docs(release): seal 1.17 protocol acceptance and milestone review

This commit is contained in:
Agent Workbench Dev
2026-11-01 00:19:52 +08:00
parent a72dc0fbc1
commit 0d1ab743f1
680 changed files with 1468073 additions and 19 deletions
+2 -2
View File
@@ -1,6 +1,6 @@
# Agent Workbench项目执行规范
1.16在feat/mobile-companion-1.16完成A–F本地正式交付,产品来源2ecdfa2、发行QA0cc8b4c/52f0d20,最终不可变20261031114604110。独立源码519/519、24类186组历史UI、实际Electron窗口7项、原生流程9组、三工程16节点及父Node8/8、React9/9、Python10/10通过;源码与便携各7组手机流程、同载荷隔离安装/重启/卸载7组、正式桌面/390px共20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件、全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。iPhone 14 Pro Max/iOS 27.2为用户提供;用户确认信任、配对、原审批、差异/报告SHA、主屏幕、锁屏、切网恢复、停止及撤销,原宿主审计/进程/文件另行核对,无手机截图。正式手机网关默认关闭、无设备/证书;QA证书与设备身份没有迁入。验收手机监听已关闭并释放4321,4320管理及原数据保留。真实模型工程和Purge推理收益继续按用户选择单独待验,本轮真实模型请求为0。原失败和逐字节证据见docs/QA-1.16.md及docs/mobile-companion-release-identity.json;下一PLAN-1.17仅方案,按既有授权逐切片实施。原4320/4321实机QA候选保持demo无Key、设备已撤销,重新体验须新配对;不把QA状态当正式设备登记。
1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。不可变Agent-Workbench-1.17.0-win-x64-20261031150548103,产品/QA输入来源a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8。原始证据见docs/model-protocols-f-release-evidence/manifest.json及docs/model-protocols-release-identity.json;下一PLAN-1.18仅方案,不把计划当已实现。
## 范围与目标
@@ -60,7 +60,7 @@
## 当前本地服务与Windows
- 正式服务入口http://127.0.0.1:4318,当前1.16.0,沿用用户手动启动的PID36552与原Provider;当前运行目录及数据身份以docs/mobile-companion-release-identity.json为准。1.15期间旧内存Key丢失是历史记录;本轮原连接配置和当前Provider均保持。普通源码/规划/Git工作不重启服务、不读取或复制Key。
- 正式服务入口http://127.0.0.1:4318,当前1.17.0,沿用用户手动启动的PID36552与原Provider;当前运行目录及数据身份以docs/model-protocols-release-identity.json为准。1.15期间旧内存Key丢失是历史记录;本轮原连接配置和当前Provider均保持。普通源码/规划/Git工作不重启服务、不读取或复制Key。
- 浏览器/fixture验证使用独立端口与数据目录;用户项目与研究对象的登录/会话配置不用于测试凭据。
- Windows执行环境以实测shell为准。现有工具使用PowerShell 5.1和UTF-8;不用&&/||拼接命令,使用换行和$LASTEXITCODE核对结果。
- 桌面与移动共用执行宿主。手机127.0.0.1指手机,远程接入通过未来经验证的配对网关。
+7
View File
@@ -1,5 +1,12 @@
# Agent Workbench 更新记录
## 1.17.0 · 模型协议、流式与能力路由
- 增加Responses实际请求/原工具ID/用量、Chat及Responses有界SSE、同线路reasoning续接和完整参数终结校验。
- 模型中心支持协议、能力声明与范围明确的检测;能力筛选先于租约,失败/取消/断流保持原账本且不重放已完成工具。
- 桌面/手机显示同一Run片段和原SHA;修复选池后主窗口仍显示离线、旧模型校验、空闲流进展延迟及原字节下载问题。
- 独立源码558项、历史UI191组、同题Node/React/Python的真实工具链、不可变发行及保留状态升级通过;真实模型收益与新iPhone流式单列待验。
## 1.16.0 · 2026-10-31 · 手机伴随端
- 增加默认关闭的局域网HTTPS网关、Windows用户加密证书、两分钟单次配对和逐设备项目/观察/审批/停止授权;手机不接收管理令牌或模型凭据。
+4 -2
View File
@@ -1,8 +1,10 @@
# 本地 Agent 工作台
1.17.0已完成本地正式升级。现在可配置Chat/Responses、流式连接和按能力筛选的模型池,直接核对原请求、拒绝原因、断流及停止结果。模型声明和实际检测分别显示;视觉/语音输入仍属后续开发。见[工作台](http://127.0.0.1:4318/?release=1.17.0)、[验收](docs/QA-1.17.md)、[发行身份](docs/model-protocols-release-identity.json)和[下一阶段](docs/PLAN-1.18.md)。
Agent Workbench 将 Claude Code 与 Cursor 的机制研究整合为独立的本地 AI 编程工作台,并内置模型池、项目记忆、Skills/MCP、REA 工件研究、受控DeepSeek Harness工程引擎、实际页面验收和有依赖的执行计划。工程索引、精确补丁、多文件整批审查、Node/Python实际检查、浏览器交互、本地Git提交及交付报告可以在同一个Task中完成。
1.16.0手机伴随端已完成本地正式交付。手机可查看同一Task/Run、核对原差异与报告、确认原审批及停止所属执行;断线恢复和设备撤销已通过真实iPhone验收。Windows安装、源码/便携独立验证及保留数据升级完成。iPhone 14 Pro Max/iOS 27.2为用户提供;用户确认信任、配对、原审批、差异/报告SHA、主屏幕、锁屏、切网恢复、停止及撤销,原宿主审计/进程/文件另行核对,无手机截图。见[本地工作台](http://127.0.0.1:4318/?release=1.16.0)、[实际验收](docs/QA-1.16.md)与[发行身份](docs/mobile-companion-release-identity.json)。
1.16.0历史交付手机伴随端。手机可查看同一Task/Run、核对原差异与报告、确认原审批及停止所属执行;断线恢复和设备撤销已通过真实iPhone验收。Windows安装、源码/便携独立验证及保留数据升级完成。iPhone 14 Pro Max/iOS 27.2为用户提供;用户确认信任、配对、原审批、差异/报告SHA、主屏幕、锁屏、切网恢复、停止及撤销,原宿主审计/进程/文件另行核对,无手机截图。见[本地工作台](http://127.0.0.1:4318/?release=1.16.0)、[实际验收](docs/QA-1.16.md)与[发行身份](docs/mobile-companion-release-identity.json)。
1.15.0历史交付Windows桌面应用、本机可选加密连接、后台任务/窗口分离、单实例、所属项目入口、可回退安装与分页巡检。495/495独立源码、23类179组历史UI、真实窗口7/原生UI9/三工程16节点、父8/8、9/9、10/10、同载荷安装/重启/卸载7组(隔离测试登记,正式产品安装器另行stage)、正式18页、1446条工件记录和1357个去重文件核对通过;当前2连接配置/1池、17项目/29全部会话/2567记录/4599事件,正式纠正升级前后每行/全导出/全部工件原字节保持(相对初始16项目,用户启动新增1个示例并将原列表顺移;体验期间另新增1个空白idle会话及对应1事件,全部原28会话/记录/工件保持);旧PID23172已在恢复检查前退出,用户体验时手动启动1.14/PID36552,随后保留这个进程/Provider升级;旧内存Key不可恢复,需在界面重新填写,9229关闭。双击安装目录AgentWorkbench.exe使用原项目和连接配置;见[本地工作台](http://127.0.0.1:4318/?release=1.15.0)、[实际验收](docs/QA-1.15.md)、[桌面说明](docs/DESKTOP.md)与[发行身份](docs/desktop-host-release-identity.json)。
@@ -26,7 +28,7 @@ Agent Workbench 将 Claude Code 与 Cursor 的机制研究整合为独立的本
## 后续产品与能力建设
1.12包含模型中心/池路由、[项目记忆](docs/PROJECT-MEMORY.md)、[任务恢复检查点](docs/TASK-CHECKPOINTS.md)、[Skills](docs/SKILLS.md)和[项目MCP](docs/MCP.md)。记忆只在来源复核、脱敏且请求预算允许时使用;恢复检查点绑定完整要求和原结果,核对后显式续接同一任务。1.13两个受限worker与1.14受控引擎/提示实验均已完成发行;1.15桌面和本机加密完整发行已验收,具体结果以[QA-1.15](docs/QA-1.15.md)为准;下一[1.16手机伴随端](docs/PLAN-1.16.md)先实现一次性配对、设备权限与断线恢复,再推进更多模型协议和多模态。
1.12包含模型中心/池路由、[项目记忆](docs/PROJECT-MEMORY.md)、[任务恢复检查点](docs/TASK-CHECKPOINTS.md)、[Skills](docs/SKILLS.md)和[项目MCP](docs/MCP.md)。记忆只在来源复核、脱敏且请求预算允许时使用;恢复检查点绑定完整要求和原结果,核对后显式续接同一任务。1.13两个受限worker与1.14受控引擎/提示实验均已完成发行;1.15桌面和本机加密完整发行已验收,具体结果以[QA-1.15](docs/QA-1.15.md)为准;1.16手机伴随端及1.17两协议/流式/能力路由均已交付;下一[1.18工程附件、视觉输入与长任务续接](docs/PLAN-1.18.md)按完整工程流程推进。
当前Git项目可以先预览已提交版本,再创建独立目录和分支;来源已有暂存、未提交及未跟踪文件保留在原项目。子工作区有自己的项目、会话和实际执行目录,创建中断后显式核对。模型准备Git交付预览后停止请求,确认这份预览后由内核完成提交和报告。具体范围见[独立工作区](docs/WORKSPACES.md)及[1.6验收](docs/QA-1.6.md)。
+8 -6
View File
@@ -1,10 +1,12 @@
# Agent Workbench 能力目标与验收矩阵
1.17当前完整交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。 见[1.17验收](QA-1.17.md)、[回顾](RETROSPECTIVE-1.17.md)与[发行身份](model-protocols-release-identity.json)。
日期:2026-10-31。1.0.0为基础里程碑,1.1.0实现M1,1.2.0增加内置Git交付,1.3.0增加工程索引和精确补丁,1.4.0增加可恢复多文件变更集,1.5.0/1.5.1增加有界请求与可用工件分页,1.6.0增加独立Git工作区和审查收尾,1.6.1增加实际Task工作依据和文件预算分页。下表保留1.0.0缺口与完整产品目标;每项完成范围以实际验证记录为准。
总体方案见 [PRODUCT-PLAN.md](PRODUCT-PLAN.md),研究问题与来源见 [RESEARCH-ROADMAP.md](RESEARCH-ROADMAP.md)。
1.16当前完整交付:独立HTTPS/PWA、单次配对、设备项目/操作权限、原Task/Run观察/审批/停止、原seq恢复、差异/报告SHA和撤销已交付。独立源码519/519、24类186组历史UI、实际Electron窗口7项、原生流程9组、三工程16节点及父Node8/8、React9/9、Python10/10通过;源码与便携各7组手机流程、同载荷隔离安装/重启/卸载7组、正式桌面/390px共20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件、全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。iPhone 14 Pro Max/iOS 27.2为用户提供;用户确认信任、配对、原审批、差异/报告SHA、主屏幕、锁屏、切网恢复、停止及撤销,原宿主审计/进程/文件另行核对,无手机截图。正式手机网关默认关闭、无设备/证书;QA证书与设备身份没有迁入。验收手机监听已关闭并释放4321,4320管理及原数据保留。真实模型工程和Purge推理收益继续按用户选择单独待验,本轮真实模型请求为0。见[1.16验收](QA-1.16.md)。
1.16历史交付:独立HTTPS/PWA、单次配对、设备项目/操作权限、原Task/Run观察/审批/停止、原seq恢复、差异/报告SHA和撤销已交付。独立源码519/519、24类186组历史UI、实际Electron窗口7项、原生流程9组、三工程16节点及父Node8/8、React9/9、Python10/10通过;源码与便携各7组手机流程、同载荷隔离安装/重启/卸载7组、正式桌面/390px共20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件、全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。iPhone 14 Pro Max/iOS 27.2为用户提供;用户确认信任、配对、原审批、差异/报告SHA、主屏幕、锁屏、切网恢复、停止及撤销,原宿主审计/进程/文件另行核对,无手机截图。正式手机网关默认关闭、无设备/证书;QA证书与设备身份没有迁入。验收手机监听已关闭并释放4321,4320管理及原数据保留。真实模型工程和Purge推理收益继续按用户选择单独待验,本轮真实模型请求为0。见[1.16验收](QA-1.16.md)。
1.15历史交付:Windows宿主/单实例、显式DPAPI保存、真实Electron窗口/所属本机入口、安装/更新/回退/卸载与分页巡检全部交付。495/495独立源码、23类179组历史UI、真实窗口7/原生UI9/三工程16节点、父8/8、9/9、10/10、同载荷安装/重启/卸载7组(隔离测试登记,正式产品安装器另行stage)、正式18页、1446条工件记录和1357个去重文件核对通过,当前2连接配置/1池、17项目/29全部会话/2567记录/4599事件,正式纠正升级前后每行/全导出/全部工件原字节保持(相对初始16项目,用户启动新增1个示例并将原列表顺移;体验期间另新增1个空白idle会话及对应1事件,全部原28会话/记录/工件保持);旧PID23172已在恢复检查前退出,用户体验时手动启动1.14/PID36552,随后保留这个进程/Provider升级;旧内存Key不可恢复,需在界面重新填写。原失败/范围见[QA-1.15](QA-1.15.md),下一真实手机前置见[PLAN-1.16](PLAN-1.16.md)。
@@ -136,17 +138,17 @@
| C11 真实检查与TestRunner | 实际Node/React/Python检查计划、测试数量/退出码/原日志、源码/范围绑定已交付;零测试/跳过不能算通过 | 识别真实项目检查;计划绑定planId,执行返回jobId;保存退出码、测试数和报告 | DS-3 / tools / M2 | 构建失败、零测试、拒绝/取消均不算通过;结果绑定相同工作区与ChangeSet |
| C12 修复与代码审查 | 实际失败反馈、有界修改/修复、无进展暂停、审查/Git门槛已交付;独立多模型复核与稳定自主完成率待验 | 从真实失败定位并有限轮修复;核对需求、diff、回归与风险;保留失败历史 | CX-1、DS-3 / tasks、workspaces / M2–M4 | 能复现的失败先出现再修复;实际重跑通过;审查不能用“模型说成功”代替 |
| C13 可交付成果 | Task所属diff/Git/检查/浏览器原工件与交付报告已交付;完整真实任务基准与专业交付指标待做 | 需求与结果对应、diff/commit、测试报告、运行方法、限制与工件引用 | WB-1 / artifacts / M2–M6 | 用户按交付说明能启动并验证;报告与实际文件、检查及commit一致 |
| C14 模型协议与路由 | 多连接兼容Chat协议池、Run固定路由、轮询/权重/租约/冷却、已验证格式导入已交付;stream:false,多协议/多模态待做 | ModelProvider能力探测、流式文本/工具参数、模型配置快照、按任务角色路由与可解释回退 | CX-2、DS-1 / models / M1、M3 | 支持/不支持的工具/流式能力可见;Run绑定实际模型;故障不串结果或重放工具 |
| C15 预算与用量 | 完整请求预检、Run/Task累计预算与unknown usage已交付;1.13增加父总预算/两个子分配、累计步数/token/工具/时限、迟到和未知请求保守记账;价格与实际费用待做 | 时间、token、费用、步骤、重试与worker总预算;无进展检测;可配置价格与未知usage标记 | CC-2、DS-2 / models、tasks / M1–M4 | 子任务/重试纳入总预算;耗尽时保存检查点;缺失usage不伪造准确费用 |
| C14 模型协议与路由 | 1.17已交付Chat/Responses、两协议有界SSE、按声明能力选池、显式范围检测、Run固定路由及已验证格式导入;保留轮询/权重/租约/冷却与原请求证据。视觉/语音实际输入待做 | ModelProvider能力声明与实测、流式文本/工具参数、模型配置快照、按任务角色路由与可解释回退 | CX-2、DS-1 / models / M1、M3 | 支持/不支持的工具/流式能力可见;Run绑定实际模型;故障不串结果或重放工具 |
| C15 预算与用量 | 完整请求预检、Run/Task累计预算与unknown usage已交付;父/子分配、累计步数/token/工具/时限及迟到/未知请求保守记账保持。1.17逐协议变体核验输出预留,保存实际原usage;价格与实际费用待做 | 时间、token、费用、步骤、重试与worker总预算;无进展检测;可配置价格与未知usage标记 | CC-2、DS-2 / models、tasks / M1–M4 | 子任务/重试纳入总预算;耗尽时保存检查点;缺失usage不伪造准确费用 |
| C16 多代理协作 | 1.13已交付两个受限worker、独立worktree、父子累计预算/停止恢复、当前成果与具体冲突、父工程审查合并;任意规模/OS沙箱待做 | 独立问题分工、上下文/预算隔离、独立worktree、结果与证据汇总、冲突合并 | CC-1、CU-3、DS-2、WB-1 / tasks、workspaces / M4 | 两个独立任务并行完成可合并;重叠编辑被发现;父任务停止可传播 |
| C17 REA能力研究 | 固定REA6.8.0的ASAR/PE三项静态操作、所属job/目标SHA256/原Evidence已交付;其余工具与运行trace待做 | 目标清单、版本/hash、inventory/analyze/trace、分析job、Evidence查看与研究结论登记 | 全部目标 / tools、artifacts / M3 | 从能力结论能追到确切工件与Evidence;静态/运行证据和未知范围分别显示 |
| C18 Purge增强实验 | 1.14来源锁定的两版本提示实验、开关/失效拒绝和同题夹具对照已交付;原完整插件与真实模型收益未验证 | 独立实验profile、兼容版本、策略分类、开关/回退和同题对照;有效机制按接口采纳 | PG-1 / engines、models、tasks / M3 | 同模型/任务/预算重复比较成功率、成本与错误修改;插件失效可恢复基线 |
| C19 外部代理引擎 | 1.14固定DeepSeek0.2.1-rc.2受控prompt/工具模型桥、所属工程/审批/预算/父审查/Git与停止已交付;Codex等适配待做 | EngineAdapter管理握手、会话/进程、事件、停止/恢复和真实能力限制 | CX-2、DS-1 / engines / M3 | 引擎事件有Task/Run映射;协议缺失能力明确;root idle不直接判工程完成 |
| C20 浏览器与界面验证 | 代理固定静态构建、真实Edge交互/DOM/PNG/原报告与Task绑定已交付;任意在线浏览与电脑操作待做 | 可选浏览器工具、页面状态/截图、请求/控制台结果、交互验收与开发预览 | CU-2、WB-1 / tools、artifacts / M3–M6 | React真实交互/失败场景可复现;结果含页面与任务身份,不只保存截图 |
| C21 桌面产品 | 1.15真实Electron窗口/后台宿主拆分、单实例、原生入口、安装/更新/回退/卸载完整发行和正式保留状态部署已交付 | Electron Windows安装/升级、后台执行宿主、工程/终端/diff/成果区和任务通知 | 产品方案 / desktop、gateway / M5 | 关闭窗口后可再查看任务;安装/升级保留数据;宿主进程生命周期明确 |
| C22 手机伴随端 | 1.16默认关闭HTTPS、一次性配对、项目/操作授权、原Task/Run审批/停止及SHA报告、PWA壳和断线恢复已实现;真实iPhone核心通过、完整独立发行待验。iPhone 14 Pro Max/iOS 27.2为用户提供,原生操作由用户确认、原审计/进程/文件另验 | PWA任务/进度/审查入口;配对网关与设备撤销;共用Run与事件游标 | 产品方案 / mobile、gateway / M5 | 实际手机连电脑并审查同一任务;断线补发不重复执行;失效设备不能授权 |
| C23 数据、配置与发布迁移 | 临时Key默认内存;1.15新增显式DPAPI保存/关闭/删除、同事务回退和安装根入口/指针回退已实际验收;正式发行已通过,设备配置待1.16 | 用户主动配置凭据、系统凭据存储、数据库迁移/回退、版本兼容和日志导出 | CX-2、DS-1 / storage、gateway、desktop / M1、M5–M6 | 升级副本与回退可验证;导出无Key;任务和来源不丢失 |
| C24 工程能力评测 | 行为/UI与Node/React/Python实际样本、失败/人工介入证据已保留;封存成套基准、重复评测与整体稳定完成率待做 | 封存真实工程任务、独立验收、重复运行和错误分类;追踪人工介入/成本/恢复 | 全部目标 / evaluation、artifacts / M1–M6 | 发布能力可链接实际任务记录;报告含失败样本与环境,不能只展示成功演示 |
| C22 手机伴随端 | 1.16默认关闭HTTPS、一次性配对、项目/操作授权、原Task/Run审批/停止及SHA报告、PWA壳和断线恢复已实现;真实iPhone核心与完整独立发行通过。iPhone 14 Pro Max/iOS 27.2为用户提供,原生操作由用户确认、原审计/进程/文件另验 | PWA任务/进度/审查入口;配对网关与设备撤销;共用Run与事件游标 | 产品方案 / mobile、gateway / M5 | 实际手机连电脑并审查同一任务;断线补发不重复执行;失效设备不能授权 |
| C23 数据、配置与发布迁移 | 临时Key默认内存;1.15新增显式DPAPI保存/关闭/删除、同事务回退和安装根入口/指针回退已实际验收;正式发行已通过,1.16设备配对/撤销已验收 | 用户主动配置凭据、系统凭据存储、数据库迁移/回退、版本兼容和日志导出 | CX-2、DS-1 / storage、gateway、desktop / M1、M5–M6 | 升级副本与回退可验证;导出无Key;任务和来源不丢失 |
| C24 工程能力评测 | 固定40份工程基准及Node/React/Python实际工具链、父审查/Git和失败记录已封存,1.17新增同题Responses SSE真实执行门槛;真实模型自主完成率、多模型收益与重复推理评测待验 | 封存真实工程任务、独立验收、重复运行和错误分类;追踪人工介入/成本/恢复 | 全部目标 / evaluation、artifacts / M1–M6 | 发布能力可链接实际任务记录;报告含失败样本与环境,固定回答不替代模型推理评测 |
### 1.6.1 M2实际工作依据切片
+5 -1
View File
@@ -1,5 +1,9 @@
# 里程碑回顾与修复审查
## 当前节点:1.17完整本地交付
1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。 见[1.17验收](QA-1.17.md)、[回顾](RETROSPECTIVE-1.17.md)与[发行身份](model-protocols-release-identity.json)。 下一[1.18方案](PLAN-1.18.md)先审查截图附件、实际视觉选路和同Task预算续接;以下1.12.1审查及1.16节点保留为历史记录。
审查开始:2026-10-30;本地交付:2026-10-31(Asia/Shanghai)。审查基线为1.12.0实现3f6797b和交付记录ca9042e;审查起始服务采用不可变发行20261030124454517。用户要求本阶段先审查、修复及记录计划,随后再开发新能力。
本报告区分历史发行证据、当前源码检查、当前隔离复验及未完成项。各版本原报告和失败记录保持原字节,不把过去的局部验证推断成整体自主交付能力。
@@ -12,7 +16,7 @@
- 修复:先以独立端口、临时项目和自有凭据夹具复现,再做相关验证;不读取真实Key、不发送真实模型请求,不普通重启正式服务。
- 收尾:本地小步提交、必要的补丁发行、最终证据与未完成能力说明。
## 当前进度
## 1.12.1审查时的进度
已核对1.12.0交付:357/357独立源码、150组完整界面和部署后8个页面通过;原PID19084/Provider以及15项目、26会话、2505记录、4476事件在升级时保持。此记录是历史发行基线,本阶段新的检查另行记录。
+3 -3
View File
@@ -1,10 +1,10 @@
# 后续执行计划
更新:2026-10-31。当前正式1.16.0手机伴随端A–F已完成。独立源码519/519、24类186组历史UI、实际Electron窗口7项、原生流程9组、三工程16节点及父Node8/8、React9/9、Python10/10通过;源码与便携各7组手机流程、同载荷隔离安装/重启/卸载7组、正式桌面/390px共20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件、全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。正式手机网关默认关闭、无设备/证书;QA证书与设备身份没有迁入。验收手机监听已关闭并释放4321,4320管理及原数据保留。真实模型工程和Purge推理收益继续按用户选择单独待验,本轮真实模型请求为0。原结果见[QA-1.16](QA-1.16.md)与[发行身份](mobile-companion-release-identity.json)。
更新:2026-10-31。1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。 见[1.17验收](QA-1.17.md)、[回顾](RETROSPECTIVE-1.17.md)与[发行身份](model-protocols-release-identity.json)。
产品目标是让用户给出需求后,工作台能保留项目基线,完成定位、计划、修改、检查、必要修复、审查和Git交付,并在停止、故障、断线后明确保留已完成结果。桌面与手机共享同一执行宿主、任务和证据。
当前接续:1.16完整发行与回顾完成,下一具体阶段为[1.17模型协议、流式与能力路由](PLAN-1.17.md),该文件仍为后续方案;先审查现有Provider/池/请求账本,再按实际协议逐切片实现。iPhone 14 Pro Max/iOS 27.2为用户提供;用户确认信任、配对、原审批、差异/报告SHA、主屏幕、锁屏、切网恢复、停止及撤销,原宿主审计/进程/文件另行核对,无手机截图。4320/4321实机QA设备已撤销,正式网关默认关闭,新的正式使用需按[手机连接说明](IPHONE-COMPANION.md)在原数据中配对。
当前接续:1.17 A–F已完成。下一[1.18工程附件、视觉输入与长任务可恢复性](PLAN-1.18.md)只记录后续方案;先复核1.17原失败、实际能力和待验边界,再按A–F完成一张截图到工程交付的路径。iPhone新增流式/附件实机与真实模型工程分别验收,不用模拟视口和固定回答替代。
## 执行顺序与完成门槛
@@ -16,7 +16,7 @@
| 2b:其它引擎与真实收益(后续独立切片) | Codex实际适配;Claude Code/Cursor/WorkBuddy机制研究;Purge真实模型对照 | 实际版本、审批/取消/模型与工具契约;已验证的受限桥和所属工作区 | 指定工程有实际原协议/工具/检查/Git与停止结果;真实有无实验同模型同题同预算,记录失败和人工介入,不用连接或固定回答夹具推断成功率 |
| 3:Windows桌面(1.15完成) | 单实例、后台宿主、项目/终端/审查入口、安装包与保留数据更新 | 宿主与客户端生命周期拆分;本机凭据加密;稳定迁移契约 | 关闭/重开窗口不丢任务;重复启动复用宿主;安装/升级/卸载的数据行为可核验;崩溃后显式恢复;旧版本可回退 |
| 4:手机伴随端(1.16完成) | PWA、一次性配对、设备身份、撤销、同Run观察/审批、断线游标补发 | 设备认证、加密传输、审批身份/幂等、速率限制;阶段3宿主 | 在真实手机完成配对和撤销;断线后按seq恢复;旧Run/过期/重复审批拒绝;手机和桌面显示同一结果;手机不直接执行宿主文件与命令 |
| 5:模型与多模态 | 更多实际协议、流式取消、按能力选路、可恢复池策略、视觉/语音/附件、项目缓存 | 每种协议的工具调用/用量/取消实测;凭据和附件隔离 | 图像、语音、文本各有真实样本;流式断开保留unknown而不重复副作用;不兼容成员不被调度;缓存绑定项目/权限/源码版本并可清除;429/401/超时/冷却有独立证据 |
| 5:模型与多模态(1.17协议/流式完成,1.18附件/视觉待做) | 两协议、流式取消、能力选路已交付;下一为图像附件、同Task续接,语音及项目缓存后续分步实现 | 每种协议的工具调用/用量/取消实测;凭据和附件隔离 | 图像、语音、文本各有真实样本;不兼容成员零请求;故障不重复副作用;缓存绑定项目/权限/源码版本并可清除;原失败可核查 |
| 6:自我迭代与专业交付 | 隔离分支生成候选、固定评测集、多模型复核、质量/成本对照、升级候选和回滚 | 可复现工程基准、完整质量门槛、阶段1隔离和阶段3更新 | 工作台可提出并验证自身改进;候选不直接改运行中的发行;回归或独立验收失败不能发布;保存原差异、指标和回退点;采纳须经用户已有权限规则 |
阶段间可并行完成只读研究和文档,涉及项目写入、权限、发行的实现按依赖推进。每个切片提交真实功能、失败分类、验收证据和本地Git提交;不自动推送远端。
+3 -1
View File
@@ -1,6 +1,6 @@
# 1.17 执行计划:模型协议、流式执行与能力路由
接续已完成独立发行和正式部署的1.16手机伴随端。当前从299fb14在feat/model-protocols-1.17实施;A–D已实现并通过专项,E操作验收中,F完整发行待执行。完成情况以QA-1.17、task_plan和原证据为准。继续保持原Task/Run、工具结果、项目字节、审批和数据,按切片开发、验证、回顾并本地提交。
A–F已完成实现、独立验收、正式状态保留升级及阶段回顾。1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。 见[1.17验收](QA-1.17.md)、[回顾](RETROSPECTIVE-1.17.md)与[发行身份](model-protocols-release-identity.json)。 本文保留原实施范围与契约,下一具体方案见[PLAN-1.18](PLAN-1.18.md)。
用户目标是让接入的多个模型在真实工程中稳定承担合适工作。当前基础是兼容Chat协议的多连接、轮询/权重、Run粘连、并发租约、冷却及请求账本,另有受控DeepSeek引擎桥;这些不证明所有品牌的原生协议、流式、视觉或语音都可用。WorkBuddy未知配置格式、未知登录态导入及真实Purge收益仍须独立核验。
@@ -19,6 +19,8 @@
| E:真实任务界面 | 协议/能力配置、检测范围、当前选路、流状态、中断核对和失败原因;沿用已有品牌与中文任务布局 | 用户可完成添加、检测、禁用、入池、执行、停止与核对全流程;实际桌面/390px/iPhone观察状态与原事件一致 |
| F:工程与发行回顾 | 固定Node/React/Python样本、故障注入、完整历史回归、独立源码/便携/安装与保留状态升级 | 原失败保留;模型协议、夹具执行、真实推理收益分别报告;版本与源码/工件/实际部署身份可追溯 |
E原计划包含iPhone新增流式实机观察,本轮完成的是桌面/390px及真实TLS配对伴随端;1.16已有的用户实机结论保持,1.17新增流式仍单列待验,不将本地A–F交付表述为新增实机通过。
## 已核对的扩展位置
1.16发行期间只读核对了当前源码,未开始1.17产品实现。`src/server/provider.mjs`的`complete`固定调用`/chat/completions`,`request`读取完整有界JSON后返回;已有请求AbortController和停止等待,不能据此宣称支持逐片输出。`src/shared/contracts/model-request.mjs`负责封存请求体与预算,新协议需要封存它实际发送的字节,而不是只保留旧Chat请求的摘要。
+30
View File
@@ -0,0 +1,30 @@
# 1.18 后续方案:工程附件、视觉输入与长任务可恢复性
本文件是后续方案,未标成已实现。先完成1.17的独立源码、便携、安装和正式状态保留验收,封存原失败与阶段回顾,再按下表分步实施。继续使用现有React、Task/Run、原请求工件、模型中心、Git审查及桌面/伴随端,不扩成另一套执行系统。
## 用户能完成的任务
用户可以给现有工程任务附一张界面截图,要求定位问题、提出修改并通过原检查。工作台明确显示附件将发送给哪条模型连接,在该Run中保留原文件身份、实际请求和原结果;需要视觉的请求不会调度到仅支持文字的池成员。桌面和手机看到同一附件与任务,停止、断线及失败不会重放已完成工具。
1.17原生窗口回看中出现了既有任务上下文预算暂停。后续先核对原预算记录和当前选路容量,明确指出是固定输入、响应预留、上下文窗口还是累计预算不足,提供同Task的可审查续接。不能只给一句“缩小范围或新建会话”,也不能静默丢掉用户需求、原工具结果或绕开预算。该观察本身不代表已定位新的产品缺陷,实施前用原脱敏证据及自有样本复现。
| 切片 | 具体能力 | 交付证据与门槛 |
| --- | --- | --- |
| A:审查与输入契约 | 审查1.17已交付能力、开放问题和原生暂停;定义Task所属附件、引用、类型/大小/数量、生命周期与预算 | 固定正常与拒绝样本;原历史无写入迁移;未知附件不能进入请求 |
| B:受控图像附件 | 首先支持PNG/JPEG静态图片,校验实际内容、尺寸、编码和上限,保存原字节/SHA及显示派生图;选择和移除均可见 | 所属项目/Task越界读取被拒绝;伪装类型、超限、文件变化、撤销均保留明确结果;不从工程目录自动搜图 |
| C:实际视觉协议与池 | 分别实现Chat/Responses实际图像输入与封存;按当前请求推导vision需求;显式检测声明与实测结果分开 | 自有HTTP逐字节证明、真实图像样本独立验收;不兼容成员零请求;故障/停止和原工具完整配对保持 |
| D:预算与续接 | 图像大小/模型计费未知量显式展示;改进长任务暂停诊断和同Task续接;附件引用不复制为重复历史正文 | 固定长任务包含文字、附件和大工具结果;完整组/原结果保持;不因为减少估算而接受超限请求;真实用量与本地估算分列 |
| E:桌面与手机完整流程 | 在现有任务输入、附件预览与执行轨迹中完成添加、核对、发送、停止、恢复、下载;维持中文工程工作台布局 | 实际桌面/390px和原生窗口验证;iPhone文件选择、旋转、弱网与恢复需用户实机配合,模拟视口结果单列 |
| F:工程与发行回顾 | 同一截图需求的Node/React/Python适用工程样本、原测试、真实页面、审批、父审查/Git、完整历史回归 | 独立源码/便携/同载荷安装、原PID/Provider/全部数据保留升级;失败保留;固定回答夹具不代替真实模型视觉判断 |
## 依赖和范围
视觉先做一张截图到工程交付的完整路径,PDF、音频转写、实时语音、视频、通用电脑操作分别作为后续切片。只有图片上传或模型返回文字,不能称作多模态工程能力已完成。附件不进入PWA离线缓存,不跟随未知客户端登录态导入,不进入其它项目的记忆或共享缓存。
后续缓存需要绑定项目、权限、源码/附件SHA、实际模型/协议和请求版本,用户能查看和清除;不缓存凭据,也不能把旧工具执行结果缓存命中当成新执行成功。语音需要单独定义录音授权、传输、转写、播放和费用边界。其它代理引擎沿用真实协议/审批/取消/工作区契约研究,不把产品名称直接视为可接入能力。
## 每个里程碑都要回顾
每个切片完成后记录实现与实际结果、原失败、尚未验证的能力和本地commit。继续保留1.17混合池保守容量、未知WorkBuddy配置、真实模型工程/Purge收益、安装校验耗时及长自定义路径预检欠项。安装提速必须通过同样的完整性和数据核对。自我迭代要以隔离分支、固定评测、多模型复核及可回退候选为基础,不能直接让模型改写正在运行的发行。
1.17正式stage从原快照到候选/数据核对约639秒;源码可见激活过程中重复核对当前版和候选完整清单。后续单独测量各阶段,并显示真实阶段进度;审查同一次操作中能否复用已绑定身份的检查结果,同时保留发布前的变化检测。该项是性能研究计划,不能直接缓存“已通过”来略过变化文件、原启动器或数据核对。
+1 -1
View File
@@ -1,6 +1,6 @@
# Agent Workbench 后续产品与执行开发方案
更新:2026-10-31。当前正式基线1.15.0,Windows桌面与可选本机加密A–F完整交付;495/495独立源码、23类179组历史UI、真实窗口7/原生UI9/三工程16节点、父8/8、9/9、10/10、同载荷安装/重启/卸载7组(隔离测试登记,正式产品安装器另行stage)、正式18页、1446条工件记录和1357个去重文件核对通过,当前2连接配置/1池、17项目/29全部会话/2567记录/4599事件,正式纠正升级前后每行/全导出/全部工件原字节保持(相对初始16项目,用户启动新增1个示例并将原列表顺移;体验期间另新增1个空白idle会话及对应1事件,全部原28会话/记录/工件保持);旧PID23172已在恢复检查前退出,用户体验时手动启动1.14/PID36552,随后保留这个进程/Provider升级;旧内存Key不可恢复,需在界面重新填写。下一[1.16手机方案](PLAN-1.16.md)保持具体未实施范围,依[后续计划](NEXT-STEPS.md)继续切片回顾和本地Git。真实模型工程/Purge收益继续按用户选择暂缓;当前能力、原证据与部署分别见[能力矩阵](CAPABILITIES.md)、[QA-1.15](QA-1.15.md)及[验证记录](VALIDATION.md)。
更新:2026-10-31。当前正式基线1.17.0。1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。 见[1.17验收](QA-1.17.md)、[回顾](RETROSPECTIVE-1.17.md)与[发行身份](model-protocols-release-identity.json)。 下一[1.18工程附件、视觉与长任务方案](PLAN-1.18.md)仍仅方案,先审查再按切片实施。
## 1. 产品目标
+12 -1
View File
