror, directory: lock.electron, explicit: true, checksumSource: 'locked electron/checksums.json', checksumFileSha256: fileDigest(path.join(run.source,'node_modules/electron/checksums.json')) };
|
||||
} else report.environment.electronDownload = { mirror: null, explicit: false, checksumSource: 'locked electron/checksums.json', checksumFileSha256: fileDigest(path.join(run.source,'node_modules/electron/checksums.json')) };
|
||||
await nodeCommand('electron-install',['node_modules/electron/install.js'],run.source,electronEnv);
|
||||
const require = createRequire(path.join(run.source,'package.json'));
|
||||
const electron = require('electron'); assert.ok(fs.statSync(electron).isFile(), 'Actual Electron binary missing');
|
||||
report.environment.electron = { executable: electron, sha256: fileDigest(electron) };
|
||||
const { browserInstallation } = await import(pathToFileURL(path.join(run.source,'src/core/browser/browser-driver.mjs')).href);
|
||||
const edge = browserInstallation(run.source); assert.equal(edge.status,'available','Actual Microsoft Edge is required; do not skip browser tests.'); report.environment.edge = edge;
|
||||
}
|
||||
stage = 'build'; await nodeCommand('build',[npmCli,'run','build']);
|
||||
report.build = builtFiles(run.source); writeNewJson(path.join(run.evidence,'build-inputs.json'),report.build);
|
||||
stage = 'tests';
|
||||
const tests = options.profile === 'windows' ? fs.readdirSync(path.join(run.source,'tests')).filter(name=>name.endsWith('.test.mjs')).sort().map(name=>'tests/'+name) : [...contracts.testFiles,'tests/official-contracts-ci.test.mjs','tests/ci-pipeline.test.mjs','tests/source-package.test.mjs'];
|
||||
const testFiles = [...new Set(tests)].sort();
|
||||
writeNewJson(path.join(run.evidence,'test-inputs.json'),testFiles.map(file=>({file,sha256:fileDigest(path.join(run.source,file))})));
|
||||
const testResult = await loggedCommand({id:'source-tests',executable:process.execPath,args:['--test','--test-concurrency=2','--test-reporter=tap',...testFiles],cwd:run.source,env,evidence:run.evidence,check:false}); report.commands.push(testResult);
|
||||
report.tests = { files: testFiles.length, totals: tapTotals(readOutput('source-tests')), exitCode:testResult.exitCode, interrupted:testResult.interrupted };
|
||||
assert.equal(testResult.exitCode,0,'Original source tests failed; preserve this run and fix the specific failure.'); assert.equal(testResult.interrupted,false); requireCompleteTests(report.tests.totals);
|
||||
if (options.profile === 'windows') { stage='actual-browser-desktop-smoke'; await nodeCommand('desktop-smoke',['scripts/ci-desktop-smoke.mjs','--evidence',path.join(run.evidence,'desktop-smoke')]); report.desktopSmoke = JSON.parse(fs.readFileSync(path.join(run.evidence,'desktop-smoke/result.json'))); assert.equal(report.desktopSmoke.passed,true); }
|
||||
stage='final-identity'; report.sourceAfter=verifySourceInputs(run.source,archive.files); assert.deepEqual(builtFiles(run.source),report.build,'Build artifacts changed during tests'); writeNewJson(path.join(run.evidence,'source-after.json'),report.sourceAfter);
|
||||
report.passed=true;
|
||||
} catch(error) {
|
||||
report.failure={stage,name:error.name,message:error.message,stack:error.stack}; process.exitCode=1;
|
||||
} finally {
|
||||
report.finishedAt=new Date().toISOString(); writeNewJson(path.join(run.evidence,'report.json'),report);
|
||||
const logFiles = fs.readdirSync(run.evidence).filter(name=>fs.statSync(path.join(run.evidence,name)).isFile()).sort();
|
||||
writeNewJson(path.join(run.evidence,'MANIFEST.json'),{schemaVersion:1,files:logFiles.map(file=>({file,bytes:fs.statSync(path.join(run.evidence,file)).size,sha256:fileDigest(path.join(run.evidence,file))})),scope:'Top-level immutable command logs and reports; source/dependencies are separate owned execution inputs.'});
|
||||
console.log(JSON.stringify({passed:report.passed,profile:options.profile,source:report.source?.commit,tests:report.tests,stage:report.failure?.stage,error:report.failure?.message,report:path.join(run.evidence,'report.json')}));
|
||||
}
|
||||
@@ -0,0 +1,12 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import { desktopExecutableIdentity } from '../src/core/desktop/app-installation.mjs';
|
||||
import { samePath } from '../src/core/desktop/file-state.mjs';
|
||||
|
||||
export function desktopLaunchCommand(appRoot, args) {
|
||||
const root = fs.realpathSync.native(appRoot), executablePath = desktopExecutableIdentity(root).path;
|
||||
// Packaged Electron loads resources/app itself; only the default Electron
|
||||
// runtime takes the compiled main program as a positional argument.
|
||||
const packaged = samePath(executablePath, path.join(root, 'desktop/AgentWorkbench.exe'));
|
||||
return { executablePath, args: [...(packaged ? [] : [path.join(root, 'dist-desktop/main.cjs')]), ...args] };
|
||||
}
|
||||
@@ -0,0 +1,230 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { randomUUID } from 'node:crypto';
|
||||
import { execFileSync } from 'node:child_process';
|
||||
import { DatabaseSync } from 'node:sqlite';
|
||||
import { client, sessionAttempts, artifactBytes, connectionDefinition, expectedOutputFields, sha } from './m4-live/common.mjs';
|
||||
import { invokeProbe, inspectorClosed } from './m4-live/inspector.mjs';
|
||||
import { publicSnapshot, digest } from './retained-host-state.mjs';
|
||||
|
||||
export const root = path.resolve('.qa/m4-claude-correction-2026-11-12');
|
||||
const origin = 'http://127.0.0.1:4318';
|
||||
const connectionId = '9ae348a5-d0a6-42ed-8664-0da5f9effe34';
|
||||
const load = relative => JSON.parse(fs.readFileSync(path.join(root, relative), 'utf8'));
|
||||
function save(relative, value) {
|
||||
const file = path.resolve(root, relative); assert.ok(file.startsWith(root + path.sep));
|
||||
fs.mkdirSync(path.dirname(file), { recursive: true });
|
||||
fs.writeFileSync(file, typeof value === 'string' || Buffer.isBuffer(value) ? value : JSON.stringify(value, null, 2) + '\n', { flag: 'wx' });
|
||||
}
|
||||
const git = args => execFileSync('git', args, { encoding: 'utf8', windowsHide: true }).trim();
|
||||
function source(relative) {
|
||||
const bytes = fs.readFileSync(path.join(root, relative)); return { path: relative, bytes: bytes.length, sha256: sha(bytes) };
|
||||
}
|
||||
function requireOpen(config) {
|
||||
for (const [key, expected] of Object.entries({ executionMode: 'unrestricted', capacityMode: 'passthrough', approvalMode: 'full', maxSteps: 0, maxRunTokens: 0, maxTaskTokens: 0 })) assert.equal(config[key], expected, key);
|
||||
}
|
||||
export function sseFrames(raw) {
|
||||
return raw.split(/\r?\n\r?\n/).flatMap(packet => {
|
||||
const data = packet.split(/\r?\n/).filter(line => line.startsWith('data:')).map(line => line.slice(5).trimStart()).join('\n');
|
||||
if (!data || data === '[DONE]') return [];
|
||||
try { return [{ value: JSON.parse(data) }]; } catch { return [{ invalidJsonSha256: sha(data) }]; }
|
||||
});
|
||||
}
|
||||
export function assessPreflight({ input, invoked, attempt, requests, body, raw, toolEvents = 0 }) {
|
||||
const outputFields = body ? Object.fromEntries(['max_tokens', 'max_output_tokens', 'max_completion_tokens'].filter(key => Object.hasOwn(body, key)).map(key => [key, body[key]])) : null;
|
||||
const frames = sseFrames(raw), starts = frames.filter(frame => frame.value?.type === 'message_start');
|
||||
const modelFields = starts.map(frame => frame.value.message?.model ?? null);
|
||||
const stopReasons = frames.filter(frame => frame.value?.type === 'message_delta').map(frame => frame.value.delta?.stop_reason).filter(value => value != null);
|
||||
const rawComplete = starts.length === 1 && frames.some(frame => frame.value?.type === 'message_stop') && frames.every(frame => !frame.invalidJsonSha256);
|
||||
const usageComplete = attempt?.usageStatus === 'reported' && frames.some(frame => frame.value?.message?.usage) && frames.some(frame => frame.value?.usage);
|
||||
const streaming = { requested: body?.stream === true, sseObserved: /^(?:event:|data:)/m.test(raw), messageStartCount: starts.length,
|
||||
messageStopCount: frames.filter(frame => frame.value?.type === 'message_stop').length, invalidJsonFrames: frames.filter(frame => frame.invalidJsonSha256).length,
|
||||
rawComplete, usageComplete, complete: rawComplete && attempt?.modelIdentity?.complete === true };
|
||||
const wirePassed = !!body && body.model === input.connection.model && JSON.stringify(outputFields) === JSON.stringify(input.expectedOutput.fields) && body.stream === true && streaming.sseObserved;
|
||||
const passed = invoked?.result.contentPassed === true && invoked.result.actualToolsExecuted === 0 && toolEvents === 0 && requests === 1 && wirePassed && streaming.complete && usageComplete && attempt?.status === 'completed'
|
||||
&& attempt?.modelIdentity?.status === 'matched' && modelFields.length === 1 && modelFields[0] === input.connection.model;
|
||||
return { passed, wirePassed, streaming, modelFields, stopReasons, outputFields,
|
||||
rawUsageFrames: frames.filter(frame => frame.value?.usage || frame.value?.message?.usage).map(frame => ({ type: frame.value.type, usage: frame.value.usage ?? frame.value.message.usage })) };
|
||||
}
|
||||
async function capture(api, sessionId, prefix) {
|
||||
const state = await api('/sessions/' + sessionId), exported = await api('/sessions/' + sessionId + '/export'), ledger = await sessionAttempts(api, sessionId);
|
||||
save(prefix + '/session.json', state); save(prefix + '/export.json', exported); save(prefix + '/attempts.json', ledger.rows); save(prefix + '/ledger-references.json', ledger);
|
||||
assert.ok(ledger.total <= 1, 'Each original case has at most one physical request.');
|
||||
const captures = [], ids = new Set();
|
||||
for (const row of ledger.rows) for (const kind of ['request', 'response']) {
|
||||
const id = row[kind + 'ArtifactId']; if (!id) continue;
|
||||
const artifact = state.execution.artifacts.find(value => value.id === id) ?? { id, sha256: row[kind + 'Sha256'] };
|
||||
const bytes = await artifactBytes(api, sessionId, artifact); assert.equal(sha(bytes), row[kind + 'Sha256']);
|
||||
const relative = prefix + '/captures/' + row.id + '-' + kind + (kind === 'request' ? '.json' : row.streaming ? '.sse' : '.txt');
|
||||
save(relative, bytes); captures.push({ attemptId: row.id, artifactId: id, kind, ...source(relative) }); ids.add(id);
|
||||
}
|
||||
for (const artifact of state.execution.artifacts) {
|
||||
if (ids.has(artifact.id) || !/^(text\/|application\/(json|x-ndjson))/.test(artifact.mediaType ?? '')) continue;
|
||||
const bytes = await artifactBytes(api, sessionId, artifact), relative = prefix + '/artifacts/' + artifact.id + (artifact.mediaType === 'application/json' ? '.json' : '.txt');
|
||||
save(relative, bytes); captures.push({ artifactId: artifact.id, kind: artifact.kind, ...source(relative) });
|
||||
}
|
||||
save(prefix + '/artifact-sources.json', captures); return { state, exported, ledger, captures };
|
||||
}
|
||||
async function captureCase(api, scope, prefix, input, invoked) {
|
||||
const captured = await capture(api, input.sessionId, prefix), attempt = captured.ledger.rows[0];
|
||||
const request = captured.captures.find(value => value.attemptId === attempt?.id && value.kind === 'request');
|
||||
const response = captured.captures.find(value => value.attemptId === attempt?.id && value.kind === 'response');
|
||||
const body = request ? load(request.path) : null, raw = response ? fs.readFileSync(path.join(root, response.path), 'utf8') : '';
|
||||
const toolEvents = captured.state.events.filter(value => value.kind === 'tool').length;
|
||||
const run = captured.state.execution.run;
|
||||
const entry = { scope, sessionId: input.sessionId, taskId: invoked?.result.taskId ?? run?.taskId ?? null, runId: invoked?.result.runId ?? run?.id ?? null,
|
||||
attemptId: attempt?.id ?? null, requestKey: attempt?.requestKey ?? null, identity: attempt?.modelIdentity ?? null, status: attempt?.status ?? 'not_dispatched',
|
||||
usageStatus: attempt?.usageStatus ?? 'not_dispatched', usage: attempt?.usage ?? null, cost: attempt?.cost ?? null,
|
||||
costScope: 'Only an explicit ledger cost; absent upstream cost remains null, never inferred from tokens.',
|
||||
requests: captured.ledger.total, actualToolsExecuted: invoked?.result.actualToolsExecuted ?? run?.toolCalls ?? null, toolEvents,
|
||||
diagnostic: invoked?.result.diagnostic ?? attempt?.diagnostic ?? captured.state.session?.lastFailure ?? null,
|
||||
expectedOutput: input.expectedOutput, ...assessPreflight({ input, invoked, attempt, requests: captured.ledger.total, body, raw, toolEvents }),
|
||||
sources: captured.captures, invocationSource: fs.existsSync(path.join(root, scope, 'invocation.json')) ? source(scope + '/invocation.json') : null };
|
||||
save(prefix + '/result.json', entry); return entry;
|
||||
}
|
||||
async function freeze() {
|
||||
assert.equal(fs.existsSync(path.join(root, 'frozen.json')), false); assert.equal(git(['status', '--porcelain']), '', 'Commit validated driver before freezing real requests.');
|
||||
const { api, boot, close } = await client(); try {
|
||||
await inspectorClosed(); requireOpen(boot.config); assert.equal(boot.version, '1.25.13');
|
||||
const host = await api('/host/status'), center = await api('/model-center'); assert.equal(host.activity.active, false);
|
||||
const connection = center.connections.find(value => value.id === connectionId); assert.ok(connection?.hasKey && connection.enabled);
|
||||
assert.equal(connection.model, 'claude-opus-5-5'); assert.equal(connection.protocol, 'anthropic-messages');
|
||||
const baseline = await publicSnapshot(origin, host.identity.dataRoot); assert.equal(baseline.active.length, 0);
|
||||
const old = JSON.parse(fs.readFileSync('.qa/m4-live-baseline-2026-11-11/audits/final-bfadc758-619b-4288-b516-95e461da314d/after.json', 'utf8'));
|
||||
const oldConnection = old.modelCenter.connections.find(value => value.id === connectionId);
|
||||
const definitionChanges = Object.keys(connectionDefinition(connection)).filter(key => digest({ value: connectionDefinition(connection)[key] }) !== digest({ value: connectionDefinition(oldConnection)[key] }));
|
||||
const observed = load('observation.json');
|
||||
save('frozen.json', { at: new Date().toISOString(), host, config: boot.config, center, connection, baseline, originalProjects: boot.projects,
|
||||
expectedOutput: expectedOutputFields(connection), driverCommit: git(['rev-parse', 'HEAD']), driverSha256: sha(fs.readFileSync('scripts/m4-claude-correction.mjs')),
|
||||
postSealPublicChanges: { configRevision: [old.modelCenter.connections.find(value => value.id === 'legacy-default').revision, boot.config.configRevision],
|
||||
registryRevision: [old.modelCenter.revision, center.revision], selectedConnectionRevision: [oldConnection.revision, connection.revision],
|
||||
credentialRevision: [oldConnection.credentialRevision, connection.credentialRevision], definitionChangedFields: definitionChanges,
|
||||
addedConnections: center.connections.filter(value => !old.modelCenter.connections.some(original => original.id === value.id)).map(value => ({ id: value.id, name: value.name, model: value.model, revision: value.revision })),
|
||||
observationRegistryRevision: observed.center.revision, actorOfAdditionalChanges: 'unestablished; preserved as observed; no connection restored' },
|
||||
scopes: ['text', 'tools'], maximumPhysicalRequests: 2, observationMsPerCase: 180000, toolsExecuted: 0, keysReadOrResaved: false, oldM4Replayed: false });
|
||||
save('driver-source.mjs', fs.readFileSync('scripts/m4-claude-correction.mjs'));
|
||||
console.log(JSON.stringify({ frozen: true, pid: host.identity.pid, revision: connection.revision, credentialRevision: connection.credentialRevision, registryRevision: center.revision, selected: center.selected, originalSessions: baseline.database.sessionIds.length, modelRequests: 0 }));
|
||||
} finally { await close(); }
|
||||
}
|
||||
async function run() {
|
||||
const frozen = load('frozen.json'); assert.equal(fs.existsSync(path.join(root, 'claim.json')), false, 'Never redispatch a claimed correction batch; reconcile its saved identities only.');
|
||||
assert.equal(sha(fs.readFileSync('scripts/m4-claude-correction.mjs')), frozen.driverSha256);
|
||||
save('claim.json', { id: randomUUID(), at: new Date().toISOString(), scopes: ['text', 'tools'], maximumPhysicalRequests: 2, noRetry: true });
|
||||
const { api, boot, close } = await client(); try {
|
||||
requireOpen(boot.config); assert.deepEqual(boot.config, frozen.config); await inspectorClosed();
|
||||
const host = await api('/host/status'); assert.equal(host.identity.pid, frozen.host.identity.pid); assert.equal(host.activity.active, false);
|
||||
assert.deepEqual(connectionDefinition((await api('/model-center')).connections.find(value => value.id === connectionId)), connectionDefinition(frozen.connection));
|
||||
const catalogId = randomUUID(); save('catalog-request.json', { id: catalogId, connectionId, revision: frozen.connection.revision });
|
||||
try { save('catalog.json', await api('/model-center/catalog', { id: connectionId, revision: frozen.connection.revision, requestId: catalogId }, 90000)); }