@@ -1,6 +1,8 @@
# 1.17 模型协议、流式执行与能力路由验收
基线299fb14/正式1.16.0,在feat/model-protocols-1.17开发。开发使用自有项目/loopback服务/数据与凭据;正式Provider与数据保持,真实模型工程与Purge收益单独待验。阶段未完整验收前不部署或宣称1.17已交付。
基线299fb14/原正式1.16.0,在feat/model-protocols-1.17开发并已完整本地交付。1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。 最终不可变Agent-Workbench-1.17.0-win-x64-20261031150548103,源码/工件/实际部署身份见[发行身份](model-protocols-release-identity.json)。
以下A–F记录各切片当时的开发结果与原失败;其中“待验”“未部署”等描述属于该时间点,最终状态以文末“F最终独立发行和正式交付”及发行身份为准。
## A:协议与能力登记
@@ -31,3 +33,12 @@ E界面已构建,正在操作验收;F独立发行与正式部署未执行。
完整历史UI25类191组通过(model-protocols-all-ui-validation.json)。开发中整套回归553项552通过/1失败,原日志保留;旧测试把破损JSON预期为工具内错误,现按完整响应契约改为合法JSON的语义参数错误,并新增两个协议整组不执行/不重放回归。最终验收须来自冻结后的独立源码整轮,不把开发中结果当最终发行通过。
新原生观察已封存protocol-native-before,正式仍1.16;用户期间新增的项目/Task以正式部署前新快照为准,不沿用旧计数、不重放原任务。私有发行执行器只复制到新路径;正式公开模型中心的新默认能力字段将逐项校验,存储行、导出、配置和原工件必须严格保持。封存脚本初次替换顺序拒绝、构建误用未安装的本地npm路径失败均保留;改为限定顺序和固定Node附带npm后构建通过。
## F最终独立发行和正式交付
1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。
387份输入、22份构建和40份原基准保持冻结;Git原blob365份严格一致,17份仅仓库声明的CRLF/LF规范化,5份示例生成文件逐字节对应已提交生成器。完整独立源码整轮558/558,源码/便携各自使用所属Node/服务和原协议验收;同载荷QA安装使用独立AppId,正式产品安装器另行stage,只升级一次原内存处理器。新公开能力默认值与每个旧字段严格比对,未泛化忽略模型中心差异。
原开发553项的1失败、首次Responses工程失败、UI及执行器失败与最终成功均按原字节封存。证据见[原件清单](model-protocols-f-release-evidence/manifest.json)。包内说明是构建时快照;最终验收和commit由独立ACCEPTANCE说明绑定,原不可变包不回写。
+11
View File
@@ -21,3 +21,14 @@
三个真实工程检查通过时,Responses子任务最初仍暂停。原因是旧证据核验把max_tokens当唯一输出预留字段;新请求变体本身和实际执行正确,但核验不兼容。修复后逐变体核对原native.maxTokens、完整请求指标和归属,原审批/失败/成果没有重放,重新独立工程完成。这样的暂停有价值,不能把它当成理由放松交付证明。
破损JSON整组拒绝与合法JSON参数校验是两个边界,旧Runtime测试现在保留后者,新增前者分别覆盖Chat和Responses及池不重放。完整历史UI191组和原工程通过不替代独立源码发行门槛;正式升级前重新捕获全部用户状态,允许公开能力字段的已知默认增量,禁止把其它差异统一忽略。
## 最终回顾与下一步
1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。
本阶段证明协议与工程执行闭环,尚未证明所有品牌、多模态或模型自主成功率。后续优先把截图附件、实际视觉选路和长任务暂停诊断做成完整工程流程;未知客户端格式、语音、缓存和通用电脑操作继续按独立契约和实际样本推进。原不可变发行保持,可回看本轮失败和逐项修复;后续方案见[PLAN-1.18](PLAN-1.18.md)。
正式安装stage实际耗时639秒,包含原数据快照、安装器、完整候选清单与保留状态核对。进度可见性和校验耗时继续作为体验欠项;后续提速须保持相同完整性门槛,不通过减少文件或数据比对来缩短时间。
封存首轮把model-protocols-ui-validation.json当目录,ENOTDIR在归档写入前拒绝。修正为只遍历实际目录,原执行器、日志和失败收据保留,使用独占seal-retry结果;没有重跑产品验收、安装或升级。
+4
View File
@@ -1,5 +1,9 @@
# Agent Workbench 验收记录
## 1.17 · 模型协议、流式执行与能力路由(完整本地交付)
1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件,全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。 见[1.17验收](QA-1.17.md)、[回顾](RETROSPECTIVE-1.17.md)与[发行身份](model-protocols-release-identity.json)。
## 1.16 · 手机伴随端(完整本地交付)
独立源码519/519、24类186组历史UI、实际Electron窗口7项、原生流程9组、三工程16节点及父Node8/8、React9/9、Python10/10通过;源码与便携各7组手机流程、同载荷隔离安装/重启/卸载7组、正式桌面/390px共20页通过。正式PID36552/Provider保持,17项目/30全部会话/2590记录/4653事件、全部表每行/会话导出/工件原字节保持;1455条工件及1366个去重文件核对通过,9229关闭。iPhone 14 Pro Max/iOS 27.2为用户提供;用户确认信任、配对、原审批、差异/报告SHA、主屏幕、锁屏、切网恢复、停止及撤销,原宿主审计/进程/文件另行核对,无手机截图。正式手机网关默认关闭、无设备/证书;QA证书与设备身份没有迁入。验收手机监听已关闭并释放4321,4320管理及原数据保留。真实模型工程和Purge推理收益继续按用户选择单独待验,本轮真实模型请求为0。原结果/失败见[QA-1.16](QA-1.16.md)、[回顾](RETROSPECTIVE-1.16.md)、[发行身份](mobile-companion-release-identity.json)及[1.17方案](PLAN-1.17.md)。
@@ -0,0 +1 @@
* -text -whitespace
@@ -0,0 +1,21 @@
{
"success": true,
"phase": "archives",
"startedAt": "2026-10-31T15:08:55.387Z",
"finishedAt": "2026-10-31T15:09:25.856Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-archives.log",
"logBytes": 815,
"logSha256": "a6773e384c8feb1e02039897939b33af8b9399a40814cd7e73ca17586a274daa",
"pointer": {
"path": ".qa/archives-validation.json",
"sha256": "7443b869ad19229173fcafe37b8048931b4b42290cfd72860f99df8ad8ee9f62"
},
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1 @@
{"success": true, "version": "1.17.0", "testedAt": "2026-10-31T15:09:25.813228+00:00", "archives": [{"kind": "portable", "path": "D:\\Ai-project\\REA\u9006\u5411\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103.zip", "files": 13578, "crcPassed": true, "manifestAndDirectoryMatch": true, "sha256": "1e13982788e516157c3a9a05500f1bd4c497dd9aa9701e519207677eaf01458d"}, {"kind": "source", "path": "D:\\Ai-project\\REA\u9006\u5411\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103.zip", "files": 13871, "crcPassed": true, "manifestAndDirectoryMatch": true, "sha256": "20263325fa0a6c8479eff01b631dc56db84fec3e2ab5eacc154798085209f7ab"}], "extractionRoot": "D:\\AWQA17\\source-check-65eb2b40faa24f9ca9165ce2743fc0ec\\Agent-Workbench-1.17.0-source-20261031150548103"}
@@ -0,0 +1,21 @@
{
"success": true,
"phase": "source",
"startedAt": "2026-10-31T15:11:18.158Z",
"finishedAt": "2026-10-31T15:43:48.402Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-source.log",
"logBytes": 1428,
"logSha256": "eb6d948245a4091ee03d05c2fe2c349d87ad5bc2c408b4c0be202f7bf3260862",
"pointer": {
"path": ".qa/source-package-validation.json",
"sha256": "95887a6cf5ba7008d1940cfb7cdf8769568d32b64d91a5673cda87a9e52d0043"
},
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1,15 @@
{"sourceRoot":"D:\\AWQA17\\s-c9bdd353","directory":"D:\\AWQA17\\s-c9bdd353\\.qa\\package-validation-a2f99936-6bc1-4ca9-afa5-eb400e6101ec","nodeVersion":"v24.22.1","python":{"version":"3.10.11","pytest":"8.4.2","path":"D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\dev-task-plans\\.qa\\python-3.10\\Scripts\\python.exe"}}
Source package: install
Source package: build
Source package: contained REA
Source package: native DeepSeek SDK
Source package: contained browser runtime
Source package: compiled controlled-engine delivery
Source package: contained companion gateway and UI
Source package: contained model protocols and mobile streams
Source package: tests
Source package: actual contained Electron window
Source package: actual contained Electron engine UI
Source package: actual contained Electron engineering projects
Source package: Responses SSE actual Node React Python engineering
{"success":true,"version":"1.17.0","sourceRoot":"D:\\AWQA17\\s-c9bdd353","directory":"D:\\AWQA17\\s-c9bdd353\\.qa\\package-validation-a2f99936-6bc1-4ca9-afa5-eb400e6101ec","tests":{"tests":558,"pass":558,"fail":0,"cancelled":0,"skipped":0,"todo":0},"nodeVersion":"v24.22.1","python":{"version":"3.10.11","pytest":"8.4.2","path":"D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\dev-task-plans\\.qa\\python-3.10\\Scripts\\python.exe"},"inputFiles":387,"buildMatchedFiles":22,"rea":true,"engine":true,"browser":true,"controlled":true}
@@ -0,0 +1,21 @@
{
"success": true,
"phase": "portable",
"startedAt": "2026-10-31T15:09:36.840Z",
"finishedAt": "2026-10-31T15:11:14.047Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-portable.log",
"logBytes": 1166941,
"logSha256": "3aa719fa307d09b4fb3b43220239b988a5f21ccfd75fc116a2ad32f2f4ed6b65",
"pointer": {
"path": ".qa/release-validation.json",
"sha256": "cffe60a361ef1ffa5d265c2d3d1025dbb4b274784527da1ea7039880a2f092f5"
},
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
File diff suppressed because one or more lines are too long
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "portable-companion",
"startedAt": "2026-10-31T15:11:14.116Z",
"finishedAt": "2026-10-31T15:11:29.955Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-portable-companion.log",
"logBytes": 806,
"logSha256": "e72c86239628734e7d996ab622964583ac81e6d2989b6b97b46af82a01a713c7",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1,8 @@
PASS desktop explicitly enables HTTPS and downloads the actual public CA without changing OS trust
PASS 390px real gateway pairing clears fragment, displays scoped task list and registers the same device in desktop
PASS phone UI confirms original Task/Run approval and the existing Runtime executes one command
PASS original paginated diff and counted check/report display verified content from the same Task
PASS disconnect and refresh recover from persistent seq without submitting a new command or Run
PASS PWA service worker stores the shell and no API/project response
PASS desktop revocation removes mobile project content immediately on the next read while preserving the original completed tool
EVIDENCE D:\Ai-project\REA逆向\Agent-Workbench\.qa\companion-ui-3a36306d-ed33-473e-8f9e-19c597c88b87
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "portable-protocols",
"startedAt": "2026-10-31T15:11:30.025Z",
"finishedAt": "2026-10-31T15:11:43.400Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-portable-protocols.log",
"logBytes": 715,
"logSha256": "8debfa35e276bc6d1cd88de4db7ca58ddb089f7183ee2fbd4024321b039602bc",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1,6 @@
PASS actual form saves Responses and declared stream, refresh preserves both and narrow editor remains operable
PASS separate text, nonce tool and actual SSE probes bind only their requested capability and own attempts
PASS actual pooled task excludes text-only before dispatch, executes original tool once and downloads SHA-matched original SSE
PASS desktop and actual paired gateway show same bounded stream; mobile stops original HTTP with zero fragment tool dispatch or replay
PASS truncated output keeps explicit unknown ledger, partial evidence and no automatic failover or incomplete tool execution
EVIDENCE D:\Ai-project\REA逆向\Agent-Workbench\.qa\model-protocols-ui-c1ff3990-c3ea-4cea-b53b-857e9f109812
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "prepare-install",
"startedAt": "2026-10-31T15:08:55.387Z",
"finishedAt": "2026-10-31T15:09:57.768Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-prepare-install.log",
"logBytes": 256,
"logSha256": "baf26c0fc665b274014a449bb166575ff604b8253b6c1e7a52edeafb9a1a7135",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1 @@
{"success":true,"appId":"9989c9b2-0642-46a4-b18f-811c9470ee92","installerSha256":"b1f6ab177436e89cda543dd3796a55c2279129b0881b940caef1e086ea0db2ce","productSetupSha256":"3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16","samePayload":true}
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "install",
"startedAt": "2026-10-31T15:11:34.709Z",
"finishedAt": "2026-10-31T15:17:20.779Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-install.log",
"logBytes": 566,
"logSha256": "bbc65bfa996e55d561e5267af96c7432d022d96a407f6421c06bc745ed2d0e79",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1,8 @@
PASS exact final installer
PASS actual installed native bootstrap and bundled execution host
PASS actual installed owned DPAPI credential and HTTP authentication
PASS actual installed owned approval and bundled npm project check
PASS actual owned Electron crash preserves final installed host and data
PASS actual installed host restart preserves every row and loads owned saved connection
PASS actual exact final installer uninstall preserves all owned user data
{"success":true,"version":"1.17.0","root":"C:\\AWQA17\\i-03c543fd601049e4bf305430392c3afa","cases":7}
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "formal-stage",
"startedAt": "2026-10-31T15:44:48.364Z",
"finishedAt": "2026-10-31T15:55:28.155Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-formal-stage.log",
"logBytes": 97,
"logSha256": "f146a633ff6c3a157c09bdb8b663977ea7444cd208773c1f5f89d2263b95ff76",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1 @@
{"success":true,"phase":"staged","version":"1.17.0","output":".qa/protocol-f-formal-stage.json"}
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "formal-upgrade",
"startedAt": "2026-10-31T15:55:28.223Z",
"finishedAt": "2026-10-31T15:57:34.259Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-formal-upgrade.log",
"logBytes": 101,
"logSha256": "ac9d65cc0bbf7b0c1f4bdad8a842bf32cf0f70277eba80f9e707bb32fd7f55a0",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1 @@
{"success":true,"phase":"upgraded","version":"1.17.0","output":".qa/protocol-f-formal-upgrade.json"}
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "formal-activate",
"startedAt": "2026-10-31T15:57:34.317Z",
"finishedAt": "2026-10-31T16:07:41.077Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-formal-activate.log",
"logBytes": 103,
"logSha256": "dc6cbbdc1db406133a71adda3193857c61b876ab49956e6c0758fb0f14d56b07",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1 @@
{"success":true,"phase":"activated","version":"1.17.0","output":".qa/protocol-f-formal-activate.json"}
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "formal-ui",
"startedAt": "2026-10-31T16:07:41.146Z",
"finishedAt": "2026-10-31T16:07:46.028Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-formal-ui.log",
"logBytes": 268,
"logSha256": "535c4b5aef973a308c0732bfebf6cd629314bd75ba51333ac6e090520c3cad0a",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1,3 @@
{"width":1280,"passedPanels":10}
{"width":390,"passedPanels":10}
{"success":true,"directory":"D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460","panels":20,"pageErrors":0,"requestFailures":0,"blockedWrites":0}
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "formal-verify",
"startedAt": "2026-10-31T16:07:46.092Z",
"finishedAt": "2026-10-31T16:09:57.326Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-formal-verify.log",
"logBytes": 100,
"logSha256": "cde4110d1f6e9f5b5ccfb3de44cb9d1b2b199fdaf5002304cc4c304ad5c9e3d9",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1 @@
{"success":true,"phase":"verified","version":"1.17.0","output":".qa/protocol-f-formal-verify.json"}
@@ -0,0 +1,18 @@
{
"success": true,
"phase": "formal-artifacts",
"startedAt": "2026-10-31T16:09:57.397Z",
"finishedAt": "2026-10-31T16:10:02.867Z",
"exitCode": 0,
"logPath": ".qa/protocol-f-formal-artifacts.log",
"logBytes": 138,
"logSha256": "7456059ce40410ae26d50a85c427c6ee46ca642b6478dddba922f0f15dffb0e3",
"pointer": null,
"unchangedInputs": true,
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"release": {
"portableRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"sourceRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16"
}
}
@@ -0,0 +1 @@
{"success":true,"total":1455,"checked":1455,"pages":15,"reportSha256":"a79d9d4dcb8d27dad0f4c4290fae1bc3fcfd11be133cb92491adaa03224d42e2"}
@@ -0,0 +1,32 @@
import fs from 'node:fs';
import assert from 'node:assert/strict';
import { createHash } from 'node:crypto';
import { execFileSync } from 'node:child_process';
const hash = bytes => createHash('sha256').update(bytes).digest('hex');
const git = args => execFileSync('git', args, { windowsHide: true, maxBuffer: 20000000 });
const baseline = JSON.parse(fs.readFileSync('docs/model-protocols-release-code-baseline.json', 'utf8'));
const tree = new Map(git(['ls-tree', '-r', '-z', baseline.codeCommit]).toString('utf8').split('\0').filter(Boolean).map(line => { const [header, name] = line.split('\t'); return [name, header.split(' ')[2]]; }));
const exact = [], normalized = [], generated = [];
const generator = 'examples/delivery-report/scripts/build.mjs'; assert.deepEqual(fs.readFileSync(generator), git(['show', baseline.codeCommit + ':' + generator]));
for (const item of baseline.inputs) {
const bytes = fs.readFileSync(item.path); assert.equal(hash(bytes), item.sha256); assert.equal(bytes.length, item.bytes);
if (tree.has(item.path)) {
const committed = git(['show', baseline.codeCommit + ':' + item.path]);
if (bytes.equals(committed)) exact.push(item.path);
else {
const normalizedBytes = Buffer.from(bytes.toString('utf8').replaceAll('\r\n', '\n'));
assert.deepEqual(Buffer.from(bytes.toString('utf8')), bytes, 'Only valid UTF-8 text may normalize.'); assert.deepEqual(normalizedBytes, committed, 'Difference beyond CRLF/LF: ' + item.path);
const attr = git(['check-attr', 'text', 'eol', '--', item.path]).toString('utf8'); assert.doesNotMatch(attr, /: text: unset/); assert.match(attr, /: eol: (lf|crlf)/);
normalized.push({ ...item, gitBytes: committed.length, gitSha256: hash(committed), attributes: attr.trim().split('\n'), scope: 'Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob.' });
}
} else {
assert.ok(/^examples\/delivery-report\/dist\/(?:[^/]+\.mjs|package\.json)$/.test(item.path), 'Unexpected uncommitted frozen input: ' + item.path);
const source = item.path.endsWith('/package.json') ? null : item.path.replace('/dist/', '/src/');
const expected = source ? git(['show', baseline.codeCommit + ':' + source]) : Buffer.from('{"type":"module"}\n'); assert.deepEqual(bytes, expected);
generated.push({ ...item, source, generator, generatorSha256: hash(fs.readFileSync(generator)) });
}
}
assert.equal(generated.length, 5);
const result = { success: true, at: new Date().toISOString(), codeCommit: baseline.codeCommit, inputFiles: baseline.inputs.length, exactInputs: exact.length, normalized, generated, scope: 'Every frozen source/QA/build input tied to the committed source or exact generated example derivation; unchanged build bytes independently recorded.' };
fs.writeFileSync('.qa/protocol-f-code-git-audit.json', JSON.stringify(result, null, 2) + '\n', { flag: 'wx' });
console.log(JSON.stringify({ success: true, inputs: baseline.inputs.length, exact: exact.length, normalized: normalized.map(item => item.path), generated: generated.length }));
@@ -0,0 +1,33 @@
import fs from 'node:fs';
import path from 'node:path';
import assert from 'node:assert/strict';
import { createHash } from 'node:crypto';
const hash = bytes => createHash('sha256').update(bytes).digest('hex');
const read = file => JSON.parse(fs.readFileSync(file, 'utf8'));
const install = read(read('.qa/protocol-f-install-location.json').result);
assert.equal(install.success, true);
const references = install.cases.flatMap(item => ['processLog', 'setupLog', 'uninstallLog'].filter(key => item[key]).map(key => ({ key, path: item[key], sha256: item[key + 'Sha256'] })));
const formal = read('.qa/protocol-f-formal-stage.json'); assert.equal(formal.success, true);
references.push({ key: 'setupLog', path: formal.setupLog, sha256: formal.setupLogSha256 });
const directory = '.qa/protocol-f-log-completions'; assert.equal(fs.existsSync(directory), false);
const checked = [], completions = [];
for (const item of references) {
const bytes = fs.readFileSync(item.path), current = hash(bytes);
if (current === item.sha256) { checked.push({ ...item, bytes: bytes.length, exact: true }); continue; }
assert.equal(item.key, 'uninstallLog', 'Only the declared uninstall completion trailer may append.');
const prefix = bytes.subarray(0, bytes.length - 199), tail = bytes.subarray(bytes.length - 199);