|
||||
catch (error) { save('catalog-error.json', { message: error.message, response: error.response ?? null, noRetry: true }); }
|
||||
finally { save('catalog-cancel.json', await api('/model-center/catalog/cancel', { id: catalogId })); }
|
||||
const projectRoot = path.join(root, 'project'); fs.mkdirSync(projectRoot); fs.writeFileSync(path.join(projectRoot, 'README.md'), 'Separate corrected Claude text/tool observations. No engineering or tool execution.\n', { flag: 'wx' });
|
||||
const project = await api('/projects', { root: projectRoot }); save('project.json', project);
|
||||
const preflights = [];
|
||||
for (const scope of ['text', 'tools']) {
|
||||
assert.deepEqual(connectionDefinition((await api('/model-center')).connections.find(value => value.id === connectionId)), connectionDefinition(frozen.connection));
|
||||
const session = await api('/sessions', { projectId: project.id });
|
||||
const input = { pid: frozen.host.identity.pid, port: 4318, runtimeRoot: boot.runtimeRoot, projectId: project.id, sessionId: session.id,
|
||||
connection: frozen.connection, scope, nonce: randomUUID(), expectedOutput: frozen.expectedOutput, observationMs: frozen.observationMsPerCase };
|
||||
save(scope + '/input.json', input); let invoked;
|
||||
try { invoked = await invokeProbe(input); save(scope + '/invocation.json', invoked); }
|
||||
catch (error) { save(scope + '/invocation-error.json', { message: error.stack, noRetry: true, reconcileOnly: true }); }
|
||||
const entry = await captureCase(api, scope, scope, input, invoked); preflights.push(entry); await inspectorClosed();
|
||||
console.log(JSON.stringify({ scope, passed: entry.passed, attemptId: entry.attemptId, identity: entry.identity?.status, modelFields: entry.modelFields, usageStatus: entry.usageStatus, tokens: entry.usage?.total_tokens ?? null, toolsExecuted: entry.actualToolsExecuted }));
|
||||
}
|
||||
const catalog = fs.existsSync(path.join(root, 'catalog.json')) ? load('catalog.json') : null;
|
||||
const requests = preflights.reduce((sum, value) => sum + value.requests, 0); assert.ok(requests <= 2);
|
||||
save('result.json', { at: new Date().toISOString(), projectId: project.id, preflights, requests,
|
||||
reportedTotalTokens: preflights.reduce((sum, value) => sum + (value.usage?.total_tokens ?? 0), 0), unknownRequests: preflights.filter(value => value.requests && value.usageStatus !== 'reported').length,
|
||||
catalog: { complete: catalog?.complete === true, models: catalog?.entries?.length ?? null, selectedListed: catalog?.entries?.some(value => value.id === frozen.connection.model) ?? false,
|
||||
source: catalog ? source('catalog.json') : source('catalog-error.json'), evidenceScope: 'Public product catalogue with supplied source metadata; no invented raw HTTP capture.' },
|
||||
selectedConnectionDefinitions: 'Frozen current connection only; no saved preference or credential mutation.', modelWeightsAuthenticated: false,
|
||||
savedConfigurationChanged: false, actualToolsExecuted: 0, engineeringOrPurgeReplayed: false, keysReadOrResaved: false });
|
||||
} catch (error) { save('run-error.json', { message: error.stack, noRetry: true, reconcileOnly: true }); throw error; }
|
||||
finally { await inspectorClosed(); await close(); }
|
||||
}
|
||||
async function reconcile() {
|
||||
const frozen = load('frozen.json'), { api, close } = await client(); const prefix = 'reconcile-' + randomUUID();
|
||||
try {
|
||||
const preflights = [];
|
||||
for (const scope of ['text', 'tools']) {
|
||||
if (!fs.existsSync(path.join(root, scope, 'input.json'))) continue;
|
||||
const input = load(scope + '/input.json'), invoked = fs.existsSync(path.join(root, scope, 'invocation.json')) ? load(scope + '/invocation.json') : null;
|
||||
preflights.push(await captureCase(api, scope, prefix + '/' + scope, input, invoked));
|
||||
}
|
||||
assert.ok(preflights.reduce((sum, row) => sum + row.requests, 0) <= frozen.maximumPhysicalRequests);
|
||||
save(prefix + '/result.json', { readOnly: true, newModelRequests: 0, preflights, originalIdentitiesOnly: true }); console.log(JSON.stringify({ prefix, cases: preflights.length, newModelRequests: 0 }));
|
||||
} finally { await close(); }
|
||||
}
|
||||
async function audit() {
|
||||
const frozen = load('frozen.json'), result = load('result.json'), prefix = 'audit-' + randomUUID(), { api, boot, close } = await client(); let receipt;
|
||||
try {
|
||||
const host = await api('/host/status'); await inspectorClosed(); assert.equal(host.activity.active, false); assert.equal(host.identity.pid, frozen.host.identity.pid); assert.equal(host.identity.instanceId, frozen.host.identity.instanceId);
|
||||
requireOpen(boot.config); assert.deepEqual(boot.config, frozen.config);
|
||||
const after = await publicSnapshot(origin, host.identity.dataRoot); save(prefix + '/after.json', after); const before = frozen.baseline;
|
||||
assert.deepEqual(after.modelCenter.selected, frozen.center.selected); assert.deepEqual(after.modelCenter.pools, frozen.center.pools); assert.equal(after.modelCenter.revision, frozen.center.revision);
|
||||
assert.deepEqual(after.modelCenter.connections.map(connectionDefinition), frozen.center.connections.map(connectionDefinition));
|
||||
const sessionIds = new Set(result.preflights.map(value => value.sessionId)); assert.equal(sessionIds.size, 2);
|
||||
assert.deepEqual(after.database.sessionIds.filter(value => !before.database.sessionIds.includes(value)).sort(), [...sessionIds].sort());
|
||||
assert.deepEqual(boot.projects.filter(value => value.id !== result.projectId), frozen.originalProjects); assert.equal(boot.projects.length, frozen.originalProjects.length + 1);
|
||||
const exports = new Map(after.exports.map(value => [value.id, value])), artifacts = new Map(after.persistent.artifacts.map(value => [value.path, value]));
|
||||
for (const value of before.exports) assert.deepEqual(exports.get(value.id), value, 'Original export: ' + value.id);
|
||||
for (const value of before.persistent.artifacts) assert.deepEqual(artifacts.get(value.path), value, 'Original artifact: ' + value.path);
|
||||
for (const value of before.persistent.files) assert.deepEqual(after.persistent.files.find(file => file.path === value.path), value, 'Original persistent file: ' + value.path);
|
||||
const db = new DatabaseSync(path.join(host.identity.dataRoot, 'workbench.sqlite'), { readOnly: true }); const addedRows = {}, sequenceChanges = []; let checkedOriginalRows = 0;
|
||||
try {
|
||||
assert.deepEqual(Object.keys(after.database.tables), Object.keys(before.database.tables));
|
||||
for (const [table, original] of Object.entries(before.database.tables)) {
|
||||
const current = after.database.tables[table]; assert.equal(current.schemaSha256, original.schemaSha256); addedRows[table] = current.count - original.count;
|
||||
const ids = new Set(original.rows.map(row => (row.kind ?? '') + ':' + row.id)), currentRows = new Map(current.rows.map(row => [(row.kind ?? '') + ':' + row.id, row]));
|
||||
for (const row of original.rows) {
|
||||
if (table === 'projects') { const value = db.prepare('SELECT * FROM projects WHERE id=?').get(row.id); assert.equal(digest({ ...value, position: value.position - 1 }), row.sha256); }
|
||||
else if (table === 'sqlite_sequence') {
|
||||
const value = db.prepare('SELECT * FROM sqlite_sequence').all().find(value => digest({ ...value, seq: row.id }) === row.sha256); assert.ok(value);
|
||||
const increment = value.name === 'events' ? after.eventCursor - before.eventCursor : value.name === 'model_attempts' ? result.requests : NaN;
|
||||
assert.ok(Number.isSafeInteger(increment)); assert.equal(value.seq, row.id + increment); sequenceChanges.push({ name: value.name, before: row.id, after: value.seq, increment });
|
||||
} else assert.equal(currentRows.get((row.kind ?? '') + ':' + row.id)?.sha256, row.sha256, table + ':' + row.id);
|
||||
checkedOriginalRows++;
|
||||
}
|
||||
if (['events', 'records', 'sessions'].includes(table)) {
|
||||
const columns = table === 'records' ? 'id,kind,session_id,project_id' : table === 'events' ? 'id,session_id,project_id' : 'id,project_id';
|
||||
for (const row of db.prepare('SELECT ' + columns + ' FROM ' + table).all()) if (!ids.has((row.kind ?? '') + ':' + row.id)) { assert.ok(sessionIds.has(table === 'sessions' ? row.id : row.session_id)); assert.equal(row.project_id, result.projectId); }
|
||||
} else if (table === 'outbox') {
|
||||
for (const row of db.prepare('SELECT outbox.seq,events.session_id FROM outbox JOIN events USING(seq) WHERE outbox.seq>?').all(before.eventCursor)) assert.ok(sessionIds.has(row.session_id));
|
||||
} else if (table === 'model_attempts') {
|
||||
assert.equal(addedRows[table], result.requests);
|
||||
for (const row of db.prepare('SELECT id,session_id FROM model_attempts').all()) if (!ids.has(':' + row.id)) assert.ok(sessionIds.has(row.session_id));
|
||||
} else if (table === 'projects') assert.equal(addedRows[table], 1);
|
||||
else assert.equal(addedRows[table], 0, 'No other new persistent rows: ' + table);
|
||||
}
|
||||
const references = new Map(db.prepare("SELECT payload FROM records WHERE kind='artifact'").all().map(row => { const value = JSON.parse(row.payload); return [value.relative, value]; }));
|
||||
for (const value of after.persistent.artifacts) if (!before.persistent.artifacts.some(original => original.path === value.path)) {
|
||||
const reference = references.get(value.path); assert.ok(reference && sessionIds.has(reference.sessionId)); assert.equal(reference.projectId, result.projectId); assert.equal(reference.sha256, value.sha256); assert.equal(reference.bytes, value.bytes);
|
||||
}
|
||||
} finally { db.close(); }
|
||||
receipt = { success: true, at: new Date().toISOString(), pid: host.identity.pid, instanceId: host.identity.instanceId, version: boot.version, originalSessions: before.database.sessionIds.length,
|
||||
originalModelRequests: before.database.tables.model_attempts.count, originalArtifacts: before.persistent.artifacts.length, checkedOriginalRows, addedRows, sequenceChanges,
|
||||
newArtifacts: after.persistent.artifacts.length - before.persistent.artifacts.length, currentSevenConnectionDefinitionsPoolsSelection: 'exact', originalExportsAndArtifactBytes: 'exact',
|
||||
originalMetadataAndCredentialRow: 'exact SHA; no credential decrypted or copied', originalProjects: 'exact after declared single front insertion', requests: result.requests, inspectorClosed: true, keysReadOrResaved: false, noRestorationOfOldHumanSettings: true };
|
||||
} catch (error) { receipt = { success: false, at: new Date().toISOString(), error: error.stack, noMutationOrRetry: true }; process.exitCode = 1; }
|
||||
finally { save(prefix + '/result.json', receipt); await close(); }
|
||||
console.log(JSON.stringify({ prefix, ...receipt }));
|
||||
}
|
||||
async function seal() {
|
||||
assert.equal(fs.existsSync(path.join(root, 'MANIFEST.json')), false);
|
||||
const result = load('result.json'), frozen = load('frozen.json'); assert.ok(result.requests <= 2);
|
||||
const auditDirs = fs.readdirSync(root).filter(value => /^audit-/.test(value)); assert.ok(auditDirs.some(value => load(value + '/result.json').success));
|
||||
const entries = []; const walk = relative => { for (const value of fs.readdirSync(path.join(root, relative), { withFileTypes: true })) {
|
||||
const filename = path.posix.join(relative.replaceAll('\\', '/'), value.name); assert.equal(value.isSymbolicLink(), false);
|
||||
if (value.isDirectory()) walk(filename); else if (value.isFile()) entries.push(source(filename)); else assert.fail('Unsupported evidence entry.');
|
||||
} }; walk(''); entries.sort((a, b) => a.path.localeCompare(b.path, 'en'));
|
||||
save('MANIFEST.json', { schemaVersion: 1, at: new Date().toISOString(), batch: 'm4-claude-correction-2026-11-12', entries });
|
||||
const receipt = { ...source('MANIFEST.json'), files: entries.length, bytes: entries.reduce((sum, value) => sum + value.bytes, 0), requests: result.requests,
|
||||
reportedTotalTokens: result.reportedTotalTokens, unknownRequests: result.unknownRequests, maximumPhysicalRequests: frozen.maximumPhysicalRequests,
|
||||
keysReadOrResaved: false, oldM4RootModified: false, engineeringOrPurgeReplayed: false };
|
||||
save('seal.json', receipt); console.log(JSON.stringify(receipt));
|
||||
}
|
||||
const action = process.argv[2];
|
||||
if (action === 'freeze') await freeze(); else if (action === 'run') await run(); else if (action === 'reconcile') await reconcile(); else if (action === 'audit') await audit(); else if (action === 'seal') await seal();
|
||||
else if (action !== undefined) throw Error('Supported actions: freeze, run, reconcile, audit, seal.');
|
||||
@@ -0,0 +1,146 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { randomUUID } from 'node:crypto';
|
||||
import { DatabaseSync } from 'node:sqlite';
|
||||
import { publicSnapshot, digest } from '../retained-host-state.mjs';
|
||||
import { root, origin, read, save, client, sha, compareConfig, compareConnectionDefinitions } from './common.mjs';
|
||||
import { inspectorClosed } from './inspector.mjs';
|
||||
|
||||
const stage = process.argv[2]; assert.ok(['node', 'react', 'python', 'final'].includes(stage));
|
||||
const prefix = 'audits/' + stage + '-' + randomUUID(), frozen = read('frozen.json'), compatibility = read('compatibility/result.json');
|
||||
const engineering = ['node', 'react', 'python'].filter(kind => fs.existsSync(path.join(root, kind, 'result.json'))).map(kind => ({ kind, ...read(kind + '/result.json') }));
|
||||
const sessionIds = new Set(compatibility.results.flatMap(value => value.preflights.map(row => row.sessionId)).concat(engineering.map(value => value.sessionId).filter(Boolean)));
|
||||
const projectIds = new Set(compatibility.results.map(value => value.projectId).concat(engineering.map(value => value.projectId).filter(Boolean)));
|
||||
const external = fs.existsSync(path.join(root, 'external-observations/declared.json')) ? read('external-observations/declared.json').sources.map(source => {
|
||||
assert.match(source.path, /^external-observations\/[a-f0-9-]{36}\/index\.json$/);
|
||||
assert.equal(sha(fs.readFileSync(path.join(root, source.path))), source.sha256);
|
||||
const value = read(source.path); assert.equal(value.m4Engineering, false); assert.equal(value.requestsReplayed, false);
|
||||
assert.equal(sessionIds.has(value.sessionId), false); assert.equal(frozen.baseline.database.sessionIds.includes(value.sessionId), false);
|
||||
assert.ok(frozen.originalProjects.some(project => project.id === value.projectId)); return value;
|
||||
}) : [];
|
||||
const allSessionIds = new Set([...sessionIds, ...external.map(value => value.sessionId)]);
|
||||
assert.equal(allSessionIds.size, sessionIds.size + external.length, 'No duplicate declared external sessions.');
|
||||
const sessionProjects = new Map(compatibility.results.flatMap(value => value.preflights.map(row => [row.sessionId, value.projectId]))
|
||||
.concat(engineering.filter(value => value.sessionId).map(value => [value.sessionId, value.projectId]), external.map(value => [value.sessionId, value.projectId])));
|
||||
const diagnosticIds = new Set(compatibility.results.map(value => read('compatibility/' + value.connectionId + '/network-start.json').job.id));
|
||||
const expectedRequests = compatibility.requests + engineering.reduce((sum, value) => sum + (value.requests ?? 0), 0);
|
||||
const externalRequests = external.reduce((sum, value) => sum + value.requests, 0), databaseRequests = expectedRequests + externalRequests;
|
||||
const { api, boot, close } = await client(); let result;
|
||||
try {
|
||||
await inspectorClosed(); const host = await api('/host/status'); assert.equal(host.activity.active, false);
|
||||
assert.equal(host.identity.pid, frozen.host.identity.pid); assert.equal(host.identity.instanceId, frozen.host.identity.instanceId); assert.equal(boot.version, '1.25.13');
|
||||
const after = await publicSnapshot(origin, host.identity.dataRoot); save(prefix + '/after.json', after); const before = frozen.baseline;
|
||||
assert.deepEqual(new Set(boot.projects.filter(value => !frozen.originalProjects.some(old => old.id === value.id)).map(value => value.id)), projectIds);
|
||||
assert.deepEqual(boot.projects.filter(value => !projectIds.has(value.id)), frozen.originalProjects);
|
||||
assert.deepEqual(after.database.sessionIds.filter(id => !before.database.sessionIds.includes(id)).sort(), [...allSessionIds].sort());
|
||||
const configComparison = compareConfig(frozen.config, boot.config), configDelta = boot.config.configRevision - frozen.config.configRevision;
|
||||
assert.equal(configDelta, engineering.filter(value => value.executed).length * 2, 'Exactly one select and restore per new engineering sample.');
|
||||
const definitions = compareConnectionDefinitions(frozen.center, after.modelCenter, frozen.config, boot.config);
|
||||
assert.deepEqual(after.modelCenter.selected, frozen.center.selected); assert.deepEqual(after.modelCenter.pools, frozen.center.pools);
|
||||
const exportMap = new Map(after.exports.map(value => [value.id, value])), artifactMap = new Map(after.persistent.artifacts.map(value => [value.path, value]));
|
||||
for (const original of before.exports) assert.deepEqual(exportMap.get(original.id), original, 'Original session export: ' + original.id);
|
||||
for (const original of before.persistent.artifacts) assert.deepEqual(artifactMap.get(original.path), original, 'Original artifact byte: ' + original.path);
|
||||
const batchExports = compatibility.results.flatMap(value => value.preflights.map(row => ({ sessionId: row.sessionId,
|
||||
source: 'compatibility/' + value.connectionId + '/' + row.scope + '/export.json' })))
|
||||
.concat(engineering.filter(value => value.sessionId).map(value => ({ sessionId: value.sessionId, source: value.kind + '/export.json' })));
|
||||
for (const value of batchExports) {
|
||||
const saved = read(value.source);
|
||||
assert.deepEqual(exportMap.get(value.sessionId), { id: value.sessionId, project: digest(saved.project), session: digest(saved.session), execution: digest(saved.execution) },
|
||||
'Every completed batch export must still match its original capture: ' + value.sessionId);
|
||||
}
|
||||
for (const value of external) {
|
||||
assert.deepEqual(exportMap.get(value.sessionId), value.exports, 'Explicit external session export: ' + value.sessionId);
|
||||
for (const artifact of value.artifacts) assert.deepEqual(artifactMap.get(artifact.path), { path: artifact.path, bytes: artifact.bytes, sha256: artifact.sha256 }, 'Explicit external artifact byte: ' + artifact.path);
|
||||
}
|
||||
const db = new DatabaseSync(path.join(host.identity.dataRoot, 'workbench.sqlite'), { readOnly: true }); const addedRows = {}, metadataChanges = [], sequenceChanges = []; let checkedOriginalRows = 0;
|
||||
try {
|
||||
assert.deepEqual(Object.keys(after.database.tables), Object.keys(before.database.tables));
|
||||
for (const [table, originalTable] of Object.entries(before.database.tables)) {
|
||||
const currentTable = after.database.tables[table]; assert.equal(currentTable.schemaSha256, originalTable.schemaSha256); addedRows[table] = currentTable.count - originalTable.count;
|
||||
const originals = new Set(originalTable.rows.map(value => (value.kind ?? '') + ':' + value.id)), currentRows = new Map(currentTable.rows.map(value => [(value.kind ?? '') + ':' + value.id, value]));
|
||||
for (const value of external) if (value.tables[table]) {
|
||||
assert.equal(value.tables[table].schemaSha256, originalTable.schemaSha256);
|
||||
for (const row of value.tables[table].rows) {
|
||||
assert.equal(originals.has((row.kind ?? '') + ':' + row.id), false, 'External observation cannot replace an original row.');
|
||||
assert.deepEqual(currentRows.get((row.kind ?? '') + ':' + row.id), row, 'Every explicitly recorded external row: ' + table + ':' + row.id);
|
||||
}
|
||||
}
|
||||
for (const original of originalTable.rows) {
|
||||
if (table === 'projects') {
|
||||
const current = db.prepare('SELECT * FROM projects WHERE id=?').get(original.id); assert.ok(current);
|
||||
assert.equal(digest({ ...current, position: current.position - projectIds.size }), original.sha256, 'Original project fields and declared front insertion positions.');
|
||||
} else if (table === 'sqlite_sequence') {
|
||||
const sequences = db.prepare('SELECT * FROM sqlite_sequence').all();
|
||||
const current = sequences.find(value => digest({ ...value, seq: original.id }) === original.sha256); assert.ok(current, 'Original sequence authority.');
|
||||
const increment = current.name === 'events' ? after.eventCursor - before.eventCursor : current.name === 'model_attempts' ? databaseRequests : NaN;
|
||||
assert.ok(Number.isSafeInteger(increment)); assert.equal(current.seq, original.id + increment);
|
||||
sequenceChanges.push({ name: current.name, original: original.id, current: current.seq, increment });
|
||||
} else if (table === 'metadata' && ['config', 'modelRegistry', 'modelNetworkDiagnostics'].includes(original.id)) {
|
||||
const current = db.prepare('SELECT * FROM metadata WHERE key=?').get(original.id), value = JSON.parse(current.value); let normalized;
|
||||
if (original.id === 'config') { assert.equal(value.configRevision, boot.config.configRevision); normalized = { ...value, configRevision: frozen.config.configRevision }; }
|
||||
else if (original.id === 'modelRegistry') {
|
||||
assert.equal(value.revision, frozen.center.revision + configDelta); assert.deepEqual(value.selected, frozen.center.selected);
|
||||
normalized = { ...value, revision: frozen.center.revision };
|
||||
} else {
|
||||
assert.ok(Array.isArray(value)); const additions = value.filter(row => diagnosticIds.has(row.id)); assert.equal(additions.length, diagnosticIds.size);
|
||||
normalized = value.filter(row => !diagnosticIds.has(row.id));
|
||||
}
|
||||
assert.equal(digest({ ...current, value: JSON.stringify(normalized) }), original.sha256, 'Original public metadata bytes after the declared increment: ' + original.id);
|
||||
metadataChanges.push({ key: original.id, originalRowSha256: original.sha256, currentRowSha256: digest(current), originalRestoredByDeclaredIncrement: true });
|
||||
} else {
|
||||
const current = currentRows.get((original.kind ?? '') + ':' + original.id); assert.ok(current, table + ': missing original ' + original.id);
|
||||
assert.equal(current.sha256, original.sha256, table + ': changed original ' + original.id);
|
||||
}
|
||||
checkedOriginalRows++;
|
||||
}
|
||||
if (['events', 'records', 'sessions'].includes(table)) {
|
||||
const columns = table === 'records' ? 'id,kind,session_id,project_id' : table === 'events' ? 'id,session_id,project_id' : 'id,project_id';
|
||||
for (const row of db.prepare('SELECT ' + columns + ' FROM ' + table).all()) if (!originals.has((row.kind ?? '') + ':' + row.id)) {
|
||||
const sessionId = table === 'sessions' ? row.id : row.session_id;
|
||||
assert.ok(allSessionIds.has(sessionId), 'Every new ' + table + ' needs a declared session.'); assert.equal(row.project_id, sessionProjects.get(sessionId));
|
||||
const observation = external.find(value => value.sessionId === sessionId);
|
||||
if (observation) assert.ok(observation.tables[table].rows.some(value => value.id === row.id && (value.kind ?? '') === (row.kind ?? '')), 'External rows must match their explicit immutable list.');
|
||||
}
|
||||
} else if (table === 'outbox') {
|
||||
for (const row of db.prepare('SELECT outbox.seq,events.session_id FROM outbox JOIN events USING(seq) WHERE outbox.seq>?').all(before.eventCursor)) {
|
||||
assert.ok(allSessionIds.has(row.session_id));
|
||||
const observation = external.find(value => value.sessionId === row.session_id);
|
||||
if (observation) assert.ok(observation.tables.outbox.rows.some(value => value.id === row.seq), 'Explicit external outbox sequence.');
|
||||
}
|
||||
} else if (table === 'model_attempts') {
|
||||
assert.equal(addedRows[table], databaseRequests);
|
||||
// The ledger stores session_id; its project authority is the existing
|
||||
// sessions relation, not a nonexistent model_attempts.project_id field.
|
||||
for (const row of db.prepare('SELECT model_attempts.id,model_attempts.session_id,sessions.project_id FROM model_attempts LEFT JOIN sessions ON sessions.id=model_attempts.session_id').all()) if (!originals.has(':' + row.id)) {
|
||||
assert.ok(allSessionIds.has(row.session_id)); assert.equal(row.project_id, sessionProjects.get(row.session_id));
|
||||
const observation = external.find(value => value.sessionId === row.session_id);
|
||||
if (observation) assert.ok(observation.tables.model_attempts.rows.some(value => value.id === row.id), 'Explicit external request ID.');
|
||||
}
|
||||
} else if (['metadata', 'sqlite_sequence'].includes(table)) assert.equal(addedRows[table], 0);
|
||||
}
|
||||
const references = new Map(db.prepare("SELECT payload FROM records WHERE kind='artifact'").all().map(row => { const value = JSON.parse(row.payload); return [value.relative, value]; }));
|
||||
const originals = new Set(before.persistent.artifacts.map(value => value.path));
|
||||
for (const artifact of after.persistent.artifacts) if (!originals.has(artifact.path)) {
|
||||
const reference = references.get(artifact.path); assert.ok(reference && allSessionIds.has(reference.sessionId) && reference.projectId === sessionProjects.get(reference.sessionId), 'New original artifact must be owned or explicitly recorded externally.');
|
||||
assert.equal(reference.sha256, artifact.sha256); assert.equal(reference.bytes, artifact.bytes);
|
||||
}
|
||||
} finally { db.close(); }
|
||||
for (const original of before.persistent.files) {
|
||||
const current = after.persistent.files.find(value => value.path === original.path); assert.ok(current);
|
||||
if (original.path === 'config.json' && current.sha256 !== original.sha256) {
|
||||
const value = JSON.parse(fs.readFileSync(path.join(host.identity.dataRoot, original.path), 'utf8'));
|
||||
assert.equal(sha(JSON.stringify({ ...value, configRevision: frozen.config.configRevision }, null, 2)), original.sha256, 'Original config file bytes after the exact config revision increment.');
|
||||
} else assert.equal(current.sha256, original.sha256, 'Original persistent file: ' + original.path);
|
||||
}
|
||||
result = { success: true, stage, version: boot.version, pid: host.identity.pid, instanceId: host.identity.instanceId, originalProjects: frozen.originalProjects.length,
|
||||
originalSessions: before.database.sessionIds.length, originalModelRequests: before.database.tables.model_attempts.count, originalArtifacts: before.persistent.artifacts.length,
|
||||
addedProjects: projectIds.size, addedSessions: sessionIds.size, addedRows, newArtifacts: after.persistent.artifacts.length - before.persistent.artifacts.length, checkedOriginalRows,
|
||||
originalSessionExports: 'exact', batchCapturedExports: { count: batchExports.length, status: 'exact' }, originalArtifactBytes: 'exact', originalRows: 'exact with explicitly hashed project-position/public-metadata/sequence increments',
|
||||
configComparison, definitions, metadataChanges, sequenceChanges, batchModelRequests: expectedRequests, externalModelRequests: externalRequests,
|
||||
externalObservations: external.map(value => ({ sessionId: value.sessionId, projectId: value.projectId, requests: value.requests, tables: Object.fromEntries(Object.entries(value.tables).map(([table, source]) => [table, source.count])), artifacts: value.artifacts.length, scope: value.scope })),
|
||||
originalCredentialRow: 'exact SHA; never decrypted or copied',
|
||||
engineering: engineering.map(value => ({ kind: value.kind, success: value.success, executed: value.executed })), inspectorClosed: true, keysReadOrResaved: false, historicalSamplesReplayed: false, purgeExecuted: false };
|
||||
} catch (error) { result = { success: false, stage, error: error.stack, keysReadOrResaved: false, noMutationOrRetry: true }; process.exitCode = 1; }
|
||||
finally { save(prefix + '/result.json', result); await close(); }
|
||||
console.log(JSON.stringify({ prefix, ...result }));
|
||||
@@ -0,0 +1,119 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { createHash } from 'node:crypto';
|
||||
import { Agent, fetch } from 'undici';
|
||||
import { messagesOutputSetting } from '../../src/shared/contracts/model-capacity.mjs';
|
||||
|
||||
export let root = path.resolve('.qa/m4-live-baseline-2026-11-11');
|
||||
export let origin = 'http://127.0.0.1:4318';
|
||||
export function configureFixture(options) {
|
||||
const target = path.resolve(options.root), url = new URL(options.origin);
|
||||
assert.equal(path.dirname(target), path.resolve('.qa'));
|
||||
assert.match(path.basename(target), /^m4-driver-[a-f0-9-]{36}$/);
|
||||
assert.equal(url.hostname, '127.0.0.1'); assert.ok(url.port && url.port !== '4318');
|
||||
root = target; origin = url.origin;
|
||||
}
|
||||
export const sha = bytes => createHash('sha256').update(bytes).digest('hex');
|
||||
export const read = relative => JSON.parse(fs.readFileSync(path.join(root, relative), 'utf8'));
|
||||
export function save(relative, value) {
|
||||
const file = path.resolve(root, relative);
|
||||
assert.ok(file.startsWith(root + path.sep), 'Evidence must stay inside its new root.');
|
||||
fs.mkdirSync(path.dirname(file), { recursive: true });
|
||||
fs.writeFileSync(file, typeof value === 'string' || Buffer.isBuffer(value) ? value : JSON.stringify(value, null, 2) + '\n', { flag: 'wx' });
|
||||
}
|
||||
export function connectionDefinition(value) {
|
||||
const { health, status, activeRequests, ...definition } = value;
|
||||
return definition;
|
||||
}
|
||||
export function compareConfig(before, after) {
|
||||
assert.ok(Number.isInteger(after.configRevision) && after.configRevision >= before.configRevision);
|
||||
assert.deepEqual({ ...after, configRevision: before.configRevision }, before);
|
||||
return { originalRevision: before.configRevision, currentRevision: after.configRevision, otherFields: 'exact' };
|
||||
}
|
||||
export function compareConnectionDefinitions(before, after, beforeConfig, afterConfig) {
|
||||
assert.deepEqual(after.connections.map(value => value.id), before.connections.map(value => value.id));
|
||||
const changes = [];
|
||||
for (let index = 0; index < before.connections.length; index++) {
|
||||
const original = connectionDefinition(before.connections[index]), current = connectionDefinition(after.connections[index]);
|
||||
if (original.id === 'legacy-default') {
|
||||
// The legacy public projection uses configRevision, rather than a saved
|
||||
// connection revision. Both deliberate select/restore saves advance it.
|
||||
const expected = original.revision + afterConfig.configRevision - beforeConfig.configRevision;
|
||||
assert.equal(current.revision, expected, 'Legacy projection tracks the exact declared config revision increment.');
|
||||
changes.push({ id: original.id, originalRevision: original.revision, currentRevision: current.revision, reason: 'public-legacy-projection-of-configRevision' });
|
||||
current.revision = original.revision;
|
||||
}
|
||||
assert.deepEqual(current, original, 'All original connection fields: ' + original.id);
|
||||
}
|
||||
return { originalFields: 'exact', declaredPublicProjectionChanges: changes };
|
||||
}
|
||||
export function expectedOutputFields(connection) {
|
||||
const fields = Object.fromEntries(['max_tokens', 'max_output_tokens', 'max_completion_tokens']
|
||||
.filter(key => Object.hasOwn(connection.nativeParameters ?? {}, key)).map(key => [key, connection.nativeParameters[key]]));
|
||||
if (connection.protocol !== 'anthropic-messages') return { fields, source: Object.keys(fields).length ? 'explicit_native' : 'passthrough_no_local_output_cap' };
|
||||
const setting = messagesOutputSetting(connection);
|
||||
if (setting.value !== undefined) fields.max_tokens = setting.value;
|
||||
return { fields, source: setting.source, declaration: setting };
|
||||
}
|
||||
export async function client() {
|
||||
const dispatcher = new Agent({ connect: { autoSelectFamily: true, timeout: 5000 } }); let token;
|
||||
const api = async (route, input, timeout = 30000) => {
|
||||
const response = await fetch(origin + '/api' + route, { dispatcher, method: input === undefined ? 'GET' : 'POST', redirect: 'error', signal: AbortSignal.timeout(timeout),
|
||||
headers: { ...(token ? { 'X-Workbench-Token': token } : {}), 'Content-Type': 'application/json', Connection: 'close' },
|
||||
...(input === undefined ? {} : { body: JSON.stringify(input) }) });
|
||||
const value = await response.json();
|
||||
if (!response.ok) throw Object.assign(new Error(route + ': ' + (value.error ?? response.status)), { status: response.status, response: value });
|
||||
return value;
|
||||
};
|
||||
const boot = await api('/bootstrap'); token = boot.token;
|
||||
api.download = async (route, timeout = 30000) => {
|
||||
const response = await fetch(origin + '/api' + route, { dispatcher, method: 'GET', redirect: 'error', signal: AbortSignal.timeout(timeout),
|
||||
headers: { 'X-Workbench-Token': token, Connection: 'close' } });
|
||||
if (!response.ok) throw Object.assign(new Error(route + ': artifact download HTTP ' + response.status), { status: response.status });
|
||||
return Buffer.from(await response.arrayBuffer());
|
||||
};
|
||||
// The bootstrap authentication token is only kept in this client's closure.
|
||||
delete boot.token;
|
||||
return { api, boot, close: () => dispatcher.close() };
|
||||
}
|
||||
export async function sessionAttempts(api, sessionId) {
|
||||
let cursor, total, snapshot; const rows = [], references = [];
|
||||
do {
|
||||
const page = await api('/model-center/attempts?sessionId=' + sessionId + '&limit=50' + (cursor ? '&cursor=' + encodeURIComponent(cursor) : ''));
|
||||
if (total === undefined) { total = page.total; snapshot = page.snapshotSequence; }
|
||||
assert.equal(page.total, total); assert.equal(page.snapshotSequence, snapshot);
|
||||
rows.push(...page.items); references.push(...page.references); cursor = page.nextCursor;
|
||||
assert.equal(page.hasMore, !!cursor);
|
||||
} while (cursor);
|
||||
assert.equal(rows.length, total); assert.equal(new Set(rows.map(value => value.id)).size, total);
|
||||
// Public pages are newest first. Store experiment receipts in dispatch order.