assert.equal(hash(prefix), item.sha256, 'The original receipt must match the exact prefix.');
const lines = tail.toString('utf8').split('\r\n'); assert.equal(lines.pop(), '');
assert.deepEqual(lines.map(line => { assert.match(line, /^\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}\.\d{3} /); return line.slice(26); }),
['Uninstallation process succeeded.', 'Removed all? Yes', 'Need to restart Windows? No', 'Log closed.']);
const archived = path.join(directory, String(completions.length + 1) + '-original-receipt-prefix-log.txt');
completions.push({ source: path.resolve(item.path), original: { path: archived, bytes: prefix.length, sha256: item.sha256 },
completed: { path: path.resolve(item.path), bytes: bytes.length, sha256: current }, appendedBytes: 199, appendedText: tail.toString('utf8'), prefix });
checked.push({ ...item, bytes: bytes.length, exact: false, exactOriginalPrefix: true });
}
fs.mkdirSync(directory);
for (const item of completions) { fs.writeFileSync(item.original.path, item.prefix, { flag: 'wx' }); delete item.prefix; }
const result = { success: true, at: new Date().toISOString(), checked, completions,
scope: 'All declared installer/process logs match their original SHA, or the uninstall receipt matches the exact original prefix followed solely by the four fixed 199-byte success/removed-all/no-restart/closed lines. Preserve both original prefix and final log; never rewrite the original receipt.' };
fs.writeFileSync(path.join(directory, 'index.json'), JSON.stringify(result, null, 2) + '\n', { flag: 'wx' });
console.log(JSON.stringify({ success: true, references: checked.length, exact: checked.filter(item => item.exact).length, completions: completions.length }));
@@ -0,0 +1,26 @@
import fs from 'node:fs';
import path from 'node:path';
import assert from 'node:assert/strict';
import { createHash } from 'node:crypto';
const hash = bytes => createHash('sha256').update(bytes).digest('hex');
const lock = JSON.parse(fs.readFileSync('package-lock.json', 'utf8')), release = JSON.parse(fs.readFileSync('releases/latest.json', 'utf8'));
const records = [];
for (const [location] of Object.entries(lock.packages)) {
if (!location.startsWith('node_modules/') || !fs.existsSync(location)) continue;
const meta = JSON.parse(fs.readFileSync(path.join(location, 'package.json'), 'utf8'));
for (const name of fs.readdirSync(location).filter(name => /^(licen[cs]e|copying)([.-].*)?$/i.test(name) && fs.statSync(path.join(location, name)).isFile())) {
const original = path.join(location, name), archived = path.join('licenses', (meta.name + '-' + meta.version + '-' + name).replace(/[^a-z0-9_.-]/gi, '_') + '.txt');
const expected = Buffer.from(fs.readFileSync(original, 'utf8')); assert.deepEqual(fs.readFileSync(archived), expected);
for (const root of [release.portableRoot, release.sourceRoot]) assert.deepEqual(fs.readFileSync(path.join(root, archived)), expected);
records.push({ original, archived, bytes: expected.length, sha256: hash(expected) });
}
}
const notices = fs.readFileSync('THIRD_PARTY_NOTICES.txt');
for (const root of [release.portableRoot, release.sourceRoot]) {
assert.deepEqual(fs.readFileSync(path.join(root, 'THIRD_PARTY_NOTICES.txt')), notices);
for (const name of fs.readdirSync('licenses')) { const bytes = fs.readFileSync(path.join('licenses', name)); assert.deepEqual(fs.readFileSync(path.join(root, 'licenses', name)), bytes); }
}
const result = { success: true, at: new Date().toISOString(), dependencyLicenseCopies: records.length, availableLicenseFiles: fs.readdirSync('licenses').length,
notices: { bytes: notices.length, sha256: hash(notices) }, records, scope: 'Generated dependency license texts match each installed locked package; every available license and notice exactly matches both immutable package directories. Original upstream whitespace is preserved, with a scoped Git whitespace exemption for those verbatim texts.' };
fs.writeFileSync('.qa/protocol-f-license-audit.json', JSON.stringify(result, null, 2) + '\n', { flag: 'wx' });
console.log(JSON.stringify({ success: true, dependencyLicenseCopies: result.dependencyLicenseCopies, availableLicenseFiles: result.availableLicenseFiles, noticesSha256: result.notices.sha256 }));
@@ -0,0 +1,84 @@
import fs from 'node:fs';
import path from 'node:path';
import net from 'node:net';
import assert from 'node:assert/strict';
import { execFile } from 'node:child_process';
import { promisify } from 'node:util';
import { DatabaseSync, backup } from 'node:sqlite';
import { publicSnapshot, compareOriginalState, hash } from './protocol-f-state.mjs';
import { InstallationManager, validateInstalledVersion } from '../src/core/desktop/install-manager.mjs';
import { HostController } from '../src/core/desktop/host-controller.mjs';
import { observeWindowsProcess } from '../src/core/desktop/process-identity.mjs';
import { projectEnvironment } from '../src/core/process-environment.mjs';
import { upgradeRunning } from '../scripts/upgrade-running.mjs';
const read = file => JSON.parse(fs.readFileSync(file, 'utf8'));
const save = (file, value) => fs.writeFileSync(file, JSON.stringify(value, null, 2) + '\n', { flag: 'wx' });
const mode = process.argv[2]; assert.ok(['stage', 'upgrade', 'activate', 'verify'].includes(mode));
const release = read('releases/latest.json'); assert.equal(release.installer.version, '1.17.0'); assert.equal(release.installer.qaFixture, null);
assert.equal(hash(fs.readFileSync(release.installer.path)), release.installer.sha256);
for (const phase of ['archives', 'source', 'portable', 'portable-companion', 'portable-protocols', 'install']) {
const result = read('.qa/protocol-f-' + phase + '-result.json'); assert.equal(result.success, true, phase);
assert.equal(result.unchangedInputs, true); assert.equal(result.release.setupSha256, release.installer.sha256);
}
const source = read('.qa/protocol-f-source-original-validation.json'); assert.equal(source.success, true); assert.ok(source.tests.tests >= 557); assert.equal(source.tests.pass, source.tests.tests); for (const key of ['fail', 'cancelled', 'skipped', 'todo']) assert.equal(source.tests[key], 0);
const location = read('.qa/protocol-f-install-location.json'), installed = read(location.result);
assert.equal(installed.success, true); assert.equal(installed.cases.length, 7); assert.equal(installed.installer.productSetup.sha256, release.installer.sha256);
const installRoot = 'C:/Users/wangy/AppData/Local/Programs/AgentWorkbench';
const originalPointer = read(path.join(installRoot, 'current.json')), dataRoot = fs.realpathSync.native(originalPointer.dataRoot);
const candidate = path.resolve(installRoot, 'versions', release.installer.candidateDirectory), url = 'http://127.0.0.1:4318', port = 4318, expectedPid = 36552;
const output = '.qa/protocol-f-formal-' + mode + '.json', beforeFile = '.qa/protocol-f-formal-before.json'; assert.equal(fs.existsSync(output), false, 'Preserve a phase result instead of repeating the operation.');
const startedAt = new Date().toISOString();
const originalProcess = () => { const value = observeWindowsProcess(expectedPid); assert.ok(value); assert.equal(value.createdAt, '2026-10-31T06:30:34.359Z'); assert.equal(hash(fs.readFileSync(value.executable)), 'd702daa815eabbff763c7d3808e235cf031cf2ca70211b56847bc72f51e240e5'); return value; };
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('An inspector is already open.')); }); socket.once('timeout', () => { socket.destroy(); reject(Error('Inspector state could not be verified.')); }); socket.once('error', error => { socket.destroy(); error.code === 'ECONNREFUSED' ? resolve(true) : reject(error); }); }); }
let result;
try {
originalProcess(); await inspectorClosed();
if (mode === 'stage') {
assert.equal(fs.existsSync(candidate), false); assert.equal(fs.existsSync(beforeFile), false);
const before = await publicSnapshot(url, dataRoot); assert.equal(before.version, '1.16.0'); assert.equal(before.pid, expectedPid); assert.deepEqual(before.active, []);
const controller = new HostController({ appRoot: before.runtimeRoot, dataRoot, port, nodePath: process.execPath }); assert.ok(await controller.connect());
const status = await controller.status(); assert.equal(status.desktopWindows.count, 0); assert.equal(status.activity.active, false);
save(beforeFile, before); save('.qa/protocol-f-formal-original-pointer.json', originalPointer);
const db = new DatabaseSync(path.join(dataRoot, 'workbench.sqlite'), { readOnly: true });
const backupPath = path.resolve('.qa/protocol-f-formal-original.sqlite'); assert.equal(fs.existsSync(backupPath), false);
try { await backup(db, backupPath); } finally { db.close(); }
const copy = new DatabaseSync(backupPath, { readOnly: true });
try { assert.equal(copy.prepare('PRAGMA quick_check').get().quick_check, 'ok'); } finally { copy.close(); }
save('.qa/protocol-f-formal-private-backup-receipt.json', { path: backupPath, bytes: fs.statSync(backupPath).size, sha256: hash(fs.readFileSync(backupPath)), credentialsExtracted: false, committed: false });
compareOriginalState(before, await publicSnapshot(url, dataRoot));
const log = path.resolve('.qa/protocol-f-formal-setup.log'); assert.equal(fs.existsSync(log), false);
const execution = await promisify(execFile)(release.installer.path, ['/VERYSILENT', '/SUPPRESSMSGBOXES', '/NORESTART', '/TASKS=desktopicon', '/DIR=' + installRoot, '/DATA_ROOT=' + dataRoot, '/PORT=' + port, '/STAGE_ONLY=1', '/LOG=' + log],
{ windowsHide: true, env: projectEnvironment(), timeout: 900000, maxBuffer: 1024 * 1024 });
const identity = validateInstalledVersion(candidate), after = await publicSnapshot(url, dataRoot), retained = compareOriginalState(before, after);
assert.equal(after.version, '1.16.0'); assert.deepEqual(read(path.join(installRoot, 'current.json')), originalPointer);
result = { success: true, phase: 'staged', version: '1.17.0', candidateIdentity: identity.identitySha256, manifestSha256: identity.manifestSha256, retained, setupLog: log, setupLogSha256: hash(fs.readFileSync(log)), setupOutputEmpty: execution.stdout === '' };
} else {
const staged = read('.qa/protocol-f-formal-stage.json'), before = read(beforeFile); assert.equal(staged.success, true); assert.equal(staged.setupSha256, release.installer.sha256);
const identity = validateInstalledVersion(candidate); assert.equal(identity.identitySha256, staged.candidateIdentity);
const current = await publicSnapshot(url, dataRoot); compareOriginalState(before, current); assert.deepEqual(current.active, []);
if (mode === 'upgrade') {
assert.equal(current.version, '1.16.0');
const info = read(path.join(dataRoot, 'service-info.json')); assert.equal(info.pid, expectedPid);
const upgraded = await upgradeRunning({ url, pid: expectedPid, currentRoot: info.appRoot, handlerRoot: current.runtimeRoot, newRoot: candidate });
assert.equal(upgraded.connectionPreserved, true);
const after = await publicSnapshot(url, dataRoot), retained = compareOriginalState(before, after); assert.equal(after.version, '1.17.0'); assert.equal(after.runtimeRoot, candidate);
await inspectorClosed(); save('.qa/protocol-f-formal-after-upgrade.json', after);
result = { success: true, phase: 'upgraded', version: '1.17.0', upgraded, retained, currentProviderPreserved: true };
} else {
assert.equal(read('.qa/protocol-f-formal-upgrade.json').success, true); assert.equal(current.version, '1.17.0'); assert.equal(current.runtimeRoot, candidate);
let activated = null;
if (mode === 'activate') { activated = await new InstallationManager({ installRoot, dataRoot, candidateRoot: candidate, port }).activate(); assert.equal(activated.phase, 'activated'); assert.equal(activated.hostRetained, false, 'Activation must not upgrade the already migrated handler again.'); }
const controller = new HostController({ appRoot: candidate, dataRoot, port, nodePath: process.execPath }); assert.ok(await controller.connect()); const status = await controller.status();
assert.equal(status.identity.pid, expectedPid); assert.equal(status.activity.active, false);
const gateway = await controller.api('/companion'); assert.equal(gateway.enabled, false); assert.deepEqual(gateway.devices, []); assert.equal(gateway.certificate, null);
const pointer = read(path.join(installRoot, 'current.json')); assert.equal(pointer.active.directory, release.installer.candidateDirectory); assert.equal(pointer.active.runtimeIdentity, identity.identitySha256);
assert.equal(fs.realpathSync.native(pointer.dataRoot), dataRoot); assert.equal(hash(fs.readFileSync(path.join(installRoot, 'AgentWorkbench.exe'))), hash(fs.readFileSync(path.join(release.portableRoot, 'AgentWorkbench.exe'))));
const after = await publicSnapshot(url, dataRoot), retained = compareOriginalState(before, after); await inspectorClosed(); save('.qa/protocol-f-formal-after-' + mode + '.json', after);
result = { success: true, phase: mode === 'activate' ? 'activated' : 'verified', version: '1.17.0', activated, identity: status.identity, pointer, retained, companion: { enabled: gateway.enabled, devices: gateway.devices.length, certificate: gateway.certificate }, currentProviderPreserved: true,
counts: { projects: after.projects, sessions: after.database.sessionIds.length, records: after.database.tables.records.count, events: after.database.tables.events.count, artifacts: after.persistent.artifacts.length }, inspectorClosed: true };
}
}
} catch (error) { result = { success: false, phase: mode, error: error.stack }; process.exitCode = 1; }
save(output, { ...result, startedAt, finishedAt: new Date().toISOString(), installRoot, dataRoot, candidate, setupSha256: release.installer.sha256, originalProcess: originalProcess(), realModelRequests: 0,
scope: 'Accepted product setup stages the existing install, one in-memory handler upgrade preserves the original PID/Provider, pointer activation and final read-only validation. Every original table/row, all session exports and artifact bytes compared; no model Key extraction or QA certificate/device migration.' });
console.log(JSON.stringify({ success: result.success, phase: result.phase, version: result.version, output, ...(result.error ? { error: result.error } : {}) }));
@@ -0,0 +1,43 @@
import fs from 'node:fs';
import assert from 'node:assert/strict';
const read=file=>fs.readFileSync(file,'utf8');
const identity=JSON.parse(read('docs/model-protocols-release-identity.json'));assert.equal(identity.success,true);
const c=identity.formal.counts,summary=`1.17.0完整本地交付:Chat/Responses实际协议、有界SSE、完整工具组核验、能力选池、显式检测与原请求证据已实现。独立源码558/558、25类191组历史UI、随包原生窗口7项/原生流程9组/桌面三工程16节点、Responses三工程13节点及两套父Node8/8、React9/9、Python10/10通过;源码与便携各7组伴随端/5组新协议界面、同载荷隔离安装/重启/卸载7组、正式20页通过。正式PID36552/Provider保持,${c.projects}项目/${c.sessions}全部会话/${c.records}记录/${c.events}事件,全部表每行/会话导出/工件原字节保持;${identity.formal.artifactRecords}条工件及${identity.formal.blobFiles}个去重文件核对通过,9229关闭。公开模型中心仅增加逐项核对的默认能力字段,原存储不迁移。正式手机网关继续关闭、无设备/证书,未迁入QA身份。真实模型工程/Purge收益继续按用户选择单列待验,本轮真实模型请求0;1.17没有新增iPhone实机流式结论。`;
const links='见[1.17验收](QA-1.17.md)、[回顾](RETROSPECTIVE-1.17.md)与[发行身份](model-protocols-release-identity.json)。';
const put=(file,code)=>fs.writeFileSync(file,code);
const insertAfterTitle=(file,text)=>{let code=read(file);const i=code.indexOf('\n');put(file,code.slice(0,i+1)+'\n'+text+'\n'+code.slice(i+1));};
const replaceParagraph=(file,prefix,text)=>{const code=read(file),paragraphs=code.split(/\r?\n\r?\n/),i=paragraphs.findIndex(p=>p.startsWith(prefix));assert.ok(i>=0,file+': '+prefix);paragraphs[i]=text;put(file,paragraphs.join('\n\n'));};
const replaceRow=(file,prefix,text)=>{const rows=read(file).split(/\r?\n/),i=rows.findIndex(row=>row.startsWith(prefix));assert.ok(i>=0,file+': '+prefix);rows[i]=text;put(file,rows.join('\n'));};
const replaceText=(file,before,after)=>{const code=read(file);assert.ok(code.includes(before),file+': missing replacement');put(file,code.replace(before,after));};
insertAfterTitle('README.md','1.17.0已完成本地正式升级。现在可配置Chat/Responses、流式连接和按能力筛选的模型池,直接核对原请求、拒绝原因、断流及停止结果。模型声明和实际检测分别显示;视觉/语音输入仍属后续开发。见[工作台](http://127.0.0.1:4318/?release=1.17.0)、[验收](docs/QA-1.17.md)、[发行身份](docs/model-protocols-release-identity.json)和[下一阶段](docs/PLAN-1.18.md)。');
replaceText('README.md','1.16.0手机伴随端已完成本地正式交付。','1.16.0历史交付手机伴随端。');
replaceText('README.md','下一[1.16手机伴随端](docs/PLAN-1.16.md)先实现一次性配对、设备权限与断线恢复,再推进更多模型协议和多模态。','1.16手机伴随端及1.17两协议/流式/能力路由均已交付;下一[1.18工程附件、视觉输入与长任务续接](docs/PLAN-1.18.md)按完整工程流程推进。');
insertAfterTitle('CHANGELOG.md','## 1.17.0 · 模型协议、流式与能力路由\n\n- 增加Responses实际请求/原工具ID/用量、Chat及Responses有界SSE、同线路reasoning续接和完整参数终结校验。\n- 模型中心支持协议、能力声明与范围明确的检测;能力筛选先于租约,失败/取消/断流保持原账本且不重放已完成工具。\n- 桌面/手机显示同一Run片段和原SHA;修复选池后主窗口仍显示离线、旧模型校验、空闲流进展延迟及原字节下载问题。\n- 独立源码558项、历史UI191组、同题Node/React/Python的真实工具链、不可变发行及保留状态升级通过;真实模型收益与新iPhone流式单列待验。');
insertAfterTitle('docs/VALIDATION.md','## 1.17 · 模型协议、流式执行与能力路由(完整本地交付)\n\n'+summary+' '+links);
insertAfterTitle('docs/CAPABILITIES.md','1.17当前完整交付:'+summary.replace(/^1\.17\.0完整本地交付:/,'')+' '+links);
put('docs/CAPABILITIES.md',read('docs/CAPABILITIES.md').replace('1.16当前完整交付:','1.16历史交付:').replace('真实iPhone核心通过、完整独立发行待验','真实iPhone核心与完整独立发行通过').replace('设备配置待1.16','1.16设备配对/撤销已验收'));
replaceRow('docs/CAPABILITIES.md','| C14 ', '| C14 模型协议与路由 | 1.17已交付Chat/Responses、两协议有界SSE、按声明能力选池、显式范围检测、Run固定路由及已验证格式导入;保留轮询/权重/租约/冷却与原请求证据。视觉/语音实际输入待做 | ModelProvider能力声明与实测、流式文本/工具参数、模型配置快照、按任务角色路由与可解释回退 | CX-2、DS-1 / models / M1、M3 | 支持/不支持的工具/流式能力可见;Run绑定实际模型;故障不串结果或重放工具 |');
replaceRow('docs/CAPABILITIES.md','| C15 ', '| C15 预算与用量 | 完整请求预检、Run/Task累计预算与unknown usage已交付;父/子分配、累计步数/token/工具/时限及迟到/未知请求保守记账保持。1.17逐协议变体核验输出预留,保存实际原usage;价格与实际费用待做 | 时间、token、费用、步骤、重试与worker总预算;无进展检测;可配置价格与未知usage标记 | CC-2、DS-2 / models、tasks / M1–M4 | 子任务/重试纳入总预算;耗尽时保存检查点;缺失usage不伪造准确费用 |');
replaceRow('docs/CAPABILITIES.md','| C24 ', '| C24 工程能力评测 | 固定40份工程基准及Node/React/Python实际工具链、父审查/Git和失败记录已封存,1.17新增同题Responses SSE真实执行门槛;真实模型自主完成率、多模型收益与重复推理评测待验 | 封存真实工程任务、独立验收、重复运行和错误分类;追踪人工介入/成本/恢复 | 全部目标 / evaluation、artifacts / M1–M6 | 发布能力可链接实际任务记录;报告含失败样本与环境,固定回答不替代模型推理评测 |');
insertAfterTitle('docs/MILESTONE-AUDIT.md','## 当前节点:1.17完整本地交付\n\n'+summary+' '+links+' 下一[1.18方案](PLAN-1.18.md)先审查截图附件、实际视觉选路和同Task预算续接;以下1.12.1审查及1.16节点保留为历史记录。');
replaceText('docs/MILESTONE-AUDIT.md','## 当前进度','## 1.12.1审查时的进度');