|
||||
return { rows: rows.reverse(), references: references.reverse(), total, snapshot };
|
||||
}
|
||||
export async function artifactBytes(api, sessionId, artifact) {
|
||||
assert.equal(typeof api.download, 'function', 'Original artifacts need the byte-preserving download path.');
|
||||
const bytes = await api.download('/artifacts/' + artifact.id + '?session=' + sessionId + '&download=1');
|
||||
assert.equal(sha(bytes), artifact.sha256); if (artifact.bytes !== undefined) assert.equal(bytes.length, artifact.bytes);
|
||||
return bytes;
|
||||
}
|
||||
export async function captureSession(api, sessionId, prefix) {
|
||||
const state = await api('/sessions/' + sessionId), exported = await api('/sessions/' + sessionId + '/export');
|
||||
save(prefix + '/session.json', state); save(prefix + '/export.json', exported);
|
||||
const ledger = await sessionAttempts(api, sessionId); save(prefix + '/attempts.json', ledger.rows); save(prefix + '/ledger-references.json', ledger);
|
||||
const captures = [], capturedIds = new Set();
|
||||
for (const row of ledger.rows) for (const kind of ['request', 'response']) {
|
||||
const id = row[kind + 'ArtifactId']; if (!id) continue;
|
||||
const artifact = state.execution.artifacts.find(value => value.id === id) ?? { id, sha256: row[kind + 'Sha256'] };
|
||||
const bytes = await artifactBytes(api, sessionId, artifact);
|
||||
assert.equal(sha(bytes), row[kind + 'Sha256']);
|
||||
const relative = prefix + '/captures/' + row.id + '-' + kind + (kind === 'request' ? '.json' : row.streaming ? '.sse' : '.txt');
|
||||
save(relative, bytes); captures.push({ attemptId: row.id, artifactId: id, kind, path: relative, sha256: sha(bytes), bytes: bytes.length }); capturedIds.add(id);
|
||||
}
|
||||
for (const artifact of state.execution.artifacts) {
|
||||
if (capturedIds.has(artifact.id) || !/^(text\/|application\/(json|x-ndjson))/.test(artifact.mediaType ?? '')) continue;
|
||||
const bytes = await artifactBytes(api, sessionId, artifact), relative = prefix + '/artifacts/' + artifact.id + (artifact.mediaType === 'application/json' ? '.json' : '.txt');
|
||||
save(relative, bytes); captures.push({ artifactId: artifact.id, kind: artifact.kind, path: relative, sha256: sha(bytes), bytes: bytes.length });
|
||||
}
|
||||
save(prefix + '/artifact-sources.json', captures);
|
||||
return { state, exported, ...ledger, captures };
|
||||
}
|
||||
@@ -0,0 +1,79 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { randomUUID } from 'node:crypto';
|
||||
import { setTimeout as delay } from 'node:timers/promises';
|
||||
import { root, save, read, client, captureSession, connectionDefinition, expectedOutputFields } from './common.mjs';
|
||||
import { invokeProbe, inspectorClosed } from './inspector.mjs';
|
||||
|
||||
const frozen = read('frozen.json'); assert.equal(fs.existsSync(path.join(root, 'compatibility')), false, 'Never replay a compatibility batch.');
|
||||
save('compatibility/claim.json', { id: randomUUID(), at: new Date().toISOString(), scopes: ['text', 'tools'], streaming: 'Separately assessed from each configured streaming request; no redundant real request.', noRetry: true });
|
||||
const { api, boot, close } = await client(), results = [];
|
||||
try {
|
||||
assert.equal(boot.version, '1.25.13'); await inspectorClosed();
|
||||
assert.equal((await api('/host/status')).identity.pid, frozen.host.identity.pid);
|
||||
for (const connection of frozen.connections) {
|
||||
const prefix = 'compatibility/' + connection.id;
|
||||
assert.deepEqual(connectionDefinition((await api('/model-center')).connections.find(value => value.id === connection.id)), connectionDefinition(connection), 'Frozen connection changed.');
|
||||
const result = { connectionId: connection.id, name: connection.name, model: connection.model, protocol: connection.protocol,
|
||||
endpointKind: new URL(connection.baseUrl).hostname === 'api.deepseek.com' ? 'official-deepseek-endpoint' : 'compatible-proxy-endpoint', preflights: [], media: 'not_exercised_in_this_slice' };
|
||||
console.log(JSON.stringify({ phase: 'compatibility', connection: connection.name, protocol: connection.protocol }));
|
||||
let started;
|
||||
try {
|
||||
started = await api('/model-center/network-diagnostics', { id: connection.id, revision: connection.revision }); save(prefix + '/network-start.json', started);
|
||||
const deadline = Date.now() + 30000; let network;
|
||||
while (true) {
|
||||
network = (await api('/model-center/network-diagnostics')).jobs.find(value => value.id === started.job.id);
|
||||
if (network?.finishedAt) break;
|
||||
if (Date.now() >= deadline) { network = (await api('/model-center/network-diagnostics/' + started.job.id + '/stop', {})).job; result.networkObservationEnded = true; break; }
|
||||
await delay(250);
|
||||
}
|
||||
save(prefix + '/network.json', network); result.network = network.status;
|
||||
assert.equal(network.httpRequests, 0); assert.equal(network.modelRequests, 0); assert.equal(network.credentialAccess, false);
|
||||
} catch (error) {
|
||||
result.network = 'error'; save(prefix + '/network-error.json', { error: error.stack, response: error.response, noRetry: true });
|
||||
if (started?.job?.id) save(prefix + '/network-owned-stop.json', await api('/model-center/network-diagnostics/' + started.job.id + '/stop', {}));
|
||||
}
|
||||
const requestId = randomUUID();
|
||||
try {
|
||||
const catalog = await api('/model-center/catalog', { id: connection.id, revision: connection.revision, requestId }, 90000);
|
||||
save(prefix + '/catalog.json', catalog); result.catalog = { complete: catalog.complete, models: catalog.entries.length, selectedListed: catalog.entries.some(value => value.id === connection.model),
|
||||
evidenceScope: 'Product catalog preview including its supplied source metadata; no invented raw HTTP artifact.' };
|
||||
} catch (error) { result.catalog = { complete: false, error: error.message, response: error.response, noRetry: true }; save(prefix + '/catalog-error.json', result.catalog); }
|
||||
finally { save(prefix + '/catalog-cancel.json', await api('/model-center/catalog/cancel', { id: requestId })); }
|
||||
const projectRoot = path.join(root, prefix, 'project'); fs.mkdirSync(projectRoot, { recursive: true });
|
||||
fs.writeFileSync(path.join(projectRoot, 'README.md'), 'M4 independent compatibility observations only. No engineering implementation.\n', { flag: 'wx' });
|
||||
const project = await api('/projects', { root: projectRoot }); result.projectId = project.id;
|
||||
for (const scope of ['text', 'tools']) {
|
||||
const session = await api('/sessions', { projectId: project.id });
|
||||
const input = { pid: frozen.host.identity.pid, port: 4318, runtimeRoot: boot.runtimeRoot, projectId: project.id, sessionId: session.id, connection,
|
||||
scope, nonce: randomUUID(), expectedOutput: expectedOutputFields(connection), observationMs: 180000 };
|
||||
save(prefix + '/' + scope + '/input.json', input); let invoked;
|
||||
try { invoked = await invokeProbe(input); save(prefix + '/' + scope + '/invocation.json', invoked); }
|
||||
catch (error) { save(prefix + '/' + scope + '/invocation-error.json', { error: error.stack, noRetry: true }); }
|
||||
const captured = await captureSession(api, session.id, prefix + '/' + scope), attempt = captured.rows[0];
|
||||
const request = captured.captures.find(value => value.attemptId === attempt?.id && value.kind === 'request');
|
||||
const response = captured.captures.find(value => value.attemptId === attempt?.id && value.kind === 'response');
|
||||
const body = request ? JSON.parse(fs.readFileSync(path.join(root, request.path), 'utf8')) : null;
|
||||
const raw = response ? fs.readFileSync(path.join(root, response.path), 'utf8') : '';
|
||||
const outputFields = body ? Object.fromEntries(['max_tokens', 'max_output_tokens', 'max_completion_tokens'].filter(key => Object.hasOwn(body, key)).map(key => [key, body[key]])) : null;
|
||||
const streaming = { requested: body?.stream === true, sseObserved: /^(?:event:|data:)/m.test(raw), complete: attempt?.status === 'completed' && attempt?.modelIdentity?.complete === true };
|
||||
const wirePassed = !!body && JSON.stringify(outputFields) === JSON.stringify(input.expectedOutput.fields) && (!body.stream || streaming.sseObserved);
|
||||
const passed = !!invoked?.result.contentPassed && invoked.result.actualToolsExecuted === 0 && captured.total === 1 && wirePassed
|
||||
&& attempt.status === 'completed' && attempt.modelIdentity?.status === 'matched' && attempt.modelIdentity.complete === true;
|
||||
const entry = { scope, passed, sessionId: session.id, taskId: invoked?.result.taskId, runId: invoked?.result.runId, identity: attempt?.modelIdentity ?? null,
|
||||
usage: attempt?.usage ?? null, usageStatus: attempt?.usageStatus ?? 'not_dispatched', status: attempt?.status ?? 'not_dispatched',
|
||||
diagnostic: invoked?.result.diagnostic ?? null, actualToolsExecuted: invoked?.result.actualToolsExecuted ?? null, requests: captured.total,
|
||||
streaming, wirePassed, outputFields, expectedOutput: input.expectedOutput, captures: captured.captures };
|
||||
save(prefix + '/' + scope + '/result.json', entry); result.preflights.push(entry);
|
||||
console.log(JSON.stringify({ connection: connection.name, scope, passed, identity: entry.identity?.status, usageStatus: entry.usageStatus, streaming, outputFields, diagnostic: entry.diagnostic?.kind }));
|
||||
assert.ok(captured.total <= 1); await inspectorClosed();
|
||||
}
|
||||
result.engineeringAllowed = result.catalog.complete && result.catalog.selectedListed && result.preflights.every(value => value.passed && value.usageStatus === 'reported');
|
||||
save(prefix + '/result.json', result); results.push(result);
|
||||
}
|
||||
const center = await api('/model-center');
|
||||
assert.deepEqual(center.selected, frozen.center.selected); assert.deepEqual(center.connections.map(connectionDefinition), frozen.center.connections.map(connectionDefinition)); assert.deepEqual(center.pools, frozen.center.pools);
|
||||
save('compatibility/result.json', { completed: true, results, requests: results.reduce((sum, row) => sum + row.preflights.reduce((n, value) => n + value.requests, 0), 0),
|
||||
connectionDefinitionsPoolsSelection: 'exact', healthChangesRetained: true, keysReadOrResaved: false, historicalSamplesReplayed: false });
|
||||
} finally { await inspectorClosed(); await close(); }
|
||||
@@ -0,0 +1,25 @@
|
||||
import fs from 'node:fs';
|
||||
import assert from 'node:assert/strict';
|
||||
import { execFileSync } from 'node:child_process';
|
||||
import { root, origin, save, client, expectedOutputFields } from './common.mjs';
|
||||
import { publicSnapshot } from '../retained-host-state.mjs';
|
||||
import { inspectorClosed } from './inspector.mjs';
|
||||
|
||||
assert.ok(fs.existsSync(root)); assert.equal(fs.existsSync(root + '/frozen.json'), false, 'Freeze this batch once.');
|
||||
assert.equal(execFileSync('git', ['status', '--porcelain'], { encoding: 'utf8', windowsHide: true }).trim(), '', 'Commit and verify the new driver before real requests.');
|
||||
const { api, boot, close } = await client();
|
||||
try {
|
||||
await inspectorClosed(); const host = await api('/host/status'), center = await api('/model-center');
|
||||
assert.equal(boot.version, '1.25.13'); assert.equal(host.activity.active, false);
|
||||
for (const [key, expected] of Object.entries({ executionMode: 'unrestricted', capacityMode: 'passthrough', approvalMode: 'full', maxSteps: 0, maxRunTokens: 0, maxTaskTokens: 0 })) assert.equal(boot.config[key], expected, key);
|
||||
const connections = center.connections.filter(value => value.enabled && value.hasKey); assert.ok(connections.length > 0, 'Restore credentials in Model Center before freezing.');
|
||||
const priority = connections.slice().sort((a, b) => {
|
||||
const rank = value => new URL(value.baseUrl).hostname === 'api.deepseek.com' ? 0 : value.protocol === 'responses' ? 1 : value.protocol === 'chat-completions' ? 2 : 3;
|
||||
return rank(a) - rank(b) || a.id.localeCompare(b.id);
|
||||
}).map(value => value.id);
|
||||
const baseline = await publicSnapshot(origin, host.identity.dataRoot);
|
||||
save('frozen.json', { observedAt: new Date().toISOString(), host, config: boot.config, center, connections, baseline, originalProjects: boot.projects,
|
||||
engineeringPriority: priority, outputFields: Object.fromEntries(connections.map(value => [value.id, expectedOutputFields(value)])),
|
||||
driverCommit: execFileSync('git', ['rev-parse', 'HEAD'], { encoding: 'utf8', windowsHide: true }).trim(), modelRequests: 0, historicalSamplesReplayed: false, keysReadOrResaved: false });
|
||||
console.log(JSON.stringify({ frozen: true, version: boot.version, pid: host.identity.pid, connections: connections.map(value => ({ id: value.id, name: value.name, protocol: value.protocol, model: value.model, revision: value.revision })), priority, modelRequests: 0 }));
|
||||
} finally { await close(); }
|
||||
@@ -0,0 +1,56 @@
|
||||
import assert from 'node:assert/strict';
|
||||
import net from 'node:net';
|
||||
import { setTimeout as delay } from 'node:timers/promises';
|
||||
import { providerPreflight } from './payload.mjs';
|
||||
|
||||
export async function inspectorClosed() {
|
||||
return new Promise((resolve, reject) => {
|
||||
const socket = net.createConnection({ host: '127.0.0.1', port: 9229 }); socket.setTimeout(1500);
|
||||
socket.once('connect', () => { socket.destroy(); reject(Error('Inspector already open.')); });
|
||||
socket.once('error', error => { socket.destroy(); error.code === 'ECONNREFUSED' ? resolve(true) : reject(error); });
|
||||
socket.once('timeout', () => { socket.destroy(); reject(Error('Inspector state unknown.')); });
|
||||
});
|
||||
}
|
||||
class Protocol {
|
||||
constructor(socket) {
|
||||
this.socket = socket; this.nextId = 1; this.pending = new Map();
|
||||
socket.addEventListener('message', event => {
|
||||
const message = JSON.parse(String(event.data)); if (!message.id) return;
|
||||
const pending = this.pending.get(message.id); if (!pending) return;
|
||||
this.pending.delete(message.id); clearTimeout(pending.timer);
|
||||
message.error ? pending.reject(Error(message.error.message)) : pending.resolve(message.result);
|
||||
});
|
||||
socket.addEventListener('close', () => { for (const pending of this.pending.values()) { clearTimeout(pending.timer); pending.reject(Error('Inspector closed; outcome unknown.')); } this.pending.clear(); });
|
||||
}
|
||||
send(method, params, timeout = 10000) {
|
||||
return new Promise((resolve, reject) => {
|
||||
const id = this.nextId++, timer = setTimeout(() => { this.pending.delete(id); reject(Error('Inspector timeout; do not replay.')); }, timeout);
|
||||
this.pending.set(id, { resolve, reject, timer }); this.socket.send(JSON.stringify({ id, method, params }));
|
||||
});
|
||||
}
|
||||
}
|
||||
export async function invokeProbe(input) {
|
||||
await inspectorClosed(); let socket, protocol, enabled = false;
|
||||
try {
|
||||
process._debugProcess(input.pid); enabled = true; let targets;
|
||||
for (let index = 0; index < 40 && !targets; index++) {
|
||||
await delay(100);
|
||||
try { targets = await (await fetch('http://127.0.0.1:9229/json/list', { signal: AbortSignal.timeout(700) })).json(); } catch {}
|
||||
}
|
||||
assert.equal(targets?.length, 1); assert.ok(targets[0].webSocketDebuggerUrl.startsWith('ws://127.0.0.1:9229/'));
|
||||
socket = new WebSocket(targets[0].webSocketDebuggerUrl);
|
||||
await new Promise((resolve, reject) => { socket.addEventListener('open', resolve, { once: true }); socket.addEventListener('error', reject, { once: true }); });
|
||||
protocol = new Protocol(socket);
|
||||
const identity = await protocol.send('Runtime.evaluate', { expression: 'process.pid', returnByValue: true }); assert.equal(identity.result.value, input.pid);
|
||||
const expression = '(' + providerPreflight.toString() + ')(' + JSON.stringify(input) + ')';
|
||||
const result = await protocol.send('Runtime.evaluate', { expression, awaitPromise: true, returnByValue: true }, input.observationMs + 30000);
|
||||
if (result.exceptionDetails) throw Error(result.exceptionDetails.exception?.description ?? result.exceptionDetails.text);
|
||||
return { result: result.result.value, expression };
|
||||
} finally {
|
||||
if (protocol && enabled) {
|
||||
await protocol.send('Runtime.evaluate', { expression: "setTimeout(() => process.getBuiltinModule('inspector').close(), 25); 'closing'", returnByValue: true });
|
||||
await delay(200);
|
||||
}
|
||||
socket?.close(); await inspectorClosed();
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,66 @@
|
||||
// A one-shot compatibility observation in a new owned session. This function
|
||||
// calls the existing Provider; it never reads or returns any credential value.