replaceParagraph('docs/PRODUCT-PLAN.md','更新:', '更新:2026-10-31。当前正式基线1.17.0。'+summary+' '+links+' 下一[1.18工程附件、视觉与长任务方案](PLAN-1.18.md)仍仅方案,先审查再按切片实施。');
replaceParagraph('docs/NEXT-STEPS.md','更新:','更新:2026-10-31。'+summary+' '+links);
replaceParagraph('docs/NEXT-STEPS.md','当前接续:','当前接续:1.17 A–F已完成。下一[1.18工程附件、视觉输入与长任务可恢复性](PLAN-1.18.md)只记录后续方案;先复核1.17原失败、实际能力和待验边界,再按A–F完成一张截图到工程交付的路径。iPhone新增流式/附件实机与真实模型工程分别验收,不用模拟视口和固定回答替代。');
replaceRow('docs/NEXT-STEPS.md','| 5:','| 5:模型与多模态(1.17协议/流式完成,1.18附件/视觉待做) | 两协议、流式取消、能力选路已交付;下一为图像附件、同Task续接,语音及项目缓存后续分步实现 | 每种协议的工具调用/用量/取消实测;凭据和附件隔离 | 图像、语音、文本各有真实样本;不兼容成员零请求;故障不重复副作用;缓存绑定项目/权限/源码版本并可清除;原失败可核查 |');
replaceParagraph('docs/PLAN-1.17.md','接续已完成','A–F已完成实现、独立验收、正式状态保留升级及阶段回顾。'+summary+' '+links+' 本文保留原实施范围与契约,下一具体方案见[PLAN-1.18](PLAN-1.18.md)。');
replaceText('docs/PLAN-1.17.md','## 已核对的扩展位置','E原计划包含iPhone新增流式实机观察,本轮完成的是桌面/390px及真实TLS配对伴随端;1.16已有的用户实机结论保持,1.17新增流式仍单列待验,不将本地A–F交付表述为新增实机通过。\n\n## 已核对的扩展位置');
replaceParagraph('docs/QA-1.17.md','基线299fb14/', '基线299fb14/原正式1.16.0,在feat/model-protocols-1.17开发并已完整本地交付。'+summary+' 最终不可变'+identity.immutableRelease+',源码/工件/实际部署身份见[发行身份](model-protocols-release-identity.json)。');
replaceText('docs/QA-1.17.md','## A:协议与能力登记','以下A–F记录各切片当时的开发结果与原失败;其中“待验”“未部署”等描述属于该时间点,最终状态以文末“F最终独立发行和正式交付”及发行身份为准。\n\n## A:协议与能力登记');
fs.appendFileSync('docs/QA-1.17.md','\n\n## F最终独立发行和正式交付\n\n'+summary+'\n\n387份输入、22份构建和40份原基准保持冻结;Git原blob365份严格一致,17份仅仓库声明的CRLF/LF规范化,5份示例生成文件逐字节对应已提交生成器。完整独立源码整轮558/558,源码/便携各自使用所属Node/服务和原协议验收;同载荷QA安装使用独立AppId,正式产品安装器另行stage,只升级一次原内存处理器。新公开能力默认值与每个旧字段严格比对,未泛化忽略模型中心差异。\n\n原开发553项的1失败、首次Responses工程失败、UI及执行器失败与最终成功均按原字节封存。证据见[原件清单](model-protocols-f-release-evidence/manifest.json)。包内说明是构建时快照;最终验收和commit由独立ACCEPTANCE说明绑定,原不可变包不回写。\n');
fs.appendFileSync('docs/RETROSPECTIVE-1.17.md','\n\n## 最终回顾与下一步\n\n'+summary+'\n\n本阶段证明协议与工程执行闭环,尚未证明所有品牌、多模态或模型自主成功率。后续优先把截图附件、实际视觉选路和长任务暂停诊断做成完整工程流程;未知客户端格式、语音、缓存和通用电脑操作继续按独立契约和实际样本推进。原不可变发行保持,可回看本轮失败和逐项修复;后续方案见[PLAN-1.18](PLAN-1.18.md)。\n');
const stage=JSON.parse(read('.qa/protocol-f-formal-stage.json')),stageSeconds=Math.round((Date.parse(stage.finishedAt)-Date.parse(stage.startedAt))/1000);
fs.appendFileSync('docs/RETROSPECTIVE-1.17.md','\n正式安装stage实际耗时'+stageSeconds+'秒,包含原数据快照、安装器、完整候选清单与保留状态核对。进度可见性和校验耗时继续作为体验欠项;后续提速须保持相同完整性门槛,不通过减少文件或数据比对来缩短时间。\n');
fs.appendFileSync('docs/RETROSPECTIVE-1.17.md','\n封存首轮把model-protocols-ui-validation.json当目录,ENOTDIR在归档写入前拒绝。修正为只遍历实际目录,原执行器、日志和失败收据保留,使用独占seal-retry结果;没有重跑产品验收、安装或升级。\n');
replaceParagraph('AGENTS.md','1.16在',summary+'不可变'+identity.immutableRelease+',产品/QA输入来源'+identity.codeCommit+'。原始证据见docs/model-protocols-f-release-evidence/manifest.json及docs/model-protocols-release-identity.json;下一PLAN-1.18仅方案,不把计划当已实现。');
put('AGENTS.md',read('AGENTS.md').replace('当前1.16.0,沿用用户手动启动的PID36552与原Provider','当前1.17.0,沿用用户手动启动的PID36552与原Provider').replace('当前运行目录及数据身份以docs/mobile-companion-release-identity.json为准','当前运行目录及数据身份以docs/model-protocols-release-identity.json为准'));
put('task_plan.md',read('task_plan.md').replace('6. [in_progress] F:三工程/故障验证、完整历史回归、独立源码/便携/安装、保留状态部署、证据/文档封存及本地Git;逐切片回顾。','6. [complete] F:'+summary+'387输入/22构建/40基准及原失败/成功原件已封存,本地Git收尾;下一PLAN-1.18仅方案。'));
replaceText('task_plan.md','专项;新增字节/超时边界复核中。','专项;新增字节/超时边界和最终独立558项均已通过。');
fs.appendFileSync('findings.md','\n\n## 1.17最终结论\n\n'+summary+'\n\n新协议不能只改Provider;实际请求预算、引擎完成核验、导入、能力筛选、原字节下载及主窗口选路必须共用同一契约。模型声明不替代实际检测,增量片段不替代终结结果。下一步按PLAN-1.18先用原证据审查任务上下文暂停,再接工程附件/视觉。\n');
fs.appendFileSync('progress.md','\n\n## 1.17 A–F完整本地交付\n\n'+summary+'\n\n最终不可变'+identity.immutableRelease+',来源'+identity.codeCommit+';封存'+identity.evidence.files+'份原件/'+identity.evidence.bytes+'字节。所有冻结输入/构建/原基准保持,文档/回顾/能力矩阵与下一1.18方案已同步。正式安装只stage/upgrade/activate各一次,后续为只读核对与原生回看。准备最终本地Git提交及独立ACCEPTANCE说明,不推送远端。\n');
console.log(JSON.stringify({success:true,version:identity.version,summary}));
@@ -0,0 +1,19 @@
import fs from 'node:fs';
import path from 'node:path';
import assert from 'node:assert/strict';
import { createHash } from 'node:crypto';
import { execFileSync } from 'node:child_process';
const folders = ['src', 'tests', 'scripts', 'benchmarks', 'public', 'examples', 'research'];
const top = ['package.json', 'package-lock.json', 'vite.config.ts', 'tsconfig.json', 'index.html', 'mobile.html', '.gitattributes', '.gitignore', 'Start-Desktop.cmd', 'Start-Workbench.cmd', 'Stop-Workbench.cmd'];
assert.equal(execFileSync('git', ['status', '--porcelain', '--', ...folders, ...top], { encoding: 'utf8', windowsHide: true }).trim(), '', 'All executable and QA inputs must be committed.');
const files = (root, relative = '') => fs.readdirSync(path.join(root, relative), { withFileTypes: true }).sort((a, b) => a.name.localeCompare(b.name)).flatMap(entry => {
const file = path.join(relative, entry.name); return entry.isDirectory() ? files(root, file) : [path.join(root, file)];
});
const rows = names => names.sort().map(file => { const bytes = fs.readFileSync(file); return { path: file.replaceAll('\\', '/'), bytes: bytes.length, sha256: createHash('sha256').update(bytes).digest('hex') }; });
const version = JSON.parse(fs.readFileSync('package.json')).version; assert.equal(version, '1.17.0');
const codeCommit = execFileSync('git', ['rev-parse', 'HEAD'], { encoding: 'utf8', windowsHide: true }).trim();
const inputs = rows(folders.flatMap(folder => files(folder)).concat(top)), builds = rows(['dist-client', 'dist-server', 'dist-desktop'].flatMap(folder => files(folder))), benchmarks = rows(files('benchmarks'));
const result = { success: true, version, codeCommit, capturedAt: new Date().toISOString(), inputs, builds, benchmarks,
scope: 'All executable source/test/QA/build inputs including mobile.html, all client/mobile/server/desktop build bytes and original benchmark inputs. Current workspace regression and current Responses engineering and native observation evidence are separate from final independently extracted source and contained release acceptance.' };
for (const file of ['docs/model-protocols-release-code-baseline.json', '.qa/model-protocols-release-code-baseline.json']) fs.writeFileSync(file, JSON.stringify(result, null, 2) + '\n', { flag: 'wx' });
console.log(JSON.stringify({ success: true, version, codeCommit, inputs: inputs.length, builds: builds.length, benchmarks: benchmarks.length }));
@@ -0,0 +1,18 @@
import fs from 'node:fs';
import path from 'node:path';
import assert from 'node:assert/strict';
import { createHash } from 'node:crypto';
import { spawn } from 'node:child_process';
import { projectEnvironment } from '../src/core/process-environment.mjs';
const hash = bytes => createHash('sha256').update(bytes).digest('hex');
const baseline = JSON.parse(fs.readFileSync('docs/model-protocols-release-code-baseline.json'));
function unchanged() { for (const item of [...baseline.inputs, ...baseline.builds, ...baseline.benchmarks]) assert.equal(hash(fs.readFileSync(item.path)), item.sha256, item.path); }
assert.equal(baseline.version, '1.17.0'); unchanged();
const logPath='.qa/protocol-f-package.log'; assert.equal(fs.existsSync(logPath), false);
fs.writeFileSync('.qa/protocol-f-prior-latest.json', fs.readFileSync('releases/latest.json'), {flag:'wx'});
const startedAt=new Date().toISOString(), log=fs.createWriteStream(logPath,{flags:'wx'});
const child=spawn(process.execPath,['scripts/package.mjs','--immutable'], {windowsHide:true, env:projectEnvironment({PATH:path.dirname(process.execPath)+path.delimiter+process.env.PATH}),stdio:['ignore','pipe','pipe']});
child.stdout.on('data',b=>log.write(b)); child.stderr.on('data',b=>log.write(b));
const exitCode=await new Promise((resolve,reject)=>{child.once('error',reject);child.once('close',resolve)}); await new Promise(resolve=>log.end(resolve)); unchanged();
const result={success:exitCode===0,exitCode,startedAt,finishedAt:new Date().toISOString(),codeCommit:baseline.codeCommit,logPath,logSha256:hash(fs.readFileSync(logPath)),unchangedInputs:true};
fs.writeFileSync('.qa/protocol-f-package-result.json',JSON.stringify(result,null,2)+'\n',{flag:'wx'}); console.log(JSON.stringify(result)); if(exitCode!==0)process.exitCode=1;
@@ -0,0 +1,24 @@
import fs from 'node:fs';
import path from 'node:path';
import assert from 'node:assert/strict';
import { randomUUID, createHash } from 'node:crypto';
import { buildWindowsInstaller } from '../scripts/build-installer.mjs';
const release = JSON.parse(fs.readFileSync('releases/latest.json', 'utf8')); assert.equal(release.installer.version, '1.17.0');
const appId = randomUUID(), destination = '.qa/protocol-f-qa-installer.json'; assert.equal(fs.existsSync(destination), false);
const installer = buildWindowsInstaller({ root: process.cwd(), portableRoot: release.portableRoot, outputRoot: path.resolve('.qa'),
name: 'protocol-installer-qa-' + appId, qaFixture: { appId, fault: 'none' } });
const source = fs.readFileSync(installer.source, 'utf8'), product = fs.readFileSync(release.installer.source, 'utf8');
const body = value => value.split('\n').filter(line => !line.startsWith('#define OutputRoot ') && !line.startsWith('#define OutputName ')).join('\n');
assert.equal(body(source).replaceAll(appId, 'E914E883-46F9-4B82-B27F-BDB2B40DCC90')
.replace('DefaultGroupName=Agent Workbench QA E914E883-46F9-4B82-B27F-BDB2B40DCC90', 'DefaultGroupName=Agent Workbench'), body(product));
fs.writeFileSync(destination, JSON.stringify({ ...installer, productSetup: release.installer,
samePayloadManifestSha256: createHash('sha256').update(fs.readFileSync(path.join(release.portableRoot, 'MANIFEST.json'))).digest('hex'),
sourceComparison: 'Only isolated QA AppId/group and output destination/name differ; entire product payload and installation behavior are exact. Product setup will separately stage the existing formal installation.' }, null, 2) + '\n', { flag: 'wx' });
let code = fs.readFileSync('.qa/verify-desktop-f-canonical-storage-guard-install.mjs', 'utf8');
for (const [before, after] of [['desktop-host-f-canonical-storage-guard-qa-installer.json', 'protocol-f-qa-installer.json'],
['desktop-host-f-canonical-storage-install-location.json', 'protocol-f-install-location.json'],
['ff4e7d22-f233-4021-829a-6f5c71933c46', appId], ['1.15.0', '1.17.0'], ['AWQA15', 'AWQA17'], ['timeout = 300000', 'timeout = 900000']]) {
assert.ok(code.includes(before), before); code = code.replaceAll(before, after);
}
fs.writeFileSync('.qa/verify-protocol-install.mjs', code, { flag: 'wx' });
console.log(JSON.stringify({ success: true, appId, installerSha256: installer.sha256, productSetupSha256: release.installer.sha256, samePayload: true }));
@@ -0,0 +1,41 @@
import fs from 'node:fs';
import assert from 'node:assert/strict';
const read = file => fs.readFileSync(file, 'utf8');
const save = (file, code) => { if (fs.existsSync(file)) assert.equal(read(file), code, 'Retain the already prepared wrapper'); else fs.writeFileSync(file, code, { flag: 'wx' }); };
const replace = (code, before, after) => { assert.ok(code.includes(before), before); return code.replaceAll(before, after); };
let code = read('.qa/freeze-companion-release-corrected.mjs');
code = replace(code, '1.16.0', '1.17.0');
code = replace(code, 'mobile-companion-corrected-code-baseline', 'model-protocols-release-code-baseline');
code = replace(code, 'actual iPhone evidence', 'current Responses engineering and native observation evidence');
save('.qa/freeze-protocol-release.mjs', code);
code = read('.qa/run-companion-phase-corrected.mjs');
code = replace(code, 'mobile-companion-corrected-code-baseline', 'model-protocols-release-code-baseline');
code = replace(code, 'companion-corrected-', 'protocol-f-');
code = replace(code, 'AWQA16', 'AWQA17');
for (const [before, after] of [['prepare-companion-installer-corrected', 'prepare-protocol-installer'], ['verify-companion-install-corrected', 'verify-protocol-install'],
['deploy-companion-formal-corrected', 'deploy-protocol-formal'], ['verify-companion-formal-ui-corrected', 'verify-protocol-formal-ui'],
['verify-companion-formal-artifacts-corrected', 'verify-protocol-formal-artifacts'], ['seal-companion-release-corrected', 'seal-protocol-release']]) code = replace(code, before, after);
code = replace(code, " 'prepare-install':", " 'portable-protocols': { args: ['scripts/verify-model-protocols-ui.mjs', '--app-root', release.portableRoot] },\n 'prepare-install':");
code = replace(code, 'const env = projectEnvironment({', 'const env = projectEnvironment({ PATH: path.dirname(process.execPath) + path.delimiter + process.env.PATH,');
save('.qa/run-protocol-phase.mjs', code);
code = read('.qa/prepare-companion-installer-corrected.mjs');
code = replace(code, '1.16.0', '1.17.0'); code = replace(code, 'companion-corrected-', 'protocol-f-');
code = replace(code, 'AWQA16', 'AWQA17'); code = replace(code, 'verify-companion-install-corrected', 'verify-protocol-install');
code = replace(code, 'companion-installer-qa-', 'protocol-installer-qa-');
save('.qa/prepare-protocol-installer.mjs', code);
code = read('.qa/deploy-companion-formal-corrected.mjs');
code = replace(code, '1.16.0', '__AWB_NEXT__'); code = replace(code, '1.15.0', '1.16.0'); code = replace(code, '__AWB_NEXT__', '1.17.0');
code = replace(code, 'companion-corrected-', 'protocol-f-'); code = replace(code, './desktop-f-state.mjs', './protocol-f-state.mjs');
code = replace(code, "'source-retry'", "'source'"); code = replace(code, 'protocol-f-source-retry-', 'protocol-f-source-');
code = replace(code, "'portable-companion', 'install'", "'portable-companion', 'portable-protocols', 'install'");
code = replace(code, 'assert.equal(source.tests.tests, 519); assert.equal(source.tests.pass, 519);',
"assert.ok(source.tests.tests >= 557); assert.equal(source.tests.pass, source.tests.tests); for (const key of ['fail', 'cancelled', 'skipped', 'todo']) assert.equal(source.tests[key], 0);");
save('.qa/deploy-protocol-formal.mjs', code);
for (const [source, target] of [['.qa/verify-companion-formal-ui-corrected.mjs', '.qa/verify-protocol-formal-ui.mjs'],
['.qa/verify-companion-formal-artifacts-corrected.mjs', '.qa/verify-protocol-formal-artifacts.mjs']]) {
code = replace(replace(read(source), 'companion-corrected-', 'protocol-f-'), '1.16.0', '1.17.0'); save(target, code);
}
console.log('Prepared new wrappers without changing any 1.16 receipt or script.');
@@ -0,0 +1,26 @@
✔ schema1 connections remain readable without inventing observed capabilities or migrating original state (993.7387ms)
✔ protocol, unsupported capabilities and stream choice persist; client claims cannot forge verified evidence (1262.8972ms)
✔ ability evidence belongs to exact connection and credential revision; unknown and unsupported reject required routing (0.2789ms)
✔ invalid declarations and inconsistent stream settings reject before registry, revision and credentials mutate (1249.8472ms)
✔ model registry keeps original credentials/endpoint intact, rejects stale edits, and forgets only volatile keys on restart (1447.1195ms)
✔ config import previews exact revisions, keeps secrets volatile, syncs only when enabled and preserves manual edits on conflicts (1141.502ms)
✔ real HTTP pool fails over 429 to another model and never repeats completed tools; attempts seal actual request bytes and Run identity (3887.3408ms)
✔ pool rotation and weights select members on new Runs; disabled and missing credentials are excluded without requests (3709.6143ms)
✔ 401 blocks the credential revision; updating the key restores availability without leaking error echoes (1282.2837ms)
✔ queued request cancellation and in-flight disable release only owned leases and do not dispatch another HTTP request (807.0351ms)
✔ compiled model-center HTTP API persists public profiles, blocks forged origins, and local operations never invoke a model (1915.174ms)
✔ model registry disk failures restore routing, pool profiles and credential references without partial updates (633.2153ms)
✔ model selection and saved configuration commit together and failures preserve the original mode and endpoint (651.1629ms)
✔ configuration imports roll back the entire batch and retain one usable preview after validation or commit failure (615.3665ms)
✔ hot upgrade adopts a retained virtual data root without reopening SQLite or changing rows, Provider or routing (1435.7982ms)
✔ failed adoption restores the original virtual data-root name and original open database (857.4231ms)
✔ hot upgrade refreshes retained model and importer methods while preserving credential and routing objects (940.5107ms)
✔ hot upgrade refuses active probes and restores retained model methods if installation fails (841.6437ms)
ℹ tests 18
ℹ suites 0
ℹ pass 18
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 16212.7253
@@ -0,0 +1,50 @@
✔ schema1 connections remain readable without inventing observed capabilities or migrating original state (1037.937ms)
✔ protocol, unsupported capabilities and stream choice persist; client claims cannot forge verified evidence (1358.2741ms)
✔ ability evidence belongs to exact connection and credential revision; unknown and unsupported reject required routing (0.2782ms)
✔ invalid declarations and inconsistent stream settings reject before registry, revision and credentials mutate (974.5523ms)
✔ model registry keeps original credentials/endpoint intact, rejects stale edits, and forgets only volatile keys on restart (1600.395ms)
✔ config import previews exact revisions, keeps secrets volatile, syncs only when enabled and preserves manual edits on conflicts (1414.7922ms)
✔ real HTTP pool fails over 429 to another model and never repeats completed tools; attempts seal actual request bytes and Run identity (3616.1198ms)
✔ pool rotation and weights select members on new Runs; disabled and missing credentials are excluded without requests (5575.1105ms)
✔ 401 blocks the credential revision; updating the key restores availability without leaking error echoes (2133.4425ms)
✔ queued request cancellation and in-flight disable release only owned leases and do not dispatch another HTTP request (1268.6183ms)
✔ compiled model-center HTTP API persists public profiles, blocks forged origins, and local operations never invoke a model (2694.4865ms)