|
||||
export async function providerPreflight(input) {
|
||||
const ensure = (value, message) => { if (!value) throw Error(message); };
|
||||
ensure(process.pid === input.pid, 'Host PID changed.');
|
||||
const servers = process._getActiveHandles().filter(h => h?.listening && h.address?.()?.address === '127.0.0.1' && h.address?.()?.port === input.port);
|
||||
ensure(servers.length === 1, 'Expected one loopback host.');
|
||||
const { store, provider, runtime } = servers[0].workbenchHostAuthority.context;
|
||||
ensure(runtime.appDirectory === input.runtimeRoot, 'Runtime changed.'); ensure(!runtime.active.size, 'Host is busy.');
|
||||
const session = store.session(input.sessionId), project = store.project(session.projectId), hub = provider.modelHub(store);
|
||||
ensure(session.projectId === input.projectId && !session.runId && !session.messages.length, 'Only a fresh owned session may run.');
|
||||
const connection = hub.view().connections.find(value => value.id === input.connection.id);
|
||||
ensure(connection?.enabled && connection.hasKey && connection.revision === input.connection.revision
|
||||
&& connection.credentialRevision === input.connection.credentialRevision && connection.model === input.connection.model
|
||||
&& connection.protocol === input.connection.protocol && connection.baseUrl === input.connection.baseUrl, 'Frozen connection changed.');
|
||||
globalThis.__workbenchM4Preflights ??= new Set();
|
||||
ensure(!globalThis.__workbenchM4Preflights.has(session.id), 'A preflight cannot be replayed.'); globalThis.__workbenchM4Preflights.add(session.id);
|
||||
const config = hub.prepare({ ...store.config, mode: 'compatible' }, { type: 'connection', id: connection.id });
|
||||
ensure(config.capacityMode === 'passthrough' && config.executionMode === 'unrestricted', 'Keep default passthrough.');
|
||||
// A single physical request is this experiment's design, not a saved quota.
|
||||
config.modelRouting.maxAttempts = 1;
|
||||
config.modelRouting.members = config.modelRouting.members.map(value => ({ ...value, modelIdentityPolicy: 'require-match',
|
||||
...(input.scope === 'streaming' ? { streaming: true } : {}) }));
|
||||
const tools = input.scope === 'tools' ? [{ type: 'function', function: { name: 'workbench_m4_echo_nonce', description: 'Echo the supplied compatibility nonce. No file or command operation.',
|
||||
parameters: { type: 'object', properties: { nonce: { type: 'string' } }, required: ['nonce'], additionalProperties: false } } }] : [];
|
||||
const prompt = input.scope === 'tools' ? 'Call workbench_m4_echo_nonce exactly once with nonce ' + input.nonce + '. No other task.' : 'Reply exactly: OK ' + input.nonce;
|
||||
const run = store.beginRun(session, prompt, false, config, { forceNewTask: true }), user = store.message(session, 'user', prompt);
|
||||
run.inputMessage = { id: user.id, providerIndex: 0, origin: 'user' }; session.providerMessages.push({ role: 'user', content: prompt });
|
||||
session.status = 'running'; const step = store.beginStep(session, run), planner = runtime.requestPlanner();
|
||||
const controller = new AbortController(), timer = setTimeout(() => controller.abort(Error('Predeclared M4 observation ended; do not replay.')), input.observationMs);
|
||||
let plan, response, diagnostic = null, event, receipts = 0; const wireIds = new Set();
|
||||
try {
|
||||
plan = planner.prepare(session, run, config, { system: 'Connection compatibility preflight only. Follow the user exactly. Do not perform project work.', tools, stepId: step.id });
|
||||
const variants = plan.body.kind === 'model-route-plan' ? plan.body.variants.map(value => value.body) : [plan.body];
|
||||
ensure(variants.length === 1, 'A frozen preflight must have one route.');
|
||||
for (const body of variants) {
|
||||
const actual = Object.fromEntries(['max_tokens', 'max_output_tokens', 'max_completion_tokens'].filter(key => Object.hasOwn(body, key)).map(key => [key, body[key]]));
|
||||
ensure(JSON.stringify(actual) === JSON.stringify(input.expectedOutput.fields), 'Output fields must equal explicit settings or the Messages protocol requirement.');
|
||||
}
|
||||
response = await provider.complete(config, plan.messages, project, controller.signal, progress => {
|
||||
if (progress.consumption) { receipts += Number(store.recordConsumption(session, run, { ...progress.consumption, requestId: plan.record.id }, plan.record.tokenReservation?.reservedTokens ?? plan.record.estimatedInputTokens)); return; }
|
||||
if (progress.phase === 'running' && progress.wireAttemptId) { wireIds.add(progress.wireAttemptId); ensure(wireIds.size === 1, 'No retry permitted.'); }
|
||||
event ??= store.event(session, { kind: 'model', source: 'model_connection', label: 'M4 单次兼容预检', phase: 'running', runId: run.id, contextPlanId: plan.record.id, startedAt: new Date().toISOString() });
|
||||
for (const key of ['modelAttemptId', 'connectionId', 'modelIdentity', 'requestUrl', 'diagnostic', 'stream', 'phase']) if (progress[key] !== undefined) event[key] = progress[key];
|
||||
plan.record.requestAttempts = wireIds.size; step.modelAttempts = wireIds.size; store.save(session);
|
||||
}, plan.record, tools, { attempts: 1, maxBudgetAttempts: 1 });
|
||||
planner.finish(session, plan.record, { response, attempts: wireIds.size });
|
||||
// An observed tool call is not executed or inserted as an unpaired call.
|
||||
if (!response.message.tool_calls?.length) session.providerMessages.push(response.message);
|
||||
store.message(session, 'assistant', response.message.content || '已收到原工具调用,预检不执行工具。'); session.status = 'completed';
|
||||
} catch (cause) {
|
||||
diagnostic = cause.diagnostic ?? { source: 'workbench', origin: 'local', stage: 'preflight', kind: controller.signal.aborted ? 'observation_stop' : 'runtime', message: provider.redact(cause.message) };
|
||||
session.status = 'failed'; session.lastFailure = { ...diagnostic, runId: run.id }; store.message(session, 'system', diagnostic.message);
|
||||
if (plan) planner.finish(session, plan.record, { error: new Error(diagnostic.message), attempts: wireIds.size });
|
||||
} finally {
|
||||
clearTimeout(timer); if (event) { event.phase = diagnostic ? 'failed' : 'completed'; event.finishedAt = new Date().toISOString(); }
|
||||
store.finishStep(session, step, diagnostic ? 'failed' : 'completed'); store.finishRun(session, run, diagnostic ? diagnostic.kind : 'm4_compatibility_preflight');
|
||||
}
|
||||
const ledger = hub.ledger.page({ filters: { sessionId: session.id }, limit: 2 }); ensure(ledger.total <= 1 && ledger.items.length === ledger.total, 'Single-request contract violated.');
|
||||
const calls = response?.message?.tool_calls ?? []; let contentPassed = false;
|
||||
try { contentPassed = input.scope === 'tools' ? calls.length === 1 && calls[0].function.name === 'workbench_m4_echo_nonce'
|
||||
&& JSON.parse(calls[0].function.arguments).nonce === input.nonce : !calls.length && response?.message?.content?.trim() === 'OK ' + input.nonce; } catch {}
|
||||
return { sessionId: session.id, taskId: run.taskId, runId: run.id, scope: input.scope, status: session.status, diagnostic,
|
||||
attempts: ledger.total, wireStarts: wireIds.size, receipts, actualToolsExecuted: run.toolCalls, contentPassed,
|
||||
outputFieldsSource: input.expectedOutput.source, savedConfigurationChanged: false, message: response?.message ?? null, usage: run.usage };
|
||||
}
|
||||
@@ -0,0 +1,42 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { execFileSync, spawnSync } from 'node:child_process';
|
||||
import { gitEnvironment } from '../../src/core/git-process.mjs';
|
||||
import { prepareProject as prepareSecondary, python, specs as secondarySpecs } from '../live-ae/project.mjs';
|
||||
import { sha } from './common.mjs';
|
||||
|
||||
// The old project module supplies only preparation/check helpers. It does not
|
||||
// load an old claim, connection, session, model result or real-batch entry point.
|
||||
export { python };
|
||||
export const specs = { node: { baseline: 'benchmarks/live-delivery-v1/node', allowed: ['src/amount.mjs', 'src/summary.mjs', 'src/report.mjs'], tests: 9 }, ...secondarySpecs };
|
||||
export function prepareProject(kind, output, save) {
|
||||
if (kind !== 'node') return prepareSecondary(kind, output, save);
|
||||
const spec = specs.node, sourceRoot = path.join(output, 'project'); assert.equal(fs.existsSync(sourceRoot), false);
|
||||
fs.mkdirSync(output, { recursive: true }); fs.cpSync(spec.baseline, sourceRoot, { recursive: true, errorOnExist: true });
|
||||
const git = args => execFileSync('git', ['--no-pager', '--no-optional-locks', '-C', sourceRoot, ...args], { encoding: 'utf8', windowsHide: true, env: gitEnvironment() });
|
||||
fs.writeFileSync(path.join(sourceRoot, 'README.md'), '# M4 independent Node baseline\nRepair the three src modules using the original nine tests. Preserve owner bytes and their staging.\n', { flag: 'wx' });
|
||||
fs.writeFileSync(path.join(sourceRoot, '.gitignore'), 'dist/\n.awb-*\nnode_modules/\n', { flag: 'wx' });
|
||||
for (const name of ['owner-staged.txt', 'owner-unstaged.txt']) fs.writeFileSync(path.join(sourceRoot, name), 'Owner baseline\n', { flag: 'wx' });
|
||||
git(['init', '--initial-branch=main']); git(['config', 'core.autocrlf', 'false']); git(['config', 'user.name', 'Workbench QA']); git(['config', 'user.email', 'qa@workbench.invalid']);
|
||||
git(['add', '--', '.']); git(['commit', '--no-gpg-sign', '-m', 'Seal M4 independent Node original baseline']); const initialHead = git(['rev-parse', 'HEAD']).trim();
|
||||
fs.writeFileSync(path.join(sourceRoot, 'owner-staged.txt'), 'Owner staged change\n'); git(['add', '--', 'owner-staged.txt']);
|
||||
fs.writeFileSync(path.join(sourceRoot, 'owner-staged.txt'), 'Owner unstaged over original staging\n');
|
||||
fs.writeFileSync(path.join(sourceRoot, 'owner-unstaged.txt'), 'Owner unstaged change\n'); fs.writeFileSync(path.join(sourceRoot, 'owner-untracked.txt'), 'Owner untracked change\n', { flag: 'wx' });
|
||||
const all = git(['ls-files', '-z']).split('\0').filter(Boolean), protectedNames = all.filter(name => !spec.allowed.includes(name)).concat('owner-untracked.txt');
|
||||
const protectedBytes = Object.fromEntries(protectedNames.map(name => [name, fs.readFileSync(path.join(sourceRoot, name))]));
|
||||
const index = git(['ls-files', '--stage', '--', ...protectedNames]);
|
||||
const protect = () => {
|
||||
for (const [name, bytes] of Object.entries(protectedBytes)) assert.deepEqual(fs.readFileSync(path.join(sourceRoot, name)), bytes, 'Protected file: ' + name);
|
||||
assert.equal(git(['ls-files', '--stage', '--', ...protectedNames]), index, 'All protected original index entries');
|
||||
};
|
||||
const check = (name, args, prefix) => {
|
||||
const execution = spawnSync(process.execPath, args, { cwd: sourceRoot, windowsHide: true, encoding: 'utf8', env: gitEnvironment(), maxBuffer: 16 * 1024 * 1024 });
|
||||
save(prefix + '/' + name + '.txt', (execution.stdout ?? '') + (execution.stderr ?? ''));
|
||||
const stats = name === 'test' ? Object.fromEntries(['tests', 'pass', 'fail', 'cancelled', 'skipped', 'todo'].map(key => [key, Number(execution.stdout?.match(new RegExp('^# ' + key + ' (\\d+)$', 'm'))?.[1] ?? NaN)])) : null;
|
||||
return { status: execution.status, signal: execution.signal, error: execution.error?.message ?? null, stats, stdoutSha256: sha(execution.stdout ?? ''), stderrSha256: sha(execution.stderr ?? '') };
|
||||
};
|
||||
const checks = prefix => { const result = { test: check('test', ['--test', '--test-reporter=tap', 'test/expense.test.mjs'], prefix), build: check('build', ['scripts/build.mjs'], prefix) }; save(prefix + '/result.json', result); protect(); return result; };
|
||||
save('source-before.json', { initialHead, status: git(['status', '--porcelain=v1']), protectedIndex: index, files: Object.fromEntries([...all, 'owner-untracked.txt'].map(name => [name, sha(fs.readFileSync(path.join(sourceRoot, name)))])) });
|
||||
return { spec, sourceRoot, git, initialHead, protect, checks };
|
||||
}
|
||||
@@ -0,0 +1,50 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { DatabaseSync } from 'node:sqlite';
|
||||
import { digest } from '../retained-host-state.mjs';
|
||||
import { root, read, save, client, sha, captureSession } from './common.mjs';
|
||||
|
||||
// Explicitly record the independently discovered greeting; do not automatically
|
||||
// accept every new session, repeat its request, or add it to engineering results.