✔ model registry disk failures restore routing, pool profiles and credential references without partial updates (956.6392ms)
✔ model selection and saved configuration commit together and failures preserve the original mode and endpoint (1156.5098ms)
✔ configuration imports roll back the entire batch and retain one usable preview after validation or commit failure (1026.335ms)
✔ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (3202.461ms)
✔ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (3745.2868ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"body":{"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial" (1968.7ms)
✔ Reject {"body":{"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","cal (2294.4894ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","cal (2126.2116ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supporte (2075.8148ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (8466.0713ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.2399ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (924.2029ms)
✔ real Chat SSE carries UTF8 and split tool JSON; Runtime executes only after DONE, stores partial progress and original usage (3669.2095ms)
✔ Responses typed SSE validates original item identity, terminal data, raw usage and tool result pair (2575.3447ms)
✔ incomplete network stream retains unknown attempt and never automatically switches pool or dispatches fragment tools (2055.3726ms)
✔ stopping an active HTTP stream closes only original request and keeps partial text without executing tools (1852.405ms)
▶ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data
✔ reject 0 (0.5471ms)
✔ reject 1 (0.1625ms)
✔ reject 2 (0.1862ms)
✔ reject 3 (4.872ms)
✔ reject 4 (0.1601ms)
✔ reject 5 (0.1423ms)
✔ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data (6.6955ms)
✔ context APIs retain exact request identity, paginate archived groups and reject foreign sessions/runs (3889.5644ms)
✔ delivery reports verify request body, sealed plan and original archived bytes, and show corruption as needing verification (3105.7535ms)
✔ hot upgrade adopts a retained virtual data root without reopening SQLite or changing rows, Provider or routing (1379.3547ms)
✔ failed adoption restores the original virtual data-root name and original open database (621.7576ms)
✔ hot upgrade refreshes retained model and importer methods while preserving credential and routing objects (655.0331ms)
✔ hot upgrade refuses active probes and restores retained model methods if installation fails (579.8463ms)
ℹ tests 40
ℹ suites 0
ℹ pass 40
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 32150.5413
@@ -0,0 +1,62 @@
✖ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (1300.0842ms)
✖ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (1644.0479ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial","call_i (1223.1276ms)
✔ Reject {"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","call_id":"f (1155.7996ms)
✔ Reject {"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","call_id":"" (1156.3402ms)
✔ Reject {"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supported"}],"us (1245.1388ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (4781.2321ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.1672ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (582.8269ms)
ℹ tests 9
ℹ suites 0
ℹ pass 7
ℹ fail 2
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 8429.7815
✖ failing tests:
test at tests\model-protocols.test.mjs:20:1
✖ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (1300.0842ms)
AssertionError [ERR_ASSERTION]: [{"id":"eb2e2af2-1bb6-4a23-bd69-bf3a92a58822","createdAt":"2026-10-31T14:05:31.890Z","taskId":"dfa2f733-60b0-4508-8fde-5d3bc2cabe63","runId":"b972b72a-3780-4e3c-aaad-b108aab9b9a5","kind":"model","source":"model_connection","label":"模型连接失败","phase":"failed","startedAt":"2026-10-31T14:05:31.890Z","contextPlanId":"ab822b86-8f30-496f-8e9e-27f72e9d736f","attempts":1,"connectionId":"0139bf69-e3c5-4538-b87b-682f69dbee68","modelAttemptId":"e5e0efbe-0584-4c67-b081-85594720bc0a","actualModel":"owned-model","output":"模型服务暂时无法处理请求(HTTP 500)。 Invalid status code: completed","diagnostic":{"endpoint":"http://127.0.0.1:49780/v1","model":"owned-model","route":"/responses","id":"caf45034-3f5a-454e-978e-272997785ea8","source":"model_connection","stage":"model_request","kind":"service","httpStatus":500,"attempts":1,"occurredAt":"2026-10-31T14:05:31.963Z","retryable":true,"title":"模型服务暂时无法处理请求","message":"模型服务暂时无法处理请求(HTTP 500)。 Invalid status code: completed","detail":"Invalid status code: completed","suggestion":"稍后重试,或检查模型服务状态。"},"durationMs":73,"finishedAt":"2026-10-31T14:05:31.963Z"},{"id":"55d20da0-da5b-478d-9dc3-7ad228c5b038","createdAt":"2026-10-31T14:05:32.098Z","taskId":"dfa2f733-60b0-4508-8fde-5d3bc2cabe63","runId":"b972b72a-3780-4e3c-aaad-b108aab9b9a5","kind":"error","source":"model_connection","label":"模型服务暂时无法处理请求","phase":"failed","output":"模型服务暂时无法处理请求(HTTP 500)。 Invalid status code: completed","diagnostic":{"endpoint":"http://127.0.0.1:49780/v1","model":"owned-model","route":"/responses","id":"caf45034-3f5a-454e-978e-272997785ea8","source":"model_connection","stage":"model_request","kind":"service","httpStatus":500,"attempts":1,"occurredAt":"2026-10-31T14:05:31.963Z","retryable":true,"title":"模型服务暂时无法处理请求","message":"模型服务暂时无法处理请求(HTTP 500)。 Invalid status code: completed","detail":"Invalid status code: completed","suggestion":"稍后重试,或检查模型服务状态。","runId":"b972b72a-3780-4e3c-aaad-b108aab9b9a5"}}]
+ actual - expected
+ 'failed'
- 'completed'
at TestContext.<anonymous> (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/tests/model-protocols.test.mjs:27:10)
at runNextTicks (node:internal/process/task_queues:65:5)
at process.processTimers (node:internal/timers:615:9)
at async Test.run (node:internal/test_runner/test:1409:7)
at async startSubtestAfterBootstrap (node:internal/test_runner/harness:387:3) {
generatedMessage: false,
code: 'ERR_ASSERTION',
actual: 'failed',
expected: 'completed',
operator: 'strictEqual',
diff: 'simple'
}
test at tests\model-protocols.test.mjs:44:1
✖ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (1644.0479ms)
AssertionError [ERR_ASSERTION]: [{"id":"b575895d-06b3-4f3f-9a65-4ed464bccd8a","createdAt":"2026-10-31T14:05:33.156Z","taskId":"c6865d55-03f8-403f-a2a2-d30f1fbbe818","runId":"4e58ce5b-c759-4de6-9e30-fb7e4c88aef7","kind":"model","source":"model_connection","label":"正在切换模型池连接","phase":"retrying","startedAt":"2026-10-31T14:05:33.156Z","contextPlanId":"be7b5e07-478b-4680-882b-9b35820fa634","attempts":2,"connectionId":"e338e924-e116-42e6-8faf-e8a3b7a8eacc","modelAttemptId":"c170d220-136d-467d-96d7-8314435888fd","output":"第 2 次请求失败,将自动重试(最多 3 次)。","diagnostic":{"endpoint":"http://127.0.0.1:49783/v1","model":"owned-model","route":"/responses","id":"bd9519fd-aac8-4015-9b73-cb8cda81bdb0","source":"model_connection","stage":"model_request","kind":"service","httpStatus":500,"attempts":1,"occurredAt":"2026-10-31T14:05:33.411Z","retryable":true,"title":"模型服务暂时无法处理请求","message":"模型服务暂时无法处理请求(HTTP 500)。 Invalid status code: completed","detail":"Invalid status code: completed","suggestion":"稍后重试,或检查模型服务状态。"},"durationMs":39,"finishedAt":"2026-10-31T14:05:33.411Z"},{"id":"57c56608-b583-4b20-bf63-3dc8b0a684b4","createdAt":"2026-10-31T14:05:33.539Z","taskId":"c6865d55-03f8-403f-a2a2-d30f1fbbe818","runId":"4e58ce5b-c759-4de6-9e30-fb7e4c88aef7","kind":"error","source":"model_connection","label":"模型服务暂时无法处理请求","phase":"failed","output":"模型服务暂时无法处理请求(HTTP 500)。 Invalid status code: completed","diagnostic":{"endpoint":"http://127.0.0.1:49783/v1","model":"owned-model","route":"/responses","id":"bd9519fd-aac8-4015-9b73-cb8cda81bdb0","source":"model_connection","stage":"model_request","kind":"service","httpStatus":500,"attempts":1,"occurredAt":"2026-10-31T14:05:33.411Z","retryable":true,"title":"模型服务暂时无法处理请求","message":"模型服务暂时无法处理请求(HTTP 500)。 Invalid status code: completed","detail":"Invalid status code: completed","suggestion":"稍后重试,或检查模型服务状态。","runId":"4e58ce5b-c759-4de6-9e30-fb7e4c88aef7"}}]
+ actual - expected
+ 'failed'
- 'completed'
at TestContext.<anonymous> (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/tests/model-protocols.test.mjs:51:10)
at runNextTicks (node:internal/process/task_queues:65:5)
at process.processTimers (node:internal/timers:615:9)
at async Test.run (node:internal/test_runner/test:1409:7)
at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
generatedMessage: false,
code: 'ERR_ASSERTION',
actual: 'failed',
expected: 'completed',
operator: 'strictEqual',
diff: 'simple'
}
@@ -0,0 +1,18 @@
✔ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (1964.1938ms)
✔ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (2478.4576ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"body":{"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial" (1265.0775ms)
✔ Reject {"body":{"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","cal (1227.6379ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","cal (1136.9406ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supporte (1321.2282ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (4951.6993ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.1418ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (597.4172ms)
ℹ tests 9
ℹ suites 0
ℹ pass 9
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 10114.327
@@ -0,0 +1,88 @@
✔ schema1 connections remain readable without inventing observed capabilities or migrating original state (944.3724ms)
✔ protocol, unsupported capabilities and stream choice persist; client claims cannot forge verified evidence (1317.1831ms)
✔ ability evidence belongs to exact connection and credential revision; unknown and unsupported reject required routing (0.2657ms)
✔ invalid declarations and inconsistent stream settings reject before registry, revision and credentials mutate (1075.5743ms)
✔ model registry keeps original credentials/endpoint intact, rejects stale edits, and forgets only volatile keys on restart (1489.4294ms)
✔ config import previews exact revisions, keeps secrets volatile, syncs only when enabled and preserves manual edits on conflicts (1367.5007ms)
✔ real HTTP pool fails over 429 to another model and never repeats completed tools; attempts seal actual request bytes and Run identity (3491.1296ms)
✔ pool rotation and weights select members on new Runs; disabled and missing credentials are excluded without requests (4968.0827ms)
✔ 401 blocks the credential revision; updating the key restores availability without leaking error echoes (2097.2378ms)
✔ queued request cancellation and in-flight disable release only owned leases and do not dispatch another HTTP request (1272.7983ms)
✔ compiled model-center HTTP API persists public profiles, blocks forged origins, and local operations never invoke a model (2366.4369ms)
✔ model registry disk failures restore routing, pool profiles and credential references without partial updates (987.9056ms)
✔ model selection and saved configuration commit together and failures preserve the original mode and endpoint (1324.5368ms)
✔ configuration imports roll back the entire batch and retain one usable preview after validation or commit failure (952.8682ms)
✖ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (2986.5252ms)
✖ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (3526.1035ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"body":{"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial" (1989.988ms)
✔ Reject {"body":{"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","cal (2082.823ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","cal (1856.9479ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supporte (1378.7842ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (7309.3255ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.169ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (1037.6843ms)
✔ context APIs retain exact request identity, paginate archived groups and reject foreign sessions/runs (4418.9194ms)
✔ delivery reports verify request body, sealed plan and original archived bytes, and show corruption as needing verification (2662.8881ms)
✔ direct code evidence remains selected while recent historical-index browsing is archived within the complete request budget (1719.2658ms)
✔ one long user turn becomes bounded complete groups with exact UTF-8 history and recent JSON arguments (3928.2304ms)
✔ tool definitions, exact system bytes, output reserve and remaining Run/Task budget are accounted before requests (757.9403ms)
✔ oversized required user instructions pause intact and can be explicitly resumed after increasing the window (1371.5941ms)
✔ all task instructions, constraints and acceptance remain in the runtime system instead of an eight-item or byte prefix (1131.8719ms)
✔ a newest manageable multi-tool group stays exact even beyond the working target; oversized earlier groups are archived (840.6565ms)
✔ invalid orphan result is rejected before archive or network preparation (609.8957ms)
✔ invalid missing result is rejected before archive or network preparation (635.866ms)
✔ invalid wrong result is rejected before archive or network preparation (530.4749ms)
✔ invalid duplicate call is rejected before archive or network preparation (592.2473ms)
✔ a group larger than the artifact limit uses verified Unicode-safe chunks without replacing tool JSON (1551.1635ms)
✔ changed or missing archived evidence stops reuse and is never silently recreated (858.6325ms)
✔ transaction failure prevents publishing a plan and any corresponding model request (849.1558ms)
✔ the Provider rejects body mutation after a sealed plan before calling HTTP (650.4978ms)
✔ HTTP failure and explicit same-Task resume reuse exact completed results and request hashes without repeating a command (4879.5458ms)
✔ a real process exit after plan persistence marks interruption and never issues an automatic model request (1180.1298ms)
ℹ tests 41
ℹ suites 0
ℹ pass 39
ℹ fail 2
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 42639.5415
✖ failing tests:
test at tests\model-protocols.test.mjs:20:1
✖ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (2986.5252ms)
AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
undefined !== 274
at TestContext.<anonymous> (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/tests/model-protocols.test.mjs:31:10)
at runNextTicks (node:internal/process/task_queues:65:5)
at process.processTimers (node:internal/timers:615:9)
at async Test.run (node:internal/test_runner/test:1409:7)
at async startSubtestAfterBootstrap (node:internal/test_runner/harness:387:3) {
generatedMessage: true,
code: 'ERR_ASSERTION',
actual: undefined,
expected: 274,
operator: 'strictEqual',
diff: 'simple'
}
test at tests\model-protocols.test.mjs:44:1
✖ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (3526.1035ms)
AssertionError [ERR_ASSERTION]: The expression evaluated to a falsy value:
assert.ok(verifyPlanEvidence(f.store, f.session, plan))
at TestContext.<anonymous> (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/tests/model-protocols.test.mjs:57:10)
at async Test.run (node:internal/test_runner/test:1409:7)
at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
generatedMessage: true,
code: 'ERR_ASSERTION',
actual: undefined,
expected: true,
operator: '==',
diff: 'simple'
}
@@ -0,0 +1,74 @@
✔ schema1 connections remain readable without inventing observed capabilities or migrating original state (1014.0062ms)
✔ protocol, unsupported capabilities and stream choice persist; client claims cannot forge verified evidence (1089.2405ms)
✔ ability evidence belongs to exact connection and credential revision; unknown and unsupported reject required routing (0.2615ms)
✔ invalid declarations and inconsistent stream settings reject before registry, revision and credentials mutate (1048.5511ms)
✔ model registry keeps original credentials/endpoint intact, rejects stale edits, and forgets only volatile keys on restart (1438.3081ms)
✔ config import previews exact revisions, keeps secrets volatile, syncs only when enabled and preserves manual edits on conflicts (1155.2807ms)
✔ real HTTP pool fails over 429 to another model and never repeats completed tools; attempts seal actual request bytes and Run identity (3464.8728ms)
✔ pool rotation and weights select members on new Runs; disabled and missing credentials are excluded without requests (5365.1931ms)
✔ 401 blocks the credential revision; updating the key restores availability without leaking error echoes (1980.7633ms)
✔ queued request cancellation and in-flight disable release only owned leases and do not dispatch another HTTP request (1169.7538ms)
✔ compiled model-center HTTP API persists public profiles, blocks forged origins, and local operations never invoke a model (2214.3522ms)
✔ model registry disk failures restore routing, pool profiles and credential references without partial updates (642.3498ms)
✔ model selection and saved configuration commit together and failures preserve the original mode and endpoint (647.6397ms)
✔ configuration imports roll back the entire batch and retain one usable preview after validation or commit failure (634.1912ms)
✖ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (1933.8954ms)
✖ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (2566.2122ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial","call_i (1908.0118ms)
✔ Reject {"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","call_id":"f (1995.5451ms)
✔ Reject {"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","call_id":"" (2060.8771ms)
✔ Reject {"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supported"}],"us (1965.412ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (7930.9742ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.159ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (653.3064ms)
ℹ tests 23
ℹ suites 0
ℹ pass 21
ℹ fail 2
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 18836.6552
✖ failing tests:
test at tests\model-protocols.test.mjs:20:1
✖ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (1933.8954ms)
AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+ actual - expected
+ 'failed'