|
||||
const sessionId = '8f896f90-50f2-4d07-8da0-3b98832c6df5';
|
||||
const attemptId = '821c6dc9-7e52-4e41-a1d9-36b4de3ecbdb';
|
||||
const prefix = 'external-observations/' + sessionId, frozen = read('frozen.json');
|
||||
assert.equal(fs.existsSync(path.join(root, prefix)), false, 'Do not overwrite an external observation.');
|
||||
assert.equal(frozen.baseline.database.sessionIds.includes(sessionId), false);
|
||||
const { api, close } = await client();
|
||||
try {
|
||||
const host = await api('/host/status');
|
||||
assert.equal(host.identity.pid, frozen.host.identity.pid); assert.equal(host.activity.active, false);
|
||||
const captured = await captureSession(api, sessionId, prefix), state = captured.state;
|
||||
assert.ok(frozen.originalProjects.some(value => value.id === state.projectId));
|
||||
assert.equal(state.status, 'completed'); assert.equal(state.execution.run.resumeOf, null);
|
||||
assert.deepEqual(state.messages.filter(value => value.role === 'user').map(value => value.content), ['你好']);
|
||||
assert.equal(state.changes.length, 0); assert.equal(state.execution.deliveries.length, 0);
|
||||
assert.equal(state.events.filter(value => value.kind === 'tool').length, 0);
|
||||
assert.deepEqual(captured.rows.map(value => value.id), [attemptId]);
|
||||
const db = new DatabaseSync(path.join(host.identity.dataRoot, 'workbench.sqlite'), { readOnly: true }), tables = {};
|
||||
db.exec('BEGIN');
|
||||
try {
|
||||
for (const table of ['sessions', 'events', 'records', 'model_attempts', 'outbox']) {
|
||||
const rows = table === 'outbox'
|
||||
? db.prepare('SELECT outbox.* FROM outbox JOIN events USING(seq) WHERE events.session_id=? ORDER BY outbox.seq').all(sessionId)
|
||||
: db.prepare('SELECT * FROM ' + table + ' WHERE ' + (table === 'sessions' ? 'id' : 'session_id') + '=? ORDER BY rowid').all(sessionId);
|
||||
const sql = db.prepare("SELECT sql FROM sqlite_schema WHERE type='table' AND name=?").get(table).sql;
|
||||
assert.equal(sha(sql), frozen.baseline.database.tables[table].schemaSha256);
|
||||
tables[table] = { count: rows.length, schemaSha256: sha(sql), rows: rows.map(row => ({ id: row.id ?? row.seq, ...(row.kind ? { kind: row.kind } : {}), sha256: digest(row) })) };
|
||||
}
|
||||
} finally { db.exec('ROLLBACK'); db.close(); }
|
||||
const index = { schemaVersion: 1, observedAt: new Date().toISOString(), sessionId, projectId: state.projectId,
|
||||
taskId: state.execution.task.id, runId: state.execution.run.id, title: state.title, scope: 'Independent new greeting in an existing project; attribution to a person is not asserted.',
|
||||
m4Engineering: false, requests: captured.total, usage: state.execution.task.usage,
|
||||
attempts: captured.rows.map(value => ({ id: value.id, status: value.status, usageStatus: value.usageStatus, modelIdentity: value.modelIdentity, requestSha256: value.requestSha256, responseSha256: value.responseSha256 })),
|
||||
exports: { id: sessionId, project: digest(captured.exported.project), session: digest(captured.exported.session), execution: digest(captured.exported.execution) },
|
||||
artifacts: state.execution.artifacts.map(value => ({ id: value.id, path: value.relative, bytes: value.bytes, sha256: value.sha256 })),
|
||||
tables, captures: captured.captures, resumedHistoricalTask: false, toolCalls: 0, changes: 0, deliveries: 0,
|
||||
keysReadOrResaved: false, requestsReplayed: false, productOrProjectMutation: false };
|
||||
const relative = prefix + '/index.json'; save(relative, index);
|
||||
save('external-observations/declared.json', { schemaVersion: 1, sources: [{ path: relative, sha256: sha(fs.readFileSync(path.join(root, relative))) }] });
|
||||
console.log(JSON.stringify({ sessionId, projectId: state.projectId, requests: index.requests, totalTokens: index.usage.totalTokens,
|
||||
tableCounts: Object.fromEntries(Object.entries(tables).map(([key, value]) => [key, value.count])), artifacts: index.artifacts.length, modelIdentity: index.attempts[0].modelIdentity.status, scope: index.scope }));
|
||||
} finally { await close(); }
|
||||
@@ -0,0 +1,98 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { randomUUID } from 'node:crypto';
|
||||
import { setTimeout as delay } from 'node:timers/promises';
|
||||
import { root, save, read, client, sha, captureSession, connectionDefinition, compareConfig, compareConnectionDefinitions } from './common.mjs';
|
||||
import { prepareProject, python, specs } from './project.mjs';
|
||||
import { verifyLiveReactClient } from '../live-react-client.mjs';
|
||||
|
||||
const kind = process.argv[2]; assert.ok(Object.hasOwn(specs, kind));
|
||||
const frozen = read('frozen.json'), compatibility = read('compatibility/result.json');
|
||||
if (kind !== 'node') assert.equal(read((kind === 'react' ? 'node' : 'react') + '/result.json').success, true, 'The previous engineering baseline must fully pass.');
|
||||
const connectionId = kind === 'node' ? frozen.engineeringPriority.find(id => compatibility.results.some(value => value.connectionId === id && value.engineeringAllowed)) : read('node/result.json').connectionId;
|
||||
assert.equal(fs.existsSync(path.join(root, kind)), false, 'Do not replay an engineering sample.');
|
||||
if (!connectionId) { save(kind + '/result.json', { success: false, executed: false, reason: 'No connection passed the predeclared compatibility gate.', purgeExecuted: false }); process.exit(0); }
|
||||
const { api, boot, close } = await client(); let session, route, selected = false, original, own, failure, result, state, selectionRevision;
|
||||
const observations = [], facts = new Set(), output = path.join(root, kind), store = (file, value) => save(kind + '/' + file, value);
|
||||
const settle = async () => {
|
||||
if (!route) return; state = await api(route);
|
||||
if (['running', 'stopping', 'awaiting_approval'].includes(state.status)) { await api(route + '/stop', {}, 60000); state = await api(route); }
|
||||
assert.ok(!['running', 'stopping', 'awaiting_approval'].includes(state.status), 'Only a quiescent owned task can be independently checked.');
|
||||
};
|
||||
try {
|
||||
const host = await api('/host/status'); assert.equal(host.activity.active, false); assert.equal(host.identity.pid, frozen.host.identity.pid); assert.equal(boot.version, '1.25.13');
|
||||
compareConfig(frozen.config, boot.config); const center = await api('/model-center'); assert.deepEqual(center.selected, frozen.center.selected); assert.deepEqual(center.pools, frozen.center.pools);
|
||||
original = center.connections.find(value => value.id === connectionId);
|
||||
assert.deepEqual(connectionDefinition(original), connectionDefinition(frozen.connections.find(value => value.id === connectionId)));
|
||||
store('claim.json', { id: randomUUID(), kind, connection: original, config: boot.config, source: specs[kind].baseline, allowed: specs[kind].allowed,
|
||||
observationMs: 600000, noNewFactMs: 180000, implementationSupplied: false, supervisorContinuation: false, localTokenCaps: false, identityPolicySaved: false });
|
||||
own = prepareProject(kind, output, store); const baseline = own.checks('baseline'); assert.equal(baseline.test.stats.tests, own.spec.tests); assert.ok(baseline.test.stats.fail > 0); assert.equal(baseline.build.status, 0);
|
||||
const project = await api('/projects', { root: own.sourceRoot }); session = await api('/sessions', { projectId: project.id }); route = '/sessions/' + session.id; store('owned.json', { project, session });
|
||||
const selection = await api('/model-center/select', { type: 'connection', id: connectionId }); selected = true; selectionRevision = selection.config.configRevision;
|
||||
const goals = {
|
||||
node: '请独立修复这个Node费用报表工程,使原九项测试和build通过,并完成本地Git交付。只修改src/amount.mjs、src/summary.mjs、src/report.mjs;按原测试理解金额、币种汇总和报表要求,不修改原测试、package配置、构建脚本、README或owner文件,保留原有暂存/未暂存/未跟踪状态。请自行读取、修改、执行检查并提交。',
|
||||
react: '独立完成这个React/TypeScript看板工程README要求,修复状态筛选、去空白搜索、实际统计、按目标任务切换与空态,保留既有样式和布局。只修改src/query.ts、src/summary.ts、src/transition.ts、src/App.tsx。自行读取源码及原九项测试、运行test/typecheck/build并完成仅这四文件的本地Git提交。依赖已按原锁离线准备;不可改原测试、配置、锁、其它源码/样式、README和owner文件,保留原暂存/未暂存/未跟踪状态。',
|
||||
python: '独立完成这个Python任务报表工程README的规范化、汇总与渲染契约。只修改normalize.py、summary.py、render.py。自行读取源码及原十项测试,运行pytest与源码编译并完成仅这三文件的本地Git提交。现成解释器为 ' + python + ',已装pytest,无需安装依赖。不可改原测试、配置、README和owner文件,保留原暂存/未暂存/未跟踪状态。',
|
||||
};
|
||||
const prompt = goals[kind] + ' 当前为完全访问权限,可用工程工具或终端自动执行,无需独立审查,不要只解释方案或口头完成,不推送远端。';
|
||||
store('goal.txt', prompt); await api(route + '/messages', { content: prompt }); const started = Date.now(); let novel = started, signature = '';
|
||||
while (true) {
|
||||
state = await api(route); own.protect();
|
||||
for (const event of state.events.filter(value => value.kind === 'tool' && value.phase === 'completed' && ['read_file', 'read_artifact', 'list_files', 'query_code', 'search_files'].includes(value.tool))) {
|
||||
const key = event.tool + JSON.stringify(event.input) + sha(event.output ?? ''); if (!facts.has(key)) { facts.add(key); novel = Date.now(); }
|
||||
}
|
||||
const current = JSON.stringify({ files: own.spec.allowed.map(name => sha(fs.readFileSync(path.join(own.sourceRoot, name)))), checks: state.execution.checks.map(value => ({ status: value.status, source: value.sourceFingerprint })) });
|
||||
if (current !== signature) { signature = current; novel = Date.now(); }
|
||||
const observation = { kind, status: state.status, requests: state.execution.task?.usage.requests, tokens: state.execution.task?.usage.totalTokens, facts: facts.size,
|
||||
changes: state.changes.length, checks: state.execution.checks.map(value => value.status), lastTool: state.events.filter(value => value.kind === 'tool').at(-1)?.tool };
|
||||
if (JSON.stringify(observation) !== JSON.stringify(observations.at(-1))) { observations.push(observation); console.log(JSON.stringify(observation)); }
|
||||
if (!['running', 'stopping', 'awaiting_approval'].includes(state.status)) break;
|
||||
if (Date.now() - started > 600000 || Date.now() - novel > 180000) { failure = { reason: 'experiment_observation_ended', elapsedMs: Date.now() - started, noNewFactMs: Date.now() - novel, productBudgetChanged: false }; break; }
|
||||
await delay(750);
|
||||
}
|
||||
await settle(); own.protect(); const final = own.checks('independent-final'); let browser = null;
|
||||
if (kind === 'react' && Object.values(final).every(value => value.status === 0)) {
|
||||
try { browser = await verifyLiveReactClient(own.sourceRoot, path.join(output, 'browser')); } catch (error) { browser = { success: false, error: error.stack }; } store('browser/result.json', browser);
|
||||
}
|
||||
const commit = own.git(['rev-parse', 'HEAD']).trim(), paths = own.git(['diff', '--name-only', own.initialHead, commit]).trim().split(/\r?\n/).filter(Boolean).sort();
|
||||
const committedCurrent = own.spec.allowed.every(name => sha(Buffer.from(own.git(['show', commit + ':' + name]))) === sha(fs.readFileSync(path.join(own.sourceRoot, name))));
|
||||
const otherUntracked = own.git(['ls-files', '--others', '--exclude-standard']).trim().split(/\r?\n/).filter(name => name && name !== 'owner-untracked.txt');
|
||||
result = { success: !failure && state.status === 'completed' && Object.values(final).every(value => value.status === 0) && final.test.stats.tests === own.spec.tests && final.test.stats.pass === own.spec.tests
|
||||
&& commit !== own.initialHead && JSON.stringify(paths) === JSON.stringify(own.spec.allowed.slice().sort()) && committedCurrent && !otherUntracked.length && (kind !== 'react' || browser?.success === true),
|
||||
kind, executed: true, sessionId: session.id, projectId: project.id, connectionId, commit, initialHead: own.initialHead, committedPaths: paths, committedCurrent, otherUntracked,
|
||||
final, browser, failure: failure ?? null, productStopReason: state.execution.run?.stopReason ?? null, ownerBytesAndIndexPreserved: true, supervisorContinuations: 0, localTokenCapsChanged: false };
|
||||
} catch (error) { result = { success: false, kind, executed: !!session, failure: { error: error.stack, noRetry: true }, sessionId: session?.id, connectionId }; }
|
||||
finally {
|
||||
const errors = [], restoration = {}; const guard = async (label, fn) => { try { await fn(); } catch (error) { errors.push({ label, error: error.stack }); } };
|
||||
if (route) {
|
||||
await guard('settle-owned-task', settle);
|
||||
await guard('capture-owned-task', async () => {
|
||||
const captured = await captureSession(api, session.id, kind);
|
||||
const identitiesPassed = captured.total > 0 && captured.rows.every(value => value.connectionId === connectionId && value.status === 'completed' && value.modelIdentity?.complete && value.modelIdentity.status === 'matched')
|
||||
&& captured.state.execution.task.usage.unknownRequests === 0 && captured.state.execution.task.usage.requests === captured.total;
|
||||
result = { ...result, success: result.success && identitiesPassed, identitiesPassed, requests: captured.total, usage: captured.state.execution.task.usage, taskId: captured.state.execution.task?.id,
|
||||
runId: captured.state.execution.run?.id, captures: captured.captures, internalDeliveries: captured.state.execution.deliveries.map(value => ({ id: value.id, status: value.status, commit: value.commit })),
|
||||
reportArtifacts: captured.state.execution.artifacts.filter(value => /^delivery-report/.test(value.kind)).map(value => ({ id: value.id, kind: value.kind, sha256: value.sha256 })) };
|
||||
store('git-status.txt', own.git(['status', '--porcelain=v1'])); store('git-log.txt', own.git(['log', '--format=fuller', '--stat', own.initialHead + '..HEAD']));
|
||||
});
|
||||
}
|
||||
await guard('restore-selection', async () => {
|
||||
if (!selected) return;
|
||||
const center = await api('/model-center'), now = await api('/bootstrap');
|
||||
if (center.selected.type === 'connection' && center.selected.id === connectionId && now.config.configRevision === selectionRevision) {
|
||||
await api('/model-center/select', frozen.center.selected); restoration.selection = frozen.center.selected;
|
||||
} else restoration.externalSelectionOrConfigChangePreserved = true;
|
||||
});
|
||||
await guard('verify-config-and-definitions', async () => {
|
||||
restoration.config = (await api('/bootstrap')).config;
|
||||
if (!restoration.externalSelectionOrConfigChangePreserved) restoration.configComparison = compareConfig(frozen.config, restoration.config);
|
||||
const center = await api('/model-center');
|
||||
restoration.connectionDefinitions = compareConnectionDefinitions(frozen.center, center, frozen.config, restoration.config);
|
||||
assert.deepEqual(center.pools, frozen.center.pools); restoration.pools = 'exact';
|
||||
});
|
||||
restoration.keysReadOrResaved = false; restoration.errors = errors; if (errors.length) result = { ...result, success: false, cleanupErrors: errors };
|
||||
store('restoration.json', restoration); store('observations.json', observations); store('result.json', result); await close();
|
||||
}
|
||||
console.log(JSON.stringify({ kind, success: result.success, requests: result.requests, usage: result.usage, commit: result.commit, failure: result.failure, cleanupErrors: result.cleanupErrors }));
|
||||
if (!result.success) process.exitCode = 1;
|
||||
@@ -0,0 +1,46 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { root, read, save, sha } from './common.mjs';
|
||||
|
||||
if (process.argv.includes('--verify')) {
|
||||
const manifest = read('MANIFEST.json'), receipt = read('seal.json');
|
||||
assert.equal(sha(fs.readFileSync(path.join(root, 'MANIFEST.json'))), receipt.manifestSha256);
|
||||
for (const entry of manifest.entries) {
|
||||
const bytes = fs.readFileSync(path.join(root, entry.path));
|
||||
assert.equal(bytes.length, entry.bytes); assert.equal(sha(bytes), entry.sha256, entry.path);
|
||||
}
|
||||
assert.equal(manifest.entries.length, receipt.files);
|
||||
assert.equal(manifest.entries.reduce((sum, value) => sum + value.bytes, 0), receipt.bytes);
|
||||
console.log(JSON.stringify({ verified: true, files: receipt.files, bytes: receipt.bytes, manifestSha256: receipt.manifestSha256 }));
|
||||
} else {
|
||||
assert.equal(fs.existsSync(path.join(root, 'MANIFEST.json')), false, 'Never overwrite a sealed M4 root.');
|
||||
const final = read('final-summary.json'); assert.equal(final.stage, 'final'); assert.equal(final.audit.success, true);
|
||||
assert.equal(final.engineering.length, 3); assert.ok(final.engineering.every(value => value.success));
|
||||
const entries = [], excluded = [];
|
||||
const walk = relative => {
|
||||
for (const entry of fs.readdirSync(path.join(root, relative), { withFileTypes: true })) {
|
||||
const file = path.posix.join(relative.replaceAll('\\', '/'), entry.name);
|
||||
if (entry.isDirectory() && entry.name === 'node_modules') {
|
||||
const lockFile = path.join(root, relative, 'package-lock.json'); assert.ok(fs.existsSync(lockFile));
|
||||
excluded.push({ path: file, scope: 'Reconstructible installed dependencies only; original package-lock and npm-ci log included.', lockSha256: sha(fs.readFileSync(lockFile)) });
|
||||
continue;
|
||||
}
|
||||
assert.equal(entry.isSymbolicLink(), false, 'Evidence traversal does not follow links: ' + file);
|
||||
if (entry.isDirectory()) walk(file);
|
||||
else if (entry.isFile()) { const bytes = fs.readFileSync(path.join(root, file)); entries.push({ path: file, bytes: bytes.length, sha256: sha(bytes) }); }
|
||||
else assert.fail('Unsupported evidence entry: ' + file);
|
||||
}
|
||||
};
|
||||
walk(''); entries.sort((a, b) => a.path.localeCompare(b.path, 'en'));
|
||||
const manifest = { schemaVersion: 1, batch: 'm4-live-baseline-2026-11-11', sealedAt: new Date().toISOString(),
|
||||
scope: 'Original observations, failures, captures, audits, projects and Git, browser PNGs, machine summaries and retained ancillary observations. No real credential extraction.',
|
||||
entries, excluded, selfAndReceipt: 'MANIFEST.json and seal.json created after entry hashing; their integrity is identified by the receipt.' };
|
||||
save('MANIFEST.json', manifest);
|
||||
const bytes = fs.readFileSync(path.join(root, 'MANIFEST.json')), receipt = { files: entries.length, bytes: entries.reduce((sum, value) => sum + value.bytes, 0),
|
||||
manifestSha256: sha(bytes), manifestBytes: bytes.length, excluded, modelRequests: final.requests, reportedTotalTokens: final.reportedTotalTokens,
|
||||
unknownRequests: final.unknownRequests, externalRequests: final.audit.externalModelRequests, keysReadOrResaved: false, oldSamplesReplayed: false, purgeExecuted: false };
|
||||
save('seal.json', receipt);
|
||||
fs.writeFileSync(path.resolve('docs/m4-live-seal-2026-11-11.json'), JSON.stringify(receipt, null, 2) + '\n', { flag: 'wx' });
|
||||
console.log(JSON.stringify(receipt));
|
||||
}
|
||||
@@ -0,0 +1,76 @@
|
||||
// Build a read-only source index from already sealed AD evidence. No model,
|
||||
// task, command, approval, retry, Git write or native-v2 envelope is produced.