- 'completed'
at TestContext.<anonymous> (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/tests/model-protocols.test.mjs:27:10)
at async Test.run (node:internal/test_runner/test:1409:7)
at async startSubtestAfterBootstrap (node:internal/test_runner/harness:387:3) {
generatedMessage: true,
code: 'ERR_ASSERTION',
actual: 'failed',
expected: 'completed',
operator: 'strictEqual',
diff: 'simple'
}
test at tests\model-protocols.test.mjs:44:1
✖ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (2566.2122ms)
AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+ actual - expected
+ 'failed'
- 'completed'
at TestContext.<anonymous> (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/tests/model-protocols.test.mjs:51:10)
at runNextTicks (node:internal/process/task_queues:65:5)
at process.processTimers (node:internal/timers:615:9)
at async Test.run (node:internal/test_runner/test:1409:7)
at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
generatedMessage: true,
code: 'ERR_ASSERTION',
actual: 'failed',
expected: 'completed',
operator: 'strictEqual',
diff: 'simple'
}
@@ -0,0 +1,46 @@
✔ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (3199.1367ms)
✔ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (3554.0138ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"body":{"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial" (2181.8467ms)
✔ Reject {"body":{"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","cal (2208.917ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","cal (1925.746ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supporte (1989.7759ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (8307.0909ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.1543ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (822.0029ms)
✔ real Chat SSE carries UTF8 and split tool JSON; Runtime executes only after DONE, stores partial progress and original usage (3728.7601ms)
✔ Responses typed SSE validates original item identity, terminal data, raw usage and tool result pair (3155.1217ms)
✔ incomplete network stream retains unknown attempt and never automatically switches pool or dispatches fragment tools (2380.5295ms)
✔ stopping an active HTTP stream closes only original request and keeps partial text without executing tools (2079.5337ms)
▶ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data
✔ reject 0 (0.8265ms)
✔ reject 1 (0.2276ms)
✔ reject 2 (0.2375ms)
✔ reject 3 (5.0659ms)
✔ reject 4 (0.1517ms)
✔ reject 5 (0.1274ms)
✔ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data (7.4981ms)
✔ direct code evidence remains selected while recent historical-index browsing is archived within the complete request budget (2683.3478ms)
✔ one long user turn becomes bounded complete groups with exact UTF-8 history and recent JSON arguments (3879.5931ms)
✔ tool definitions, exact system bytes, output reserve and remaining Run/Task budget are accounted before requests (709.4956ms)
✔ oversized required user instructions pause intact and can be explicitly resumed after increasing the window (1281.3723ms)
✔ all task instructions, constraints and acceptance remain in the runtime system instead of an eight-item or byte prefix (1072.6324ms)
✔ a newest manageable multi-tool group stays exact even beyond the working target; oversized earlier groups are archived (892.4577ms)
✔ invalid orphan result is rejected before archive or network preparation (652.0142ms)
✔ invalid missing result is rejected before archive or network preparation (571.7852ms)
✔ invalid wrong result is rejected before archive or network preparation (610.133ms)
✔ invalid duplicate call is rejected before archive or network preparation (619.9279ms)
✔ a group larger than the artifact limit uses verified Unicode-safe chunks without replacing tool JSON (1610.2754ms)
✔ changed or missing archived evidence stops reuse and is never silently recreated (799.0783ms)
✔ transaction failure prevents publishing a plan and any corresponding model request (788.6798ms)
✔ the Provider rejects body mutation after a sealed plan before calling HTTP (629.4382ms)
✔ HTTP failure and explicit same-Task resume reuse exact completed results and request hashes without repeating a command (5516.4905ms)
✔ a real process exit after plan persistence marks interruption and never issues an automatic model request (1351.5032ms)
ℹ tests 36
ℹ suites 0
ℹ pass 36
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 35257.3316
@@ -0,0 +1,39 @@
✔ owned CC Switch SQLite imports the selected TOML Responses wire_api and leaves unknown clients untouched (1434.5284ms)
✔ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (4628.0791ms)
✔ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (4708.4754ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"body":{"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial" (2367.0582ms)
✔ Reject {"body":{"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","cal (2476.4776ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","cal (2317.129ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supporte (2229.2344ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (9390.7418ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.1644ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (1040.2795ms)
✔ capability routing skips text-only members before acquiring a lease and records exact refusal in the original Run (2401.3744ms)
✔ all incompatible members reject with no network request, no lease, preserved Task and explicit non-dispatched ledger (2350.372ms)
✔ explicit synthetic tool probe verifies only actual matching call, never executes it and loses verification on revision changes (1549.1779ms)
✔ plain text probe cannot verify tools; wrong tool nonce and client supplied evidence cannot claim verified support (1817.464ms)
✔ a disabled connection invalidates an in-flight probe even when toggle mutates its original registry object (925.1592ms)
✔ real Chat SSE carries UTF8 and split tool JSON; Runtime executes only after DONE, stores partial progress and original usage (4639.8283ms)
✔ Responses typed SSE validates original item identity, terminal data, raw usage and tool result pair (4264.8592ms)
✔ incomplete network stream retains unknown attempt and never automatically switches pool or dispatches fragment tools (2665.4274ms)
✔ stopping an active HTTP stream closes only original request and keeps partial text without executing tools (1507.8489ms)
✔ explicit streaming probe uses actual SSE despite unknown declaration and binds its own original attempt (840.1404ms)
✔ stream timeout closes the original HTTP response, preserves redacted capture and never retries (1045.3911ms)
✔ SSE accepts CR-only final event boundaries and retains exact invalid UTF8 bytes without dispatch (1.144ms)
▶ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data
✔ reject 0 (0.2171ms)
✔ reject 1 (0.1405ms)
✔ reject 2 (0.1213ms)
✔ reject 3 (5.2154ms)
✔ reject 4 (0.1261ms)
✔ reject 5 (0.1235ms)
✔ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data (6.4556ms)
ℹ tests 29
ℹ suites 0
ℹ pass 29
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 24266.1287
@@ -0,0 +1,47 @@
✔ pairing is short lived, single-use, generation-bound and validates exact permissions (1107.2974ms)
✔ device transaction failure neither consumes the ticket nor leaves a credential (971.6891ms)
✔ devices have project and action boundaries; rotation, expiration, revocation and host changes reject old authority (1084.1224ms)
✔ persistent registry stores hashes and a bounded audit, never raw credentials (967.9301ms)
✔ real TLS verifies IP SAN, CA chain, encrypted restart and public fingerprint (1195.4954ms)
✔ TLS sealing failure leaves listener closed and no plaintext or successful TLS registration (968.7866ms)
✔ HTTP pairing is concurrent single-use; secure cookie and strict origin/host checks (1062.3071ms)
✔ gateway never exposes management routes, arbitrary assets or unpaired project data (1249.5189ms)
✔ authorized project events paginate from original seq without leaking other project events or payloads (1700.9864ms)
✔ remote task projection excludes raw model history, host configuration, tool input and credentials (1442.3281ms)
✔ permission checks apply to direct remote mutation, not only UI controls (1047.2542ms)
✔ approval fingerprint and exact task/revision are bound; duplicate requests never execute twice (1540.1497ms)
✔ unknown original operation is recorded and never replayed after response failure (1245.2744ms)
✔ revocation and gateway closure reject old access without invoking runtime stop (777.859ms)
✔ rate limits reject pairing brute force and do not persist candidate secrets (750.6093ms)
✔ real Windows CurrentUser DPAPI persists only a sealed CA and reopens it (2582.997ms)
✔ mobile approval executes the original Runtime command exactly once and rejects the later desktop decision (1981.9998ms)
✔ desktop rejection wins a concurrent mobile request and original command remains unexecuted (1623.7516ms)
✔ phone stop waits for the original process exit and prevents later heartbeats and command replay (2266.9916ms)
✔ Run replacement during parent-stop wait is rejected rather than stopping the new Run (631.3227ms)
✔ diff pages preserve source SHA; report reads verify blobs and exclude unrelated artifact kinds (959.4308ms)
✔ hot upgrade refuses an active gateway and restores closed gateway ownership on failure (640.9049ms)
✔ damaged device permissions or expiry fail closed and never overwrite the original registry (699.9645ms)
✔ device revocation destroys its actual authenticated TLS connection (790.7691ms)
✔ real Chat SSE carries UTF8 and split tool JSON; Runtime executes only after DONE, stores partial progress and original usage (4304.0643ms)
✔ Responses typed SSE validates original item identity, terminal data, raw usage and tool result pair (3748.2139ms)
✔ incomplete network stream retains unknown attempt and never automatically switches pool or dispatches fragment tools (2501.0255ms)
✔ stopping an active HTTP stream closes only original request and keeps partial text without executing tools (2356.3092ms)
✔ explicit streaming probe uses actual SSE despite unknown declaration and binds its own original attempt (1192.6798ms)
✔ stream timeout closes the original HTTP response, preserves redacted capture and never retries (1412.0866ms)
✔ SSE accepts CR-only final event boundaries and retains exact invalid UTF8 bytes without dispatch (1.9856ms)
▶ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data
✔ reject 0 (0.3728ms)
✔ reject 1 (0.2031ms)
✔ reject 2 (0.1649ms)
✔ reject 3 (4.8382ms)
✔ reject 4 (0.1347ms)
✔ reject 5 (0.119ms)
✔ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data (6.4917ms)
ℹ tests 38
ℹ suites 0
ℹ pass 38
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 29527.5584
@@ -0,0 +1,34 @@
✔ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (3865.633ms)
✔ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (3960.2019ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"body":{"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial" (1785.3953ms)
✔ Reject {"body":{"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","cal (2155.3008ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","cal (1944.6009ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supporte (2156.3372ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (8042.6422ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.1578ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (931.3403ms)
✔ capability routing skips text-only members before acquiring a lease and records exact refusal in the original Run (2287.0886ms)
✔ all incompatible members reject with no network request, no lease, preserved Task and explicit non-dispatched ledger (2077.0141ms)
✔ explicit synthetic tool probe verifies only actual matching call, never executes it and loses verification on revision changes (1159.9859ms)
✔ plain text probe cannot verify tools; wrong tool nonce and client supplied evidence cannot claim verified support (1076.1412ms)
✔ real Chat SSE carries UTF8 and split tool JSON; Runtime executes only after DONE, stores partial progress and original usage (4006.8909ms)
✔ Responses typed SSE validates original item identity, terminal data, raw usage and tool result pair (3648.3519ms)
✔ incomplete network stream retains unknown attempt and never automatically switches pool or dispatches fragment tools (2541.1146ms)
✔ stopping an active HTTP stream closes only original request and keeps partial text without executing tools (1301.2753ms)
▶ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data
✔ reject 0 (0.8398ms)
✔ reject 1 (0.203ms)
✔ reject 2 (0.2546ms)
✔ reject 3 (5.3118ms)
✔ reject 4 (0.177ms)
✔ reject 5 (0.1507ms)
✔ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data (7.9792ms)
ℹ tests 24
ℹ suites 0
ℹ pass 24
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 18348.2088
@@ -0,0 +1,52 @@
✔ schema1 connections remain readable without inventing observed capabilities or migrating original state (892.4845ms)
✔ protocol, unsupported capabilities and stream choice persist; client claims cannot forge verified evidence (966.5389ms)
✔ ability evidence belongs to exact connection and credential revision; unknown and unsupported reject required routing (0.3516ms)
✔ invalid declarations and inconsistent stream settings reject before registry, revision and credentials mutate (681.927ms)
✔ model registry keeps original credentials/endpoint intact, rejects stale edits, and forgets only volatile keys on restart (1259.4085ms)
✔ config import previews exact revisions, keeps secrets volatile, syncs only when enabled and preserves manual edits on conflicts (996.4614ms)
✔ real HTTP pool fails over 429 to another model and never repeats completed tools; attempts seal actual request bytes and Run identity (4172.9802ms)
✔ pool rotation and weights select members on new Runs; disabled and missing credentials are excluded without requests (5369.0176ms)
✔ 401 blocks the credential revision; updating the key restores availability without leaking error echoes (2136.1904ms)
✔ queued request cancellation and in-flight disable release only owned leases and do not dispatch another HTTP request (1285.1659ms)
✔ compiled model-center HTTP API persists public profiles, blocks forged origins, and local operations never invoke a model (2522.6777ms)
✔ model registry disk failures restore routing, pool profiles and credential references without partial updates (1125.4403ms)
✔ model selection and saved configuration commit together and failures preserve the original mode and endpoint (1857.4609ms)
✔ configuration imports roll back the entire batch and retain one usable preview after validation or commit failure (874.6812ms)
✔ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (3872.5619ms)
✔ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (3712.8468ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"body":{"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial" (2077.5477ms)
✔ Reject {"body":{"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","cal (2176.6625ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","cal (2038.6461ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supporte (1629.5944ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (7923.2614ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.1597ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (982.9388ms)
✔ capability routing skips text-only members before acquiring a lease and records exact refusal in the original Run (3118.5909ms)
✔ all incompatible members reject with no network request, no lease, preserved Task and explicit non-dispatched ledger (1579.9588ms)
✔ explicit synthetic tool probe verifies only actual matching call, never executes it and loses verification on revision changes (997.4663ms)
✔ plain text probe cannot verify tools; wrong tool nonce and client supplied evidence cannot claim verified support (1226.9391ms)
✔ real Chat SSE carries UTF8 and split tool JSON; Runtime executes only after DONE, stores partial progress and original usage (4102.3699ms)
✔ Responses typed SSE validates original item identity, terminal data, raw usage and tool result pair (3736.7539ms)
✔ incomplete network stream retains unknown attempt and never automatically switches pool or dispatches fragment tools (2254.8904ms)
✔ stopping an active HTTP stream closes only original request and keeps partial text without executing tools (1331.031ms)
▶ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data
✔ reject 0 (0.6407ms)
✔ reject 1 (0.1739ms)
✔ reject 2 (0.2062ms)
✔ reject 3 (4.8471ms)
✔ reject 4 (0.1743ms)
✔ reject 5 (0.1468ms)
✔ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data (6.8946ms)
✔ hot upgrade adopts a retained virtual data root without reopening SQLite or changing rows, Provider or routing (1591.9706ms)
✔ failed adoption restores the original virtual data-root name and original open database (834.5372ms)
✔ hot upgrade refreshes retained model and importer methods while preserving credential and routing objects (1088.6735ms)
✔ hot upgrade refuses active probes and restores retained model methods if installation fails (900.0117ms)
ℹ tests 42
ℹ suites 0
ℹ pass 42
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 33273.7612
@@ -0,0 +1,20 @@
> local-agent-workbench@1.17.0 build
> tsc --noEmit && vite build && node scripts/build-mobile.mjs && node scripts/build-server.mjs
vite v7.3.7 building client environment for production...
transforming...
✓ 1725 modules transformed.
rendering chunks...
computing gzip size...
dist-client/.vite/manifest.json 0.55 kB │ gzip: 0.22 kB
dist-client/index.html 0.58 kB │ gzip: 0.40 kB
dist-client/mobile.html 0.82 kB │ gzip: 0.46 kB
dist-client/assets/mobile-BOUNcCmV.css 11.42 kB │ gzip: 3.04 kB
dist-client/assets/desktop-5MRuZeKa.css 135.46 kB │ gzip: 25.93 kB
dist-client/assets/mobile-Dz6Bo2B9.js 21.74 kB │ gzip: 7.75 kB
dist-client/assets/x-Cc3T3VuL.js 225.51 kB │ gzip: 70.77 kB
dist-client/assets/desktop-DYTmXJDO.js 281.89 kB │ gzip: 79.46 kB
✓ built in 1.15s
Mobile shell built: 3 allowed assets; API and project data are never cached.