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { execFileSync } from 'node:child_process';
|
||||
import { createHash } from 'node:crypto';
|
||||
|
||||
const repository = process.cwd(), evidence = path.resolve('.qa/live-validation-ad');
|
||||
const output = path.resolve('.qa/coordination-readonly-2026-11-11-engineering-ad-r2');
|
||||
assert.equal(fs.existsSync(output), false, 'Do not overwrite a source handoff.');
|
||||
const sha = bytes => createHash('sha256').update(bytes).digest('hex');
|
||||
const parse = file => JSON.parse(fs.readFileSync(path.join(evidence, file), 'utf8'));
|
||||
const originalManifestBytes = fs.readFileSync(path.join(evidence, 'FINAL-MANIFEST.json'));
|
||||
const originalManifest = JSON.parse(originalManifestBytes), members = new Map(originalManifest.files.map(value => [value.path, value]));
|
||||
const result = parse('node/result.json'), state = parse('node/session.json'), frozen = parse('current.json');
|
||||
const ids = { projectId: state.projectId, sessionId: state.id, taskId: state.execution.task.id, runId: state.execution.run.id };
|
||||
assert.equal(ids.taskId, result.taskId); assert.equal(ids.sessionId, result.sessionId);
|
||||
assert.equal(state.execution.run.taskId, ids.taskId);
|
||||
fs.mkdirSync(output); const sources = [], missing = [];
|
||||
const addSealed = relative => {
|
||||
const member = members.get(relative); assert.ok(member, 'Original manifest member: ' + relative);
|
||||
const file = path.join(evidence, relative), bytes = fs.readFileSync(file);
|
||||
assert.equal(bytes.length, member.bytes); assert.equal(sha(bytes), member.sha256);
|
||||
sources.push({ path: file, bytes: bytes.length, sha256: sha(bytes), origin: 'original-ad-seal', originalRelative: relative });
|
||||
};
|
||||
for (const file of ['node/goal.txt', 'node/contract.json', 'node/source-before.json', 'node/result.json', 'node/session.json', 'node/export.json', 'node/attempts.json', 'node/git-log.txt', 'node/git-status.txt',
|
||||
'node/baseline/test.txt', 'node/baseline/build.txt', 'node/baseline/result.json', 'node/independent-final/test.txt', 'node/independent-final/build.txt', 'node/independent-final/result.json',
|
||||
'node/expense-project/package.json', 'node/expense-project/test/expense.test.mjs', 'node/expense-project/scripts/build.mjs', ...result.committedPaths.map(file => 'node/expense-project/' + file), ...result.captures.map(value => value.path)]) addSealed(file);
|
||||
const plan = execFileSync('git', ['show', frozen.driverCommit + ':docs/LIVE-VALIDATION-AD-PLAN.md'], { windowsHide: true });
|
||||
const planPath = path.join(output, 'original-predeclared-plan.md'); fs.writeFileSync(planPath, plan, { flag: 'wx' });
|
||||
sources.push({ path: planPath, bytes: plan.length, sha256: sha(plan), origin: 'original-git-blob', commit: frozen.driverCommit, repositoryPath: 'docs/LIVE-VALIDATION-AD-PLAN.md' });
|
||||
const artifactIds = new Set(state.execution.checks.flatMap(value => (value.results ?? []).map(row => row.logArtifactId)).filter(Boolean));
|
||||
for (const value of state.execution.artifacts) if (/^delivery-report/.test(value.kind)) artifactIds.add(value.id);
|
||||
const originalArtifacts = [];
|
||||
for (const id of artifactIds) {
|
||||
const artifact = state.execution.artifacts.find(value => value.id === id);
|
||||
if (!artifact) { missing.push({ artifactId: id, reason: 'Check log descriptor absent from the sealed session artifact preview; independent-final logs remain in the read scope.' }); continue; }
|
||||
for (const [key, expected] of Object.entries(ids)) assert.equal(artifact[key], expected, 'Artifact ownership: ' + key);
|
||||
assert.match(artifact.relative, /^[a-f0-9]{2}\/[a-f0-9]{64}\.blob$/);
|
||||
const originalPath = path.join(frozen.host.identity.dataRoot, 'artifacts', artifact.relative);
|
||||
if (!fs.existsSync(originalPath)) { missing.push({ artifactId: id, kind: artifact.kind, reason: 'Original artifact file unavailable', expectedSha256: artifact.sha256 }); continue; }
|
||||
const bytes = fs.readFileSync(originalPath); assert.equal(bytes.length, artifact.bytes); assert.equal(sha(bytes), artifact.sha256);
|
||||
const target = path.join(output, id + (artifact.mediaType === 'application/json' ? '.json' : '.txt'));
|
||||
fs.writeFileSync(target, bytes, { flag: 'wx' });
|
||||
sources.push({ path: target, bytes: bytes.length, sha256: sha(bytes), origin: 'byte-exact-copy-of-original-owned-artifact', artifactId: id, kind: artifact.kind });
|
||||
originalArtifacts.push({ ...artifact, sharedPath: target, originalPath, copiedBytesExact: true });
|
||||
}
|
||||
const report = originalArtifacts.find(value => value.kind === 'delivery-report');
|
||||
const reportBody = report ? JSON.parse(fs.readFileSync(report.sharedPath, 'utf8')) : null;
|
||||
const metadata = {
|
||||
schemaVersion: 1, kind: 'workbench-readonly-engineering-source-index', generatedAt: new Date().toISOString(),
|
||||
sourceName: 'agent-workbench/ad/node/' + ids.taskId + '/' + ids.runId, ids, originalVersion: '1.25.8', originalDriverCommit: frozen.driverCommit,
|
||||
purpose: 'Read-only acceptance of an existing independent Node task. No incoming restricted delegation or new M4 task is represented.',
|
||||
goalPath: path.join(evidence, 'node/goal.txt'), predeclaredPlanPath: planPath, originalTaskContract: state.execution.task.contract,
|
||||
originalStructuredAcceptance: state.execution.task.acceptance,
|
||||
originalReportArtifact: report ? { id: report.id, path: report.sharedPath, sha256: report.sha256, bytes: report.bytes, status: reportBody?.status ?? null } : null,
|
||||
reportMetadata: originalArtifacts.filter(value => /^delivery-report/.test(value.kind)), checkLogMetadata: originalArtifacts.filter(value => artifactIds.has(value.id) && !/^delivery-report/.test(value.kind)),
|
||||
engineering: { sourceRoot: result.sourceRoot, initialHead: result.initialHead, commit: result.commit, committedPaths: result.committedPaths, committedCurrent: result.committedCurrent,
|
||||
actualIndependentChecks: result.final, ownerBytesAndIndexPreserved: result.allOriginalOwnerBytesAndIndexPreserved, supervisorContinuations: result.supervisionContinuations },
|
||||
originalOutcome: { independentEngineeringSuccess: result.success, sessionStatus: state.status, taskStatus: state.execution.task.status, runStatus: state.execution.run.status,
|
||||
terminalGitObserved: true, nativeDeliveryRows: state.execution.deliveries.length, internalDeliveryReportStatus: reportBody?.status ?? 'missing',
|
||||
taskVerification: state.execution.task.verification, productStopReason: result.productStopReason, failure: result.failure, usage: result.usage,
|
||||
missing, batchNetworkAndClaudeErrors: 'Preserved separately in QA-LIVE-VALIDATION-AD.md; the Node subset does not replace the whole-batch outcome.' },
|
||||
readiness: { originalTaskAndRunAvailable: true, originalReportsCopied: !!report, nativeEngineeringEvidenceV2Envelope: false, restrictedDelegationReceiver: 'not_established_by_this_source',
|
||||
originalKeyLookup: 'not_established_by_this_source', currentM4Engineering: 'not_dispatched_at_source_index_creation', allConsumerContractFieldsSatisfied: false,
|
||||
gaps: ['No producer-native mnemo.engineering-evidence.v2 envelope or receipt is claimed.', 'The original task structured acceptance array is empty; the raw goal, predeclared plan and original tests carry its conditions.', 'Original terminal Git success and internal in_progress report remain distinct.'] },
|
||||
readScope: { mode: 'read-only-existing-evidence', allowedFiles: sources, originalManifestPath: path.join(evidence, 'FINAL-MANIFEST.json'), originalManifestSha256: sha(originalManifestBytes),
|
||||
excludes: ['Any credential value or credential store', 'Whole formal data directory or SQLite database', 'Other user projects or unrelated sessions', 'Replay, retry, execution, write or native-v2 certification'] },
|
||||
actions: { realModelRequests: 0, engineeringDispatches: 0, gitWrites: 0, originalArtifactsChanged: false, keysReadOrResaved: false, messagesSentToOtherThreads: 0 },
|
||||
};
|
||||
const metadataPath = path.join(repository, 'docs/integration/WORKBENCH-AD-NODE-READONLY-SOURCE-2026-11-11.json');
|
||||
fs.writeFileSync(metadataPath, JSON.stringify(metadata, null, 2) + '\n', { flag: 'wx' });
|
||||
const metadataBytes = fs.readFileSync(metadataPath);
|
||||
fs.writeFileSync(path.join(output, 'SOURCE-MANIFEST.json'), JSON.stringify({ schemaVersion: 1, metadataPath, metadataSha256: sha(metadataBytes), sources, originalArtifacts, missing }, null, 2) + '\n', { flag: 'wx' });
|
||||
console.log(JSON.stringify({ metadataPath, output, sourceName: metadata.sourceName, reportStatus: metadata.originalOutcome.internalDeliveryReportStatus, terminalGit: result.commit, nativeV2Envelope: false, allowedFiles: sources.length, missing: missing.length, realModelRequests: 0 }));
|
||||
@@ -0,0 +1,99 @@
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import assert from 'node:assert/strict';
|
||||
import { root, read, save, sha } from './common.mjs';
|
||||
|
||||
const stage = process.argv[2], auditPrefix = process.argv[3];
|
||||
assert.ok(['node', 'react', 'python', 'final'].includes(stage));
|
||||
assert.match(auditPrefix ?? '', new RegExp('^audits/' + stage + '-[a-f0-9-]{36}$'));
|
||||
const audit = read(auditPrefix + '/result.json'); assert.equal(audit.success, true);
|
||||
const source = relative => {
|
||||
const bytes = fs.readFileSync(path.join(root, relative));
|
||||
return { path: relative, bytes: bytes.length, sha256: sha(bytes) };
|
||||
};
|
||||
const compatibility = read('compatibility/result.json');
|
||||
const compatibilitySummary = {
|
||||
requests: compatibility.requests, unknownRequests: compatibility.results.flatMap(value => value.preflights).filter(value => value.usageStatus !== 'reported').length,
|
||||
reportedTotalTokens: compatibility.results.flatMap(value => value.preflights).reduce((sum, value) => sum + (value.usage?.total_tokens ?? 0), 0),
|
||||
source: source('compatibility/result.json'),
|
||||
connections: compatibility.results.map(value => ({ connectionId: value.connectionId, name: value.name, model: value.model, protocol: value.protocol,
|
||||
endpointKind: value.endpointKind, network: value.network, catalog: value.catalog, engineeringAllowed: value.engineeringAllowed,
|
||||
sources: ['network.json', 'catalog.json', 'result.json'].filter(file => fs.existsSync(path.join(root, 'compatibility', value.connectionId, file)))
|
||||
.map(file => source('compatibility/' + value.connectionId + '/' + file)),
|
||||
preflights: value.preflights.map(row => ({ scope: row.scope, passed: row.passed, sessionId: row.sessionId, taskId: row.taskId, runId: row.runId,
|
||||
identity: row.identity, status: row.status, usageStatus: row.usageStatus, usage: row.usage, actualToolsExecuted: row.actualToolsExecuted,
|
||||
streaming: row.streaming, wirePassed: row.wirePassed, outputFields: row.outputFields, expectedOutput: row.expectedOutput,
|
||||
sources: row.captures.map(capture => ({ ...capture, ...source(capture.path) })) })) })),
|
||||
};
|
||||
const engineering = [];
|
||||
for (const kind of ['node', 'react', 'python']) {
|
||||
if (!fs.existsSync(path.join(root, kind, 'result.json'))) continue;
|
||||
const result = read(kind + '/result.json');
|
||||
if (!result.executed) { engineering.push({ kind, result, source: source(kind + '/result.json') }); continue; }
|
||||
const session = read(kind + '/session.json'), attempts = read(kind + '/attempts.json');
|
||||
const metrics = {};
|
||||
for (const key of ['input', 'output', 'total', 'cacheRead', 'cacheWrite', 'uncached', 'reasoning']) {
|
||||
const values = attempts.map(value => value.usage?.tokenAccounting?.metrics?.[key]);
|
||||
const known = values.filter(value => Number.isSafeInteger(value?.tokens) && ['reported', 'derived'].includes(value.status));
|
||||
metrics[key] = { knownSumTokens: known.length ? known.reduce((sum, value) => sum + value.tokens, 0) : null,
|
||||
knownRequests: known.length, unknownRequests: values.length - known.length,
|
||||
statuses: [...new Set(values.map(value => value?.status ?? 'unknown'))] };
|
||||
}
|
||||
const reads = [], seenReads = new Set();
|
||||
for (const event of session.events.filter(value => value.kind === 'tool' && value.phase === 'completed' && value.tool === 'read_file')) {
|
||||
let output; try { output = JSON.parse(event.output); } catch { output = null; }
|
||||
const row = { eventId: event.id, path: output?.path ?? event.input?.path, startLine: output?.start_line ?? null, endLine: output?.end_line ?? null,
|
||||
fileSha256: output?.sha256 ?? null, outputSha256: sha(event.output ?? ''), parsed: output !== null };
|
||||
const key = JSON.stringify([row.path, row.startLine, row.endLine, row.fileSha256 ?? row.outputSha256]);
|
||||
row.sameByteRangePreviouslyRead = seenReads.has(key); seenReads.add(key); reads.push(row);
|
||||
}
|
||||
const completedTools = session.events.filter(value => value.kind === 'tool' && value.phase === 'completed');
|
||||
const wire = attempts.map(value => {
|
||||
const capture = result.captures.find(item => item.attemptId === value.id && item.kind === 'request');
|
||||
assert.ok(capture); const body = read(capture.path);
|
||||
return { attemptId: value.id, model: body.model, streaming: body.stream === true,
|
||||
outputFields: Object.fromEntries(['max_tokens', 'max_output_tokens', 'max_completion_tokens'].filter(key => Object.hasOwn(body, key)).map(key => [key, body[key]])),
|
||||
nativeReasoning: Object.fromEntries(['thinking', 'reasoning', 'reasoning_effort'].filter(key => Object.hasOwn(body, key)).map(key => [key, body[key]])),
|
||||
toolsCount: body.tools?.length ?? 0, toolsSchemaSha256: sha(JSON.stringify(body.tools ?? [])), source: source(capture.path) };
|
||||
});
|
||||
const reports = result.reportArtifacts.filter(value => value.kind === 'delivery-report').map(value => {
|
||||
const file = result.captures.find(capture => capture.artifactId === value.id).path, report = read(file);
|
||||
return { artifactId: value.id, status: report.status, taskStatus: report.task?.status, lifecycle: report.lifecycle,
|
||||
gitOutcome: report.gitOutcome, terminalGit: report.terminalGit,
|
||||
deliveries: report.deliveries?.map(item => ({ id: item.id, status: item.status, commit: item.commit })), source: source(file) };
|
||||
});
|
||||
engineering.push({ kind, success: result.success, executed: result.executed, projectId: result.projectId, sessionId: result.sessionId,
|
||||
taskId: result.taskId, runId: result.runId, connectionId: result.connectionId, requests: result.requests, usage: result.usage,
|
||||
startedAt: session.execution.run.startedAt, finishedAt: session.execution.run.finishedAt,
|
||||
elapsedMs: Date.parse(session.execution.run.finishedAt) - Date.parse(session.execution.run.startedAt),
|
||||
stopReason: result.productStopReason, failure: result.failure, identitiesPassed: result.identitiesPassed,
|
||||
commit: result.commit, initialHead: result.initialHead, committedPaths: result.committedPaths, committedCurrent: result.committedCurrent,
|
||||
otherUntracked: result.otherUntracked, final: result.final, browser: result.browser,
|
||||
browserSources: ['browser/result.json', 'browser/visual-review.json'].filter(file => fs.existsSync(path.join(root, kind, file)))
|
||||
.map(file => source(kind + '/' + file)).concat((result.browser?.screenshots ?? []).map(file => source(path.relative(root, file).replaceAll('\\', '/')))),
|
||||
ownerBytesAndIndexPreserved: result.ownerBytesAndIndexPreserved, supervisorContinuations: result.supervisorContinuations,
|
||||
localTokenCapsChanged: result.localTokenCapsChanged, internalDeliveries: result.internalDeliveries, reports,
|
||||
accounting: metrics, reads: { calls: reads.length, distinctByteRanges: seenReads.size, repeatedSameByteRanges: reads.filter(value => value.sameByteRangePreviouslyRead).length, observations: reads },
|
||||
tools: Object.fromEntries([...new Set(completedTools.map(value => value.tool))].map(tool => [tool, completedTools.filter(value => value.tool === tool).length])),
|
||||
wire, toolsSchemaVariants: [...new Set(wire.map(value => value.toolsSchemaSha256))],
|
||||
sources: ['claim.json', 'goal.txt', 'source-before.json', 'result.json', 'session.json', 'export.json', 'attempts.json', 'restoration.json', 'git-status.txt', 'git-log.txt']
|
||||
.map(file => source(kind + '/' + file)), captures: result.captures.map(value => ({ ...value, ...source(value.path) })) });
|
||||
}
|
||||
const auditHistory = fs.readdirSync(path.join(root, 'audits')).filter(name => fs.existsSync(path.join(root, 'audits', name, 'result.json')))
|
||||
.map(name => { const file = 'audits/' + name + '/result.json', value = read(file); return { stage: value.stage, success: value.success, error: value.error ?? null, source: source(file) }; });
|
||||
const summary = { schemaVersion: 1, stage, generatedAt: new Date().toISOString(), evidenceRoot: path.relative(process.cwd(), root).replaceAll('\\', '/'),
|
||||
frozen: source('frozen.json'), compatibility: compatibilitySummary, engineering,
|
||||
requests: compatibilitySummary.requests + engineering.reduce((sum, value) => sum + (value.requests ?? 0), 0),
|
||||
reportedTotalTokens: compatibilitySummary.reportedTotalTokens + engineering.reduce((sum, value) => sum + (value.usage?.totalTokens ?? 0), 0),
|
||||
unknownRequests: compatibilitySummary.unknownRequests + engineering.reduce((sum, value) => sum + (value.usage?.unknownRequests ?? 0), 0),
|
||||
audit: { ...audit, source: source(auditPrefix + '/result.json'), after: source(auditPrefix + '/after.json') }, auditHistory,
|
||||
drivers: fs.readdirSync(path.resolve('scripts/m4-live')).filter(name => name.endsWith('.mjs')).sort().map(name => { const file = 'scripts/m4-live/' + name, bytes = fs.readFileSync(file); return { path: file, bytes: bytes.length, sha256: sha(bytes) }; }),
|
||||
keysReadOrResaved: false, historicalSamplesReplayed: false, productSourceChanged: false, releaseOrRestart: false, purgeExecuted: false,
|
||||
unverified: ['Real Claude identity on the configured proxy', 'Long-term provider reliability', 'Real Mnemo production binding and semantic benefit',
|
||||
'Physical iPhone', 'General repeated-read improvement', 'R/W v2 live exclusion boundary', 'Causal cache benefit', 'Purge benefit'] };
|
||||
save(stage + '-summary.json', summary);
|
||||
const name = stage === 'node' ? 'compatibility-node' : stage;
|
||||
const file = path.resolve('docs/m4-' + name + '-2026-11-11.json');
|
||||
fs.writeFileSync(file, JSON.stringify(summary, null, 2) + '\n', { flag: 'wx' });
|
||||
console.log(JSON.stringify({ file, stage, requests: summary.requests, reportedTotalTokens: summary.reportedTotalTokens, unknownRequests: summary.unknownRequests,
|
||||
engineering: engineering.map(value => ({ kind: value.kind, success: value.success, requests: value.requests, elapsedMs: value.elapsedMs, cacheRead: value.accounting?.cacheRead, reads: value.reads?.calls, repeated: value.reads?.repeatedSameByteRanges })), audit: audit.success }));
|
||||
@@ -0,0 +1,155 @@
|
||||
// Fixed local responses verify the new driver and real tool execution only.