Server bundle built.
@@ -0,0 +1,20 @@
> local-agent-workbench@1.16.0 build
> tsc --noEmit && vite build && node scripts/build-mobile.mjs && node scripts/build-server.mjs
vite v7.3.7 building client environment for production...
transforming...
✓ 1725 modules transformed.
rendering chunks...
computing gzip size...
dist-client/.vite/manifest.json 0.55 kB │ gzip: 0.22 kB
dist-client/index.html 0.58 kB │ gzip: 0.40 kB
dist-client/mobile.html 0.82 kB │ gzip: 0.46 kB
dist-client/assets/mobile-CPi1xsK3.css 11.10 kB │ gzip: 2.97 kB
dist-client/assets/desktop-BGfi7Mev.css 135.24 kB │ gzip: 25.89 kB
dist-client/assets/mobile-D_V22qXj.js 21.74 kB │ gzip: 7.75 kB
dist-client/assets/x-Cc3T3VuL.js 225.51 kB │ gzip: 70.77 kB
dist-client/assets/desktop-Da_uXwdD.js 280.60 kB │ gzip: 79.05 kB
✓ built in 1.12s
Mobile shell built: 3 allowed assets; API and project data are never cached.
Server bundle built.
@@ -0,0 +1,20 @@
> local-agent-workbench@1.17.0 build
> tsc --noEmit && vite build && node scripts/build-mobile.mjs && node scripts/build-server.mjs
vite v7.3.7 building client environment for production...
transforming...
✓ 1725 modules transformed.
rendering chunks...
computing gzip size...
dist-client/.vite/manifest.json 0.55 kB │ gzip: 0.22 kB
dist-client/index.html 0.58 kB │ gzip: 0.40 kB
dist-client/mobile.html 0.82 kB │ gzip: 0.46 kB
dist-client/assets/mobile-BOUNcCmV.css 11.42 kB │ gzip: 3.04 kB
dist-client/assets/desktop-5MRuZeKa.css 135.46 kB │ gzip: 25.93 kB
dist-client/assets/mobile-Dz6Bo2B9.js 21.74 kB │ gzip: 7.75 kB
dist-client/assets/x-Cc3T3VuL.js 225.51 kB │ gzip: 70.77 kB
dist-client/assets/desktop-BSNXF-id.js 282.30 kB │ gzip: 79.56 kB
✓ built in 1.18s
Mobile shell built: 3 allowed assets; API and project data are never cached.
Server bundle built.
@@ -0,0 +1,35 @@
✔ model registry keeps original credentials/endpoint intact, rejects stale edits, and forgets only volatile keys on restart (2370.3633ms)
✔ config import previews exact revisions, keeps secrets volatile, syncs only when enabled and preserves manual edits on conflicts (2117.705ms)
✔ real HTTP pool fails over 429 to another model and never repeats completed tools; attempts seal actual request bytes and Run identity (6897.5971ms)
✔ pool rotation and weights select members on new Runs; disabled and missing credentials are excluded without requests (9250.5979ms)
✔ 401 blocks the credential revision; updating the key restores availability without leaking error echoes (3217.6818ms)
✔ queued request cancellation and in-flight disable release only owned leases and do not dispatch another HTTP request (2598.7156ms)
✔ compiled model-center HTTP API persists public profiles, blocks forged origins, and local operations never invoke a model (4360.1888ms)
✔ model registry disk failures restore routing, pool profiles and credential references without partial updates (1893.4263ms)
✔ model selection and saved configuration commit together and failures preserve the original mode and endpoint (2166.135ms)
✔ configuration imports roll back the entire batch and retain one usable preview after validation or commit failure (1609.896ms)
✔ HTTP download preserves original malformed SSE bytes and artifact SHA instead of decoding invalid UTF8 (3952.9458ms)
✔ real Chat SSE carries UTF8 and split tool JSON; Runtime executes only after DONE, stores partial progress and original usage (7105.3473ms)
✔ Responses typed SSE validates original item identity, terminal data, raw usage and tool result pair (6291.606ms)
✔ incomplete network stream retains unknown attempt and never automatically switches pool or dispatches fragment tools (4041.245ms)
✔ stopping an active HTTP stream closes only original request and keeps partial text without executing tools (4006.955ms)
✔ explicit streaming probe uses actual SSE despite unknown declaration and binds its own original attempt (2517.4313ms)
✔ coalesced initial SSE events publish the latest partial state even when the upstream stays idle (4539.5055ms)
✔ stream timeout closes the original HTTP response, preserves redacted capture and never retries (2557.65ms)
✔ SSE accepts CR-only final event boundaries and retains exact invalid UTF8 bytes without dispatch (0.9326ms)
▶ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data
✔ reject 0 (0.1952ms)
✔ reject 1 (0.234ms)
✔ reject 2 (0.2449ms)
✔ reject 3 (5.2352ms)
✔ reject 4 (0.1967ms)
✔ reject 5 (0.1693ms)
✔ stream parser rejects truncation, finish length, malformed JSON, duplicate sequence, changed final item and oversized data (6.8177ms)
ℹ tests 26
ℹ suites 0
ℹ pass 26
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 36601.3161
@@ -0,0 +1,20 @@
> local-agent-workbench@1.16.0 build
> tsc --noEmit && vite build && node scripts/build-mobile.mjs && node scripts/build-server.mjs
vite v7.3.7 building client environment for production...
transforming...
✓ 1725 modules transformed.
rendering chunks...
computing gzip size...
dist-client/.vite/manifest.json 0.55 kB │ gzip: 0.22 kB
dist-client/index.html 0.58 kB │ gzip: 0.40 kB
dist-client/mobile.html 0.82 kB │ gzip: 0.46 kB
dist-client/assets/mobile-BOUNcCmV.css 11.42 kB │ gzip: 3.04 kB
dist-client/assets/desktop-BGfi7Mev.css 135.24 kB │ gzip: 25.89 kB
dist-client/assets/mobile-Dz6Bo2B9.js 21.74 kB │ gzip: 7.75 kB
dist-client/assets/x-Cc3T3VuL.js 225.51 kB │ gzip: 70.77 kB
dist-client/assets/desktop-xnPDmsUV.js 281.02 kB │ gzip: 79.19 kB
✓ built in 1.17s
Mobile shell built: 3 allowed assets; API and project data are never cached.
Server bundle built.
@@ -0,0 +1,6 @@
PASS actual form saves Responses and declared stream, refresh preserves both and narrow editor remains operable
PASS separate text, nonce tool and actual SSE probes bind only their requested capability and own attempts
PASS actual pooled task excludes text-only before dispatch, executes original tool once and downloads SHA-matched original SSE
PASS desktop and actual paired gateway show same bounded stream; mobile stops original HTTP with zero fragment tool dispatch or replay
PASS truncated output keeps explicit unknown ledger, partial evidence and no automatic failover or incomplete tool execution
EVIDENCE D:\Ai-project\REA逆向\Agent-Workbench\.qa\model-protocols-ui-e98e4005-43e7-4171-936e-24b8df2b4b6b
@@ -0,0 +1,6 @@
PASS actual form saves Responses and declared stream, refresh preserves both and narrow editor remains operable
PASS separate text, nonce tool and actual SSE probes bind only their requested capability and own attempts
PASS actual pooled task excludes text-only before dispatch, executes original tool once and downloads SHA-matched original SSE
PASS desktop and actual paired gateway show same bounded stream; mobile stops original HTTP with zero fragment tool dispatch or replay
PASS truncated output keeps explicit unknown ledger, partial evidence and no automatic failover or incomplete tool execution
EVIDENCE D:\Ai-project\REA逆向\Agent-Workbench\.qa\model-protocols-ui-5c599342-cbbd-4019-8865-ee1d7b4570e4
@@ -0,0 +1,17 @@
EVIDENCE D:\Ai-project\REA逆向\Agent-Workbench\.qa\model-protocols-ui-5784abd2-38f1-46fa-a2fb-304438702da0
node:internal/modules/run_main:107
triggerUncaughtException(
^
locator.selectOption: Timeout 30000ms exceeded.
Call log:
- waiting for getByRole('dialog', { name: '模型中心', exact: true }).getByRole('form', { name: '编辑模型连接', exact: true }).getByLabel('接口协议', { exact: true })
at D:\Ai-project\REA逆向\Agent-Workbench\scripts\verify-model-protocols-ui.mjs:63:50 {
log: [
" - waiting for getByRole('dialog', { name: '模型中心', exact: true }).getByRole('form', { name: '编辑模型连接', exact: true }).getByLabel('接口协议', { exact: true })"
],
name: 'TimeoutError'
}
Node.js v24.22.1
@@ -0,0 +1,95 @@
PASS actual form saves Responses and declared stream, refresh preserves both and narrow editor remains operable
PASS separate text, nonce tool and actual SSE probes bind only their requested capability and own attempts
PASS actual pooled task excludes text-only before dispatch, executes original tool once and downloads SHA-matched original SSE
EVIDENCE D:\Ai-project\REA逆向\Agent-Workbench\.qa\model-protocols-ui-fedc8bdc-5ea4-467f-bb15-93678b03d837
node:internal/modules/run_main:107
triggerUncaughtException(
^
ExpectError: expect(locator).toContainText(expected) failed
Locator: locator('.mobile-model-stream')
Expected substring: "正在核对当前项目"
Timeout: 5000ms
Error: element(s) not found
Call log:
- Expect "toContainText" locator('.mobile-model-stream') with timeout 5000ms
- waiting for locator('.mobile-model-stream')
at captureRawStack (D:\Ai-project\REA逆向\Agent-Workbench\node_modules\playwright-core\lib\coreBundle.js:9282:17)
at callMatcherAsStep (D:\Ai-project\REA逆向\Agent-Workbench\node_modules\playwright\lib\matchers\expect.js:13316:57)
at Object.toContainText (D:\Ai-project\REA逆向\Agent-Workbench\node_modules\playwright\lib\matchers\expect.js:13308:23)
at D:\Ai-project\REA逆向\Agent-Workbench\scripts\verify-model-protocols-ui.mjs:126:56 {
matcherResult: {
name: 'toContainText',
expected: '正在核对当前项目',
message: 'expect(locator).toContainText(expected) failed\n' +
'\n' +
"Locator: locator('.mobile-model-stream')\n" +
'Expected substring: "正在核对当前项目"\n' +
'Timeout: 5000ms\n' +
'Error: element(s) not found\n' +
'\n' +
'Call log:\n' +
` - Expect "toContainText" locator('.mobile-model-stream') with timeout 5000ms\n` +
" - waiting for locator('.mobile-model-stream')\n",
pass: false,
actual: undefined,
log: [
` - Expect "toContainText" locator('.mobile-model-stream') with timeout 5000ms`,
" - waiting for locator('.mobile-model-stream')"
],
timeout: 5000,
ariaSnapshot: '- banner:\n' +
' - strong: Agent Workbench\n' +
' - text: 手机伴随端 · Windows 执行宿主\n' +
' - button "刷新任务"\n' +
'- text: 已同步 · 10/31 22:29 观察\n' +
'- main:\n' +
' - button "项目任务"\n' +
' - text: project\n' +
' - heading "受控完整流程:读取项目后结束。" [level=1]\n' +
' - text: 执行中 修订 49\n' +
' - group: 完整任务目标\n' +
' - group: 原任务与运行\n' +
' - navigation "任务内容":\n' +
' - button "执行进展"\n' +
' - button "文件差异"\n' +
' - button "检查与报告"\n' +
' - heading "原执行记录" [level=2]\n' +
' - text: 0 / 0 工具完成\n' +
' - list:\n' +
' - listitem:\n' +
' - strong: 模型响应已收到\n' +
' - text: 已完成\n' +
' - group: 查看原输出片段\n' +
' - listitem:\n' +
' - strong: 查看项目文件\n' +
' - text: 已完成\n' +
' - group: 查看原输出片段\n' +
' - listitem:\n' +
' - strong: 模型响应已收到\n' +
' - text: 已完成\n' +
' - group: 查看原输出片段\n' +
' - listitem:\n' +
' - strong: 请求模型服务\n' +
' - text: 执行中\n' +
' - strong: 正在接收模型输出\n' +
' - text: 16 字符 · 片段不执行工具 正在核对当前项目,尚未执行工具。\n' +
' - group: 查看原输出片段\n' +
' - article:\n' +
' - text: 代理回复 · 10/31 22:29\n' +
' - paragraph: 正在核对当前项目。\n' +
' - article:\n' +
' - text: 代理回复 · 10/31 22:29\n' +
' - paragraph: 项目读取完成。\n' +
' - paragraph: 手机显示最近60条执行记录和8条回复;完整记录保留在电脑原会话。\n' +
' - text: 任务累计用量\n' +
' - strong: 0 tokens / 1,000,000\n' +
' - text: 停止仍由电脑等待所属工具退出\n' +
' - button "停止原任务"'
}
}
Node.js v24.22.1
@@ -0,0 +1,20 @@
> local-agent-workbench@1.17.0 build
> tsc --noEmit && vite build && node scripts/build-mobile.mjs && node scripts/build-server.mjs
vite v7.3.7 building client environment for production...
transforming...
✓ 1725 modules transformed.
rendering chunks...
computing gzip size...
dist-client/.vite/manifest.json 0.55 kB │ gzip: 0.22 kB
dist-client/index.html 0.58 kB │ gzip: 0.40 kB
dist-client/mobile.html 0.82 kB │ gzip: 0.46 kB
dist-client/assets/mobile-BOUNcCmV.css 11.42 kB │ gzip: 3.04 kB
dist-client/assets/desktop-5MRuZeKa.css 135.46 kB │ gzip: 25.93 kB
dist-client/assets/mobile-Dz6Bo2B9.js 21.74 kB │ gzip: 7.75 kB
dist-client/assets/x-Cc3T3VuL.js 225.51 kB │ gzip: 70.77 kB
dist-client/assets/desktop-DYTmXJDO.js 281.89 kB │ gzip: 79.46 kB
✓ built in 1.21s
Mobile shell built: 3 allowed assets; API and project data are never cached.
Server bundle built.
@@ -0,0 +1,20 @@
> local-agent-workbench@1.16.0 build
> tsc --noEmit && vite build && node scripts/build-mobile.mjs && node scripts/build-server.mjs
vite v7.3.7 building client environment for production...
transforming...
✓ 1725 modules transformed.
rendering chunks...
computing gzip size...
dist-client/.vite/manifest.json 0.55 kB │ gzip: 0.22 kB
dist-client/index.html 0.58 kB │ gzip: 0.40 kB
dist-client/mobile.html 0.82 kB │ gzip: 0.46 kB
dist-client/assets/mobile-BOUNcCmV.css 11.42 kB │ gzip: 3.04 kB
dist-client/assets/desktop-5MRuZeKa.css 135.46 kB │ gzip: 25.93 kB
dist-client/assets/mobile-Dz6Bo2B9.js 21.74 kB │ gzip: 7.75 kB
dist-client/assets/x-Cc3T3VuL.js 225.51 kB │ gzip: 70.77 kB
dist-client/assets/desktop-DWk52Vcz.js 281.80 kB │ gzip: 79.44 kB
✓ built in 1.13s
Mobile shell built: 3 allowed assets; API and project data are never cached.
Server bundle built.
@@ -0,0 +1,19 @@
PASS actual form saves Responses and declared stream, refresh preserves both and narrow editor remains operable
PASS separate text, nonce tool and actual SSE probes bind only their requested capability and own attempts
EVIDENCE D:\Ai-project\REA逆向\Agent-Workbench\.qa\model-protocols-ui-1c306e8c-d4a8-416a-a73d-d63d03edc7de
node:internal/modules/run_main:107
triggerUncaughtException(
^
locator.check: Timeout 30000ms exceeded.
Call log:
- waiting for getByRole('dialog', { name: '模型中心', exact: true }).getByRole('form', { name: '编辑模型池', exact: true }).locator('.mc-member-choice').filter({ hasText: '只支持文本' }).locator('input[type=checkbox]')
at D:\Ai-project\REA逆向\Agent-Workbench\scripts\verify-model-protocols-ui.mjs:89:151 {
log: [
" - waiting for getByRole('dialog', { name: '模型中心', exact: true }).getByRole('form', { name: '编辑模型池', exact: true }).locator('.mc-member-choice').filter({ hasText: '只支持文本' }).locator('input[type=checkbox]')"
],
name: 'TimeoutError'
}
Node.js v24.22.1
@@ -0,0 +1,20 @@
> local-agent-workbench@1.17.0 build
> tsc --noEmit && vite build && node scripts/build-mobile.mjs && node scripts/build-server.mjs
vite v7.3.7 building client environment for production...
transforming...
✓ 1725 modules transformed.
rendering chunks...
computing gzip size...
dist-client/.vite/manifest.json 0.55 kB │ gzip: 0.22 kB
dist-client/index.html 0.58 kB │ gzip: 0.40 kB
dist-client/mobile.html 0.82 kB │ gzip: 0.46 kB
dist-client/assets/mobile-BOUNcCmV.css 11.42 kB │ gzip: 3.04 kB
dist-client/assets/desktop-5MRuZeKa.css 135.46 kB │ gzip: 25.93 kB
dist-client/assets/mobile-Dz6Bo2B9.js 21.74 kB │ gzip: 7.75 kB
dist-client/assets/x-Cc3T3VuL.js 225.51 kB │ gzip: 70.77 kB
dist-client/assets/desktop-BSNXF-id.js 282.30 kB │ gzip: 79.56 kB
✓ built in 1.25s
Mobile shell built: 3 allowed assets; API and project data are never cached.
Server bundle built.