|
||||
// They are not evidence of a provider's intelligence or autonomous delivery.
|
||||
import fs from 'node:fs';
|
||||
import path from 'node:path';
|
||||
import http from 'node:http';
|
||||
import assert from 'node:assert/strict';
|
||||
import { randomUUID } from 'node:crypto';
|
||||
import { fixture, launchWorkbench, completion, call, assertToolPairs } from '../../tests/helpers.mjs';
|
||||
import { root, save, client, configureFixture, captureSession, artifactBytes, expectedOutputFields, connectionDefinition } from './common.mjs';
|
||||
import { invokeProbe, inspectorClosed } from './inspector.mjs';
|
||||
|
||||
const output = path.resolve('.qa/m4-driver-' + randomUUID()), disposers = [], t = { after: fn => disposers.push(fn) };
|
||||
fs.mkdirSync(output); const f = fixture(t, { modernDefaults: true }); f.store.saveConfig({ approvalMode: 'full', mode: 'compatible' });
|
||||
const malformedBytes = Buffer.from([100, 97, 116, 97, 58, 32, 0xff, 10, 10]);
|
||||
const malformedArtifact = f.store.artifactStore.write(f.session, null, malformedBytes, { kind: 'model-stream-response', mediaType: 'text/event-stream' });
|
||||
f.store.core.artifact.set(malformedArtifact.id, malformedArtifact); f.store.save(f.session); f.store.close();
|
||||
let host, captureClient, result, engineeringRequests = 0; const requests = [], cases = [], sockets = new Set();
|
||||
const files = {
|
||||
'src/amount.mjs': `export function amountToCents(input) {
|
||||
if (!['string','number'].includes(typeof input) || typeof input === 'number' && !Number.isFinite(input)) throw new TypeError('amount');
|
||||
const match = String(input).trim().match(/^([+-]?)([0-9]+)(?:\\.([0-9]{1,2}))?$/);
|
||||
if (!match) throw new TypeError('amount');
|
||||
const n = (BigInt(match[2]) * 100n + BigInt((match[3] || '').padEnd(2,'0'))) * (match[1] === '-' ? -1n : 1n);
|
||||
if (n > BigInt(Number.MAX_SAFE_INTEGER) || n < BigInt(Number.MIN_SAFE_INTEGER)) throw new TypeError('unsafe');
|
||||
return n === 0n ? 0 : Number(n);
|
||||
}\n`,
|
||||
'src/summary.mjs': `import {amountToCents} from './amount.mjs';
|
||||
export function summarize(rows) {
|
||||
if (!Array.isArray(rows)) throw new TypeError('rows'); const map = new Map();
|
||||
for (const row of rows) {
|
||||
if (!row || typeof row.currency !== 'string') throw new TypeError('row');
|
||||
const currency = row.currency.trim().toUpperCase(); if (!/^[A-Z]{3}$/.test(currency)) throw new TypeError('currency');
|
||||
const item = map.get(currency) || {currency,cents:0,count:0};
|
||||
const cents = item.cents + amountToCents(row.amount); if (!Number.isSafeInteger(cents)) throw new TypeError('unsafe');
|
||||
map.set(currency,{currency,cents,count:item.count+1});
|
||||
}
|
||||
return [...map.values()].sort((a,b)=>a.currency.localeCompare(b.currency));
|
||||
}\n`,
|
||||
'src/report.mjs': `export function formatReport(rows) {return rows.map(({currency,cents,count})=>currency+' '+(cents<0?'-':'')+Math.floor(Math.abs(cents)/100)+'.'+String(Math.abs(cents)%100).padStart(2,'0')+' ('+count+')').join('\\n');}\n`,
|
||||
};
|
||||
const data = value => 'data: ' + (typeof value === 'string' ? value : JSON.stringify(value)) + '\n\n';
|
||||
function streamBody(protocol, model, nonce, tool) {
|
||||
if (protocol === 'chat-completions') {
|
||||
const delta = tool ? { tool_calls: [{ index: 0, ...call('call_' + nonce, 'workbench_m4_echo_nonce', { nonce }) }] } : { content: 'OK ' + nonce };
|
||||
return data({ model, choices: [{ index: 0, delta, finish_reason: null }] })
|
||||
+ data({ model, choices: [{ index: 0, delta: {}, finish_reason: tool ? 'tool_calls' : 'stop' }] })
|
||||
+ data({ model, choices: [], usage: { prompt_tokens: 20, completion_tokens: 5, total_tokens: 25, prompt_tokens_details: { cached_tokens: 4 } } }) + data('[DONE]');
|
||||
}
|
||||
if (protocol === 'responses') {
|
||||
let sequence = 0; const native = (type, value) => data({ type, sequence_number: sequence++, ...value });
|
||||
let bytes = native('response.created', { response: { id: 'resp_' + nonce, model } }), output;
|
||||
if (tool) {
|
||||
const item = { type: 'function_call', id: 'fc_' + nonce, call_id: 'call_' + nonce, name: 'workbench_m4_echo_nonce', arguments: '', status: 'in_progress' };
|
||||
bytes += native('response.output_item.added', { output_index: 0, item });
|
||||
bytes += native('response.function_call_arguments.delta', { item_id: item.id, delta: JSON.stringify({ nonce }) });
|
||||
output = [{ ...item, status: 'completed', arguments: JSON.stringify({ nonce }) }];
|
||||
bytes += native('response.function_call_arguments.done', { item_id: item.id, arguments: output[0].arguments });
|
||||
} else {
|
||||
bytes += native('response.output_text.delta', { delta: 'OK ' + nonce });
|
||||
output = [{ type: 'message', id: 'msg_' + nonce, role: 'assistant', status: 'completed', content: [{ type: 'output_text', text: 'OK ' + nonce }] }];
|
||||
}
|
||||
return bytes + native('response.completed', { response: { id: 'resp_' + nonce, model, status: 'completed', output, usage: { input_tokens: 20, output_tokens: 5, total_tokens: 25, input_tokens_details: { cached_tokens: 4 } } } });
|
||||
}
|
||||
const event = (type, value = {}) => 'event: ' + type + '\n' + data({ type, ...value });
|
||||
const initial = { id: 'msg_' + nonce, type: 'message', role: 'assistant', model, content: [], stop_reason: null, stop_sequence: null, usage: { input_tokens: 16, cache_read_input_tokens: 4, cache_creation_input_tokens: 0, output_tokens: 0 } };
|
||||
const block = tool ? { type: 'tool_use', id: 'call_' + nonce, name: 'workbench_m4_echo_nonce', input: {} } : { type: 'text', text: '' };
|
||||
const delta = tool ? { type: 'input_json_delta', partial_json: JSON.stringify({ nonce }) } : { type: 'text_delta', text: 'OK ' + nonce };
|
||||
return event('message_start', { message: initial }) + event('content_block_start', { index: 0, content_block: block })
|
||||
+ event('content_block_delta', { index: 0, delta }) + event('content_block_stop', { index: 0 })
|
||||
+ event('message_delta', { delta: { stop_reason: tool ? 'tool_use' : 'end_turn', stop_sequence: null }, usage: { output_tokens: 5 } }) + event('message_stop');
|
||||
}
|
||||
const server = http.createServer(async (request, response) => {
|
||||
try {
|
||||
const chunks = []; for await (const chunk of request) chunks.push(chunk);
|
||||
const body = JSON.parse(Buffer.concat(chunks)); requests.push({ path: request.url, body });
|
||||
const protocol = request.url.endsWith('/messages') ? 'anthropic-messages' : request.url.endsWith('/responses') ? 'responses' : 'chat-completions';
|
||||
if (protocol === 'anthropic-messages') assert.equal(body.max_tokens, 128000, 'Messages uses the exact official specification, not the old 2048.');
|
||||
else assert.ok(['max_tokens', 'max_output_tokens', 'max_completion_tokens'].every(key => !Object.hasOwn(body, key)));
|
||||
if (body.model === 'm4-error') { response.writeHead(400, { 'Content-Type': 'application/json', 'x-request-id': 'owned-m4-error' }); response.end('{"error":{"message":"owned-m4-original-error","type":"invalid_request_error"}}'); return; }
|
||||
if (body.model === 'm4-unknown') { response.writeHead(200, { 'Content-Type': 'text/event-stream' }); response.write(data({ model: body.model, choices: [{ index: 0, delta: { content: 'incomplete' }, finish_reason: null }] })); return; }
|
||||
if (body.model === 'm4-node') {
|
||||
let answer; engineeringRequests++;
|
||||
if (engineeringRequests === 1) answer = completion(Object.entries(files).map(([name, content], index) => call('owned-file-' + index, 'propose_file', { path: name, content, reason: 'Fixed local fixture only.' })));
|
||||
else if (engineeringRequests === 2) answer = completion([call('owned-plan', 'plan_checks', { names: ['test', 'build'] })]);
|
||||
else if (engineeringRequests === 3) {
|
||||
const sessions = await host.api('/sessions'), active = sessions.find(value => value.status === 'running'); assert.ok(active);
|
||||
const state = await host.api('/sessions/' + active.id), plan = state.execution.plans.at(-1); assert.ok(plan);
|
||||
answer = completion([call('owned-check', 'run_checks', { plan_id: plan.id })]);
|
||||
} else if (engineeringRequests === 4) {
|
||||
answer = completion([call('owned-terminal-git', 'run_command', { command: "git -c core.hooksPath=NUL commit --only --no-gpg-sign -m 'Fixed M4 local driver fixture' -- src/amount.mjs src/summary.mjs src/report.mjs\nif ($LASTEXITCODE -ne 0) { throw 'fixture commit failed' }", reason: 'Verify actual local Git with original owner staging preserved.' })]);
|
||||
} else answer = completion([], 'Fixed local fixture finished.');
|
||||
response.writeHead(200, { 'Content-Type': 'application/json' }); response.end(JSON.stringify({ ...answer, model: body.model })); return;
|
||||
}
|
||||
const nonce = JSON.stringify(body.messages ?? body.input).match(/(?:nonce |OK )([a-f0-9-]{36})/)?.[1]; assert.ok(nonce);
|
||||
const tool = body.tools?.length > 0;
|
||||
if (body.stream) { response.writeHead(200, { 'Content-Type': 'text/event-stream' }); response.end(streamBody(protocol, body.model, nonce, tool)); }
|
||||
else { response.writeHead(200, { 'Content-Type': 'application/json' }); response.end(JSON.stringify({ ...completion([], 'OK ' + nonce), model: body.model })); }
|
||||
} catch (error) { response.writeHead(500, { 'Content-Type': 'application/json' }); response.end(JSON.stringify({ error: { message: error.message } })); }
|
||||
});
|
||||
server.on('connection', socket => { sockets.add(socket); socket.on('close', () => sockets.delete(socket)); });
|
||||
await new Promise(resolve => server.listen(0, '127.0.0.1', resolve));
|
||||
disposers.push(() => new Promise(resolve => { for (const socket of sockets) socket.destroy(); server.close(resolve); }));
|
||||
try {
|
||||
await inspectorClosed();
|
||||
const observation = JSON.parse(fs.readFileSync('.qa/m4-live-baseline-2026-11-11/observation.json', 'utf8'));
|
||||
host = await launchWorkbench(t, f.data, {}, { appRoot: observation.boot.runtimeRoot }); configureFixture({ root: output, origin: host.url });
|
||||
captureClient = await client(); disposers.push(() => captureClient.close());
|
||||
const downloaded = await artifactBytes(captureClient.api, f.session.id, malformedArtifact); assert.deepEqual(downloaded, malformedBytes);
|
||||
save('byte-preservation/original.bin', downloaded); save('byte-preservation/metadata.json', malformedArtifact);
|
||||
cases.push({ prefix: 'byte-preservation', expected: 'exact-invalid-utf8-bytes', requests: 0, passed: true });
|
||||
const health = await host.api('/host/status'), beforeSelection = (await host.api('/model-center')).selected;
|
||||
const runProbe = async (connection, scope, prefix, expected = 'success', observationMs = 30000) => {
|
||||
const session = await host.api('/sessions', { projectId: f.project.id }), nonce = randomUUID();
|
||||
const input = { pid: health.identity.pid, port: Number(new URL(host.url).port), runtimeRoot: host.bootstrap.runtimeRoot, projectId: f.project.id, sessionId: session.id,
|
||||
connection, nonce, scope, expectedOutput: expectedOutputFields(connection), observationMs };
|
||||
save(prefix + '/input.json', input); const invoked = await invokeProbe(input); save(prefix + '/invocation.json', invoked);
|
||||
const captured = await captureSession(captureClient.api, session.id, prefix), attempt = captured.rows[0];
|
||||
assert.equal(captured.total, 1); assert.equal(invoked.result.actualToolsExecuted, 0); assert.equal(invoked.result.wireStarts, 1);
|
||||
assertToolPairs(captured.exported.session.providerMessages);
|
||||
if (expected === 'success') { assert.equal(invoked.result.contentPassed, true, JSON.stringify(invoked.result)); assert.equal(attempt.status, 'completed'); assert.equal(attempt.modelIdentity.status, 'matched'); assert.equal(attempt.modelIdentity.complete, true); assert.equal(attempt.usageStatus, 'reported'); }
|
||||
else { assert.equal(attempt.status, expected === 'unknown' ? 'interrupted' : 'failed'); assert.equal(attempt.usageStatus, 'unknown'); assert.equal(invoked.result.contentPassed, false); }
|
||||
if (expected === 'error') assert.ok(captured.captures.some(value => value.kind === 'response' && fs.readFileSync(path.join(root, value.path), 'utf8').includes('owned-m4-original-error')));
|
||||
assert.deepEqual((await host.api('/model-center')).selected, beforeSelection);
|
||||
assert.deepEqual(connectionDefinition((await host.api('/model-center')).connections.find(value => value.id === connection.id)), connectionDefinition(connection));
|
||||
cases.push({ prefix, protocol: connection.protocol, scope, expected, requests: 1, passed: true });
|
||||
};
|
||||
const baseUrl = 'http://127.0.0.1:' + server.address().port + '/v1';
|
||||
for (const [protocol, model] of [['chat-completions', 'm4-chat'], ['responses', 'm4-responses'], ['anthropic-messages', 'claude-opus-5-5']]) {
|
||||
const connection = await host.api('/model-center/connections', { name: 'Owned M4 ' + protocol, model, protocol, baseUrl, streaming: true, apiKey: 'owned-m4-fixture-secret' });
|
||||
for (const scope of ['text', 'tools']) await runProbe(connection, scope, 'preflight/' + protocol + '-' + scope);
|
||||
}
|
||||
const json = await host.api('/model-center/connections', { name: 'Owned M4 JSON', model: 'm4-json', protocol: 'chat-completions', baseUrl, streaming: false, apiKey: 'owned-m4-fixture-secret' });
|
||||
await runProbe(json, 'text', 'preflight/json'); await runProbe(json, 'streaming', 'preflight/streaming-override');
|
||||
for (const [model, expected] of [['m4-error', 'error'], ['m4-unknown', 'unknown']]) {
|
||||
const connection = await host.api('/model-center/connections', { name: 'Owned ' + model, model, protocol: 'chat-completions', baseUrl, streaming: true, apiKey: 'owned-m4-fixture-secret' });
|
||||
await runProbe(connection, 'text', 'preflight/' + model, expected, expected === 'unknown' ? 750 : 30000);
|
||||
}
|
||||
const connection = await host.api('/model-center/connections', { name: 'Owned M4 Node', model: 'm4-node', protocol: 'chat-completions', baseUrl, streaming: false, apiKey: 'owned-m4-fixture-secret' });
|
||||
const c = await client();
|
||||
try {
|
||||
const center = await c.api('/model-center');
|
||||
save('frozen.json', { fixture: true, host: await c.api('/host/status'), config: c.boot.config, center, connections: [connection], engineeringPriority: [connection.id] });
|
||||
} finally { await c.close(); }
|
||||
save('compatibility/result.json', { fixture: true, results: [{ connectionId: connection.id, engineeringAllowed: true }] });
|
||||
const argv = process.argv[2]; process.argv[2] = 'node'; await import('./run.mjs'); process.argv[2] = argv;
|
||||
const node = JSON.parse(fs.readFileSync(path.join(root, 'node/result.json'), 'utf8')); assert.equal(node.success, true, JSON.stringify({ failure: node.failure, final: node.final, cleanupErrors: node.cleanupErrors, requests: node.requests }));
|
||||
assert.equal(node.requests, 5); assert.equal(node.final.test.stats.pass, 9); assert.equal(node.ownerBytesAndIndexPreserved, true); assert.equal(node.identitiesPassed, true);
|
||||
assert.equal(requests.length, 15); result = { success: true, cases, fixtureRequests: requests.length, engineeringRequests, realModelRequests: 0,
|
||||
installedAppRoot: observation.boot.runtimeRoot, version: host.bootstrap.version, driverRunsOriginalNodeChecksAndGit: true, fixedImplementationOnlyInFixture: true };
|
||||
} catch (error) { result = { success: false, error: error.stack, cases, fixtureRequests: requests.length, engineeringRequests, realModelRequests: 0 }; process.exitCode = 1; }
|
||||
finally {
|
||||
await inspectorClosed(); for (const dispose of disposers.reverse()) await dispose();
|
||||
fs.writeFileSync(path.join(output, 'result.json'), JSON.stringify(result, null, 2) + '\n', { flag: 'wx' });
|
||||
}
|
||||
console.log(JSON.stringify({ output, ...result }));
|
||||
Some files were not shown because too many files have changed in this diff Show More
Reference in New Issue
Block a user