@@ -0,0 +1,56 @@
PASS actual form saves Responses and declared stream, refresh preserves both and narrow editor remains operable
PASS separate text, nonce tool and actual SSE probes bind only their requested capability and own attempts
PASS actual pooled task excludes text-only before dispatch, executes original tool once and downloads SHA-matched original SSE
EVIDENCE D:\Ai-project\REA逆向\Agent-Workbench\.qa\model-protocols-ui-f6eeae60-da72-43f4-90bf-de84b1c04468
node:internal/modules/run_main:107
triggerUncaughtException(
^
ExpectError: expect(locator).toContainText(expected) failed
Locator: getByLabel('模型输出片段', { exact: true }).last()
Expected substring: "正在核对当前项目"
Received string: "正在接收Responses · 0 字符片段仅供查看。完整响应核对前不执行工具;中断后保留原记录。"
Timeout: 5000ms
Call log:
- Expect "toContainText" getByLabel('模型输出片段', { exact: true }).last() with timeout 5000ms
- waiting for getByLabel('模型输出片段', { exact: true }).last()
9 × locator resolved to <div aria-label="模型输出片段" class="model-stream-preview">…</div>
- unexpected value "正在接收Responses · 0 字符片段仅供查看。完整响应核对前不执行工具;中断后保留原记录。"
at captureRawStack (D:\Ai-project\REA逆向\Agent-Workbench\node_modules\playwright-core\lib\coreBundle.js:9282:17)
at callMatcherAsStep (D:\Ai-project\REA逆向\Agent-Workbench\node_modules\playwright\lib\matchers\expect.js:13316:57)
at Object.toContainText (D:\Ai-project\REA逆向\Agent-Workbench\node_modules\playwright\lib\matchers\expect.js:13308:23)
at D:\Ai-project\REA逆向\Agent-Workbench\scripts\verify-model-protocols-ui.mjs:122:67 {
matcherResult: {
name: 'toContainText',
expected: '正在核对当前项目',
message: 'expect(locator).toContainText(expected) failed\n' +
'\n' +
"Locator: getByLabel('模型输出片段', { exact: true }).last()\n" +
'Expected substring: "正在核对当前项目"\n' +
'Received string: "正在接收Responses · 0 字符片段仅供查看。完整响应核对前不执行工具;中断后保留原记录。"\n' +
'Timeout: 5000ms\n' +
'\n' +
'Call log:\n' +
` - Expect "toContainText" getByLabel('模型输出片段', { exact: true }).last() with timeout 5000ms\n` +
" - waiting for getByLabel('模型输出片段', { exact: true }).last()\n" +
' 9 × locator resolved to <div aria-label="模型输出片段" class="model-stream-preview">…</div>\n' +
' - unexpected value "正在接收Responses · 0 字符片段仅供查看。完整响应核对前不执行工具;中断后保留原记录。"\n',
pass: false,
actual: '正在接收Responses · 0 字符片段仅供查看。完整响应核对前不执行工具;中断后保留原记录。',
log: [
` - Expect "toContainText" getByLabel('模型输出片段', { exact: true }).last() with timeout 5000ms`,
" - waiting for getByLabel('模型输出片段', { exact: true }).last()",
' 9 × locator resolved to <div aria-label="模型输出片段" class="model-stream-preview">…</div>',
' - unexpected value "正在接收Responses · 0 字符片段仅供查看。完整响应核对前不执行工具;中断后保留原记录。"'
],
timeout: 5000,
ariaSnapshot: '- strong: 正在接收\n' +
'- text: Responses · 0 字符\n' +
'- paragraph: 片段仅供查看。完整响应核对前不执行工具;中断后保留原记录。'
}
}
Node.js v24.22.1
@@ -0,0 +1,259 @@
UI regression: verify-model-protocols-ui.mjs
PASS verify-model-protocols-ui.mjs: 5 groups
UI regression: verify-companion-ui.mjs
PASS verify-companion-ui.mjs: 7 groups
UI regression: verify-controlled-ui.mjs
PASS verify-controlled-ui.mjs: 9 groups
UI regression: verify-purge-ui.mjs
PASS verify-purge-ui.mjs: 9 groups
UI regression: verify-workers-ui.mjs
PASS verify-workers-ui.mjs: 10 groups
UI regression: verify-task-plans-ui.mjs
PASS verify-task-plans-ui.mjs: 8 groups
UI regression: verify-browser-ui.mjs
PASS verify-browser-ui.mjs: 10 groups
UI regression: verify-ui.mjs
PASS verify-ui.mjs: 10 groups
UI regression: verify-diagnostics-ui.mjs
PASS verify-diagnostics-ui.mjs: 9 groups
UI regression: verify-execution-ui.mjs
PASS verify-execution-ui.mjs: 7 groups
UI regression: verify-delivery-ui.mjs
PASS verify-delivery-ui.mjs: 9 groups
UI regression: verify-index-ui.mjs
PASS verify-index-ui.mjs: 8 groups
UI regression: verify-change-sets-ui.mjs
PASS verify-change-sets-ui.mjs: 11 groups
UI regression: verify-context-ui.mjs
PASS verify-context-ui.mjs: 11 groups
UI regression: verify-workspaces-ui.mjs
PASS verify-workspaces-ui.mjs: 9 groups
UI regression: verify-python-ui.mjs
PASS verify-python-ui.mjs: 7 groups
UI regression: verify-react-delivery.mjs
PASS verify-react-delivery.mjs: 6 groups
UI regression: verify-research-ui.mjs
PASS verify-research-ui.mjs: 11 groups
UI regression: verify-engines-ui.mjs
PASS verify-engines-ui.mjs: 10 groups
UI regression: verify-model-center-ui.mjs
PASS verify-model-center-ui.mjs: 6 groups
UI regression: verify-project-memory-ui.mjs
PASS verify-project-memory-ui.mjs: 4 groups
UI regression: verify-task-checkpoints-ui.mjs
PASS verify-task-checkpoints-ui.mjs: 3 groups
UI regression: verify-skills-ui.mjs
PASS verify-skills-ui.mjs: 4 groups
UI regression: verify-mcp-ui.mjs
PASS verify-mcp-ui.mjs: 6 groups
UI regression: verify-mcp-recovery-ui.mjs
PASS verify-mcp-recovery-ui.mjs: 2 groups
{
"success": true,
"version": "1.17.0",
"at": "2026-10-31T14:54:40.155Z",
"groups": 191,
"results": [
{
"script": "verify-model-protocols-ui.mjs",
"success": true,
"groups": 5,
"durationMs": 22412,
"evidence": "docs/model-protocols-model-protocols-ui-validation.json",
"log": ".qa/model-protocols-verify-model-protocols-ui.log"
},
{
"script": "verify-companion-ui.mjs",
"success": true,
"groups": 7,
"durationMs": 24040,
"evidence": "docs/model-protocols-companion-ui-validation.json",
"log": ".qa/model-protocols-verify-companion-ui.log"
},
{
"script": "verify-controlled-ui.mjs",
"success": true,
"groups": 9,
"durationMs": 88501,
"evidence": "docs/model-protocols-controlled-final-ui-validation.json",
"log": ".qa/model-protocols-verify-controlled-ui.log"
},
{
"script": "verify-purge-ui.mjs",
"success": true,
"groups": 9,
"durationMs": 32355,
"evidence": "docs/model-protocols-purge-final-ui-validation.json",
"log": ".qa/model-protocols-verify-purge-ui.log"
},
{
"script": "verify-workers-ui.mjs",
"success": true,
"groups": 10,
"durationMs": 173264,
"evidence": "docs/model-protocols-workers-ui-validation.json",
"log": ".qa/model-protocols-verify-workers-ui.log"
},
{
"script": "verify-task-plans-ui.mjs",
"success": true,
"groups": 8,
"durationMs": 17386,
"evidence": "docs/model-protocols-task-ui-validation.json",
"log": ".qa/model-protocols-verify-task-plans-ui.log"
},
{
"script": "verify-browser-ui.mjs",
"success": true,
"groups": 10,
"durationMs": 27875,
"evidence": "docs/model-protocols-browser-ui-validation.json",
"log": ".qa/model-protocols-verify-browser-ui.log"
},
{
"script": "verify-ui.mjs",
"success": true,
"groups": 10,
"durationMs": 24543,
"evidence": "docs/model-protocols-basic-ui-validation.json",
"log": ".qa/model-protocols-verify-ui.log"
},
{
"script": "verify-diagnostics-ui.mjs",
"success": true,
"groups": 9,
"durationMs": 23892,
"evidence": "docs/model-protocols-diagnostics-ui-validation.json",
"log": ".qa/model-protocols-verify-diagnostics-ui.log"
},
{
"script": "verify-execution-ui.mjs",
"success": true,
"groups": 7,
"durationMs": 32631,
"evidence": "docs/model-protocols-execution-ui-validation.json",
"log": ".qa/model-protocols-verify-execution-ui.log"
},
{
"script": "verify-delivery-ui.mjs",
"success": true,
"groups": 9,
"durationMs": 56215,
"evidence": "docs/model-protocols-delivery-ui-validation.json",
"log": ".qa/model-protocols-verify-delivery-ui.log"
},
{
"script": "verify-index-ui.mjs",
"success": true,
"groups": 8,
"durationMs": 8665,
"evidence": "docs/model-protocols-index-ui-validation.json",
"log": ".qa/model-protocols-verify-index-ui.log"
},
{
"script": "verify-change-sets-ui.mjs",
"success": true,
"groups": 11,
"durationMs": 16725,
"evidence": "docs/model-protocols-change-sets-ui-validation.json",
"log": ".qa/model-protocols-verify-change-sets-ui.log"
},
{
"script": "verify-context-ui.mjs",
"success": true,
"groups": 11,
"durationMs": 25214,
"evidence": "docs/model-protocols-context-ui-validation.json",
"log": ".qa/model-protocols-verify-context-ui.log"
},
{
"script": "verify-workspaces-ui.mjs",
"success": true,
"groups": 9,
"durationMs": 30360,
"evidence": "docs/model-protocols-workspaces-ui-validation.json",
"log": ".qa/model-protocols-verify-workspaces-ui.log"
},
{
"script": "verify-python-ui.mjs",
"success": true,
"groups": 7,
"durationMs": 36071,
"evidence": "docs/model-protocols-python-ui-validation.json",
"log": ".qa/model-protocols-verify-python-ui.log"
},
{
"script": "verify-react-delivery.mjs",
"success": true,
"groups": 6,
"durationMs": 62073,
"evidence": "docs/model-protocols-react-delivery-validation.json",
"log": ".qa/model-protocols-verify-react-delivery.log"
},
{
"script": "verify-research-ui.mjs",
"success": true,
"groups": 11,
"durationMs": 39718,
"evidence": "docs/model-protocols-research-ui-validation.json",
"log": ".qa/model-protocols-verify-research-ui.log"
},
{
"script": "verify-engines-ui.mjs",
"success": true,
"groups": 10,
"durationMs": 18470,
"evidence": "docs/model-protocols-engines-ui-validation.json",
"log": ".qa/model-protocols-verify-engines-ui.log"
},
{
"script": "verify-model-center-ui.mjs",
"success": true,
"groups": 6,
"durationMs": 14851,
"evidence": "docs/model-protocols-model-center-ui-validation.json",
"log": ".qa/model-protocols-verify-model-center-ui.log"
},
{
"script": "verify-project-memory-ui.mjs",
"success": true,
"groups": 4,
"durationMs": 9629,
"evidence": "docs/model-protocols-project-memory-ui-validation.json",
"log": ".qa/model-protocols-verify-project-memory-ui.log"
},
{
"script": "verify-task-checkpoints-ui.mjs",
"success": true,
"groups": 3,
"durationMs": 10062,
"evidence": "docs/model-protocols-task-checkpoints-ui-validation.json",
"log": ".qa/model-protocols-verify-task-checkpoints-ui.log"
},
{
"script": "verify-skills-ui.mjs",
"success": true,
"groups": 4,
"durationMs": 9400,
"evidence": "docs/model-protocols-skills-ui-validation.json",
"log": ".qa/model-protocols-verify-skills-ui.log"
},
{
"script": "verify-mcp-ui.mjs",
"success": true,
"groups": 6,
"durationMs": 12386,
"evidence": "docs/model-protocols-mcp-ui-validation.json",
"log": ".qa/model-protocols-verify-mcp-ui.log"
},
{
"script": "verify-mcp-recovery-ui.mjs",
"success": true,
"groups": 2,
"durationMs": 7948,
"evidence": "docs/model-protocols-mcp-recovery-ui-validation.json",
"log": ".qa/model-protocols-verify-mcp-recovery-ui.log"
}
],
"scope": "Every verifier owns an independent port and synthetic data directory. Synthetic provider fixtures verify protocols; configured-model acceptance is separate. No production service or real key read/copied."
}
@@ -0,0 +1,69 @@
UI regression: verify-model-protocols-ui.mjs
PASS verify-model-protocols-ui.mjs: 5 groups
UI regression: verify-companion-ui.mjs
PASS verify-companion-ui.mjs: 7 groups
UI regression: verify-controlled-ui.mjs
PASS verify-controlled-ui.mjs: 9 groups
UI regression: verify-purge-ui.mjs
file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/verify-all-ui.mjs:76
child.once('close', code => { clearTimeout(timer); code === 0 && !timedOut ? resolve(tail) : reject(new Error(script + (timedOut ? ' timed out' : ' failed (' + code + ')') + ': ' + tail)); });
^
Error: verify-purge-ui.mjs failed (1): node:internal/modules/run_main:107
triggerUncaughtException(
^
AssertionError [ERR_ASSERTION]: Purge 来源不完整或无法核对:ENOENT: no such file or directory, lstat 'D:\Ai-project\REA逆向\dsh-purge-1.1.78-windows\package.json'
+ actual - expected
+ 'unavailable'
- 'available'
at file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/verify-purge-ui.mjs:21:68
at Array.map (<anonymous>)
at file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/verify-purge-ui.mjs:19:38
at ModuleJob.run (node:internal/modules/esm/module_job:561:25)
at async node:internal/modules/esm/loader:647:26
at async asyncRunEntryPointWithESMLoader (node:internal/modules/run_main:101:5) {
generatedMessage: false,
code: 'ERR_ASSERTION',
actual: 'unavailable',
expected: 'available',
operator: 'strictEqual',
diff: 'simple'
}
Node.js v24.22.1
at ChildProcess.<anonymous> (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/scripts/verify-all-ui.mjs:76:107)
at ChildProcess.wrapper (node:events:639:12)
at ChildProcess.emit (node:events:514:28)
at maybeClose (node:internal/child_process:1172:16)
at ChildProcess._handle.onexit (node:internal/child_process:335:5)
Node.js v24.22.1
@@ -0,0 +1,24 @@
{
"success": true,
"version": "1.17.0",
"testedAt": "2026-10-31T15:09:25.813228+00:00",
"archives": [
{
"kind": "portable",
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-win-x64-20261031150548103.zip",
"files": 13578,
"crcPassed": true,
"manifestAndDirectoryMatch": true,
"sha256": "1e13982788e516157c3a9a05500f1bd4c497dd9aa9701e519207677eaf01458d"
},
{
"kind": "source",
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.17.0-source-20261031150548103.zip",
"files": 13871,
"crcPassed": true,
"manifestAndDirectoryMatch": true,
"sha256": "20263325fa0a6c8479eff01b631dc56db84fec3e2ab5eacc154798085209f7ab"
}
],
"extractionRoot": "D:\\AWQA17\\source-check-65eb2b40faa24f9ca9165ce2743fc0ec\\Agent-Workbench-1.17.0-source-20261031150548103"
}
@@ -0,0 +1,51 @@
import fs from 'node:fs';
import path from 'node:path';
import assert from 'node:assert/strict';
import { spawn } from 'node:child_process';
import { createHash } from 'node:crypto';
import { projectEnvironment } from '../src/core/process-environment.mjs';
const hash = bytes => createHash('sha256').update(bytes).digest('hex');
const phase = process.argv[2];
const python = 'D:/Ai-project/REA逆向/Agent-Workbench/.qa/dev-task-plans/.qa/python-3.10/Scripts/python.exe';
const release = JSON.parse(fs.readFileSync('releases/latest.json', 'utf8'));
const choices = {
archives: { executable: python, args: ['scripts/verify-archives.py', '--extract-source'], pointer: '.qa/archives-validation.json' },
source: { args: ['scripts/verify-source-package.mjs', '--native-engines'], pointer: '.qa/source-package-validation.json' },
'source-retry': { args: ['scripts/verify-source-package.mjs', '--native-engines'], pointer: '.qa/source-package-validation.json' },
portable: { args: ['scripts/verify-release.mjs', '--native-engines'], pointer: '.qa/release-validation.json' },
'portable-companion': { args: ['scripts/verify-companion-ui.mjs', '--app-root', release.portableRoot] },
'portable-protocols': { args: ['scripts/verify-model-protocols-ui.mjs', '--app-root', release.portableRoot] },
'prepare-install': { args: ['.qa/prepare-protocol-installer.mjs'] },
install: { args: ['.qa/verify-protocol-install.mjs'] },
'formal-stage': { args: ['.qa/deploy-protocol-formal.mjs', 'stage'] },
'formal-upgrade': { args: ['.qa/deploy-protocol-formal.mjs', 'upgrade'] },
'formal-activate': { args: ['.qa/deploy-protocol-formal.mjs', 'activate'] },
'formal-verify': { args: ['.qa/deploy-protocol-formal.mjs', 'verify'] },
'formal-ui': { args: ['.qa/verify-protocol-formal-ui.mjs'] },
'formal-artifacts': { args: ['.qa/verify-protocol-formal-artifacts.mjs', '.qa/protocol-f-formal-verify.json', '.qa/protocol-f-formal-artifacts.json'] },
seal: { args: ['.qa/seal-protocol-release.mjs'] },
};
const choice = choices[phase]; assert.ok(choice, 'Select a defined isolated acceptance phase.');
const prefix = '.qa/protocol-f-' + phase, logPath = prefix + '.log', receiptPath = prefix + '-result.json';
for (const file of [logPath, receiptPath]) assert.equal(fs.existsSync(file), false, 'Keep existing evidence: ' + file);
const baseline = JSON.parse(fs.readFileSync('docs/model-protocols-release-code-baseline.json', 'utf8'));
function unchanged() { for (const item of [...baseline.inputs, ...baseline.builds, ...baseline.benchmarks]) { const bytes = fs.readFileSync(item.path); assert.equal(bytes.length, item.bytes, item.path); assert.equal(hash(bytes), item.sha256, item.path); } }
unchanged();
const startedAt = new Date().toISOString(), log = fs.createWriteStream(logPath, { flags: 'wx' });
const env = projectEnvironment({ PATH: path.dirname(process.execPath) + path.delimiter + process.env.PATH, AGENT_WORKBENCH_TEST_PYTHON: python,
AGENT_WORKBENCH_ARCHIVE_QA_ROOT: 'D:/AWQA17',
AGENT_WORKBENCH_DEEPSEEK_ROOT: 'D:/Ai-project/DeepSeek-Harness-Desktop/deepseek-harness',
AGENT_WORKBENCH_PURGE_ROOT: 'D:/Ai-project/DeepSeek-Harness-Desktop/plugins/dsh-purge' });
const child = spawn(choice.executable || process.execPath, choice.args, { cwd: process.cwd(), env, windowsHide: true, stdio: ['ignore', 'pipe', 'pipe'] });
child.stdout.on('data', bytes => log.write(bytes)); child.stderr.on('data', bytes => log.write(bytes));
const exitCode = await new Promise((resolve, reject) => { child.once('error', reject); child.once('close', resolve); });
await new Promise(resolve => log.end(resolve));
let pointer = null;
if (choice.pointer && fs.existsSync(choice.pointer)) { const bytes = fs.readFileSync(choice.pointer); pointer = { path: choice.pointer, sha256: hash(bytes) }; fs.writeFileSync(prefix + '-original-validation.json', bytes, { flag: 'wx' }); }
let unchangedInputs = true, error;
try { unchanged(); } catch (failure) { unchangedInputs = false; error = failure.stack; }
const receipt = { success: exitCode === 0 && unchangedInputs, phase, startedAt, finishedAt: new Date().toISOString(), exitCode, logPath,
logBytes: fs.statSync(logPath).size, logSha256: hash(fs.readFileSync(logPath)), pointer, unchangedInputs,
codeCommit: baseline.codeCommit, release: { portableRoot: release.portableRoot, sourceRoot: release.sourceRoot, setupSha256: release.installer.sha256 }, ...(error ? { error } : {}) };
fs.writeFileSync(receiptPath, JSON.stringify(receipt, null, 2) + '\n', { flag: 'wx' });
console.log(JSON.stringify(receipt)); if (!receipt.success) process.exitCode = 1;
@@ -0,0 +1,24 @@
✔ saved engine budgets verify all Chat/Responses route variants and reject a mismatched output allowance (592.287ms)
✔ owned CC Switch SQLite imports the selected TOML Responses wire_api and leaves unknown clients untouched (710.0598ms)
✔ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (1527.7255ms)
✔ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (1797.027ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"body":{"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial" (994.4985ms)
✔ Reject {"body":{"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","cal (1145.3472ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","cal (1076.7906ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supporte (1121.9989ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (4339.4911ms)
▶ invalid JSON in a nonstream Chat or Responses tool group prevents every tool and pool replay
✔ chat-completions (1174.9831ms)
✔ responses (1170.1157ms)
✔ invalid JSON in a nonstream Chat or Responses tool group prevents every tool and pool replay (2345.4991ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.2085ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (678.5131ms)
ℹ tests 14
ℹ suites 0
ℹ pass 14
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 12110.7319
@@ -0,0 +1,256 @@
{
"success": true,
"at": "2026-10-31T15:08:32.114Z",
"codeCommit": "a72dc0fbc1f3430f46b12ba0b1dbbe2c457056d8",
"inputFiles": 387,
"exactInputs": 365,
"normalized": [
{
"path": "scripts/package.mjs",
"bytes": 11486,
"sha256": "a083e2b6058c35641de30ceee3f8d4226afccbd7569f7d9cbd9d3ca216eadbfe",
"gitBytes": 11349,
"gitSha256": "bb34f3de1c67ce6411d7ff9fdbe304da1b2f269a337957b98d6a407b94b38a2d",
"attributes": [
"scripts/package.mjs: text: auto",
"scripts/package.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/qa-evidence.mjs",
"bytes": 843,
"sha256": "f22edec525e9658325857033b65b22a509e583b1a6c9fe4bc1ba378e738898e7",
"gitBytes": 826,
"gitSha256": "0a168d76a2c6242b8f42509b1f609f845449e924dd6e2155f2df3e8800bfdc36",
"attributes": [
"scripts/qa-evidence.mjs: text: auto",
"scripts/qa-evidence.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/upgrade-running.mjs",
"bytes": 13535,
"sha256": "ef92a10b5b51e3c01626444cb6e70c45955bf89c0dbc76c21c0d591de5ebfe72",
"gitBytes": 13369,
"gitSha256": "5c38b30d43ae936b0f825baaeef1fd59e4670c8ff1086768332e83d0bb6fcc20",
"attributes": [
"scripts/upgrade-running.mjs: text: auto",
"scripts/upgrade-running.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/verify-all-ui.mjs",
"bytes": 10330,
"sha256": "850a50677eee7606c6e93f9a5f3ca8bcea15bd9ece42ec3f03e1018d6e4e1be5",
"gitBytes": 10224,
"gitSha256": "12bdadf3040f34df7bd164f4c1ecd80c6188e86ee0609388e032128404a71252",
"attributes": [
"scripts/verify-all-ui.mjs: text: auto",
"scripts/verify-all-ui.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/verify-change-sets-ui.mjs",
"bytes": 16307,
"sha256": "2e1187971fd642e19123bf435727087944561a23d0a7d20d95f73614ce40b31c",
"gitBytes": 16166,
"gitSha256": "bb04dfc9498e95bba0a0333faeff66a410739b7b079f1bac33217b2dcda8f9d4",
"attributes": [
"scripts/verify-change-sets-ui.mjs: text: auto",
"scripts/verify-change-sets-ui.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/verify-context-ui.mjs",
"bytes": 18691,
"sha256": "b44da14d62ceb9c9da599c37dc08addebe251e1febb2d40f129060a1e14cc2c4",
"gitBytes": 18495,
"gitSha256": "596d4c00c08490e825b4732fab5312ac17d12336e27cc83a68cf55ce891af60d",
"attributes": [
"scripts/verify-context-ui.mjs: text: auto",
"scripts/verify-context-ui.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/verify-controlled-release.mjs",
"bytes": 21832,
"sha256": "a364de8b21ddcb210560434af8c90efaee7592c13a2ae2b83e50ae72bc2a2bd8",
"gitBytes": 21633,
"gitSha256": "c12791d6429cb798e527b60cfb3354c2bf94cf9f6158bd252399d04521833bcc",
"attributes": [
"scripts/verify-controlled-release.mjs: text: auto",
"scripts/verify-controlled-release.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/verify-release.mjs",
"bytes": 17494,
"sha256": "7ab21c22bbcc52b37334e1bf18c7a2e786af6ad87708fcdd8e4541ab6480b99f",
"gitBytes": 17312,
"gitSha256": "9501a4c7f88da3f43e179b7bd43447bfe7141c9de8a7b718be6f9655044b364c",
"attributes": [
"scripts/verify-release.mjs: text: auto",
"scripts/verify-release.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/verify-source-package.mjs",
"bytes": 16523,
"sha256": "f6a4fe1bf440d386c463d6b4dfc285b94dfd6f2b4657b7414c4bcc80b302a647",
"gitBytes": 16347,
"gitSha256": "6b029198efdffe133a9e70a1e7b39e0e0e90569e4d062b0b16af25af5a95ad75",
"attributes": [
"scripts/verify-source-package.mjs: text: auto",
"scripts/verify-source-package.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/verify-ui.mjs",
"bytes": 18880,
"sha256": "d9e7b3e8eb0ff80168acfd5d320685db6d4e1f53827cac4b96cffd5e54c758b6",
"gitBytes": 18623,
"gitSha256": "cd57dd67aada66f667a4e62b7a8f76a79d3c09acd7ba7c658b89e6d38dc6be9f",
"attributes": [
"scripts/verify-ui.mjs: text: auto",
"scripts/verify-ui.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "scripts/verify-workers-ui.mjs",
"bytes": 18945,
"sha256": "a65ff088b3429f2f9e8d46198f24eae65862d1591461f7da100911303ed4993a",
"gitBytes": 18798,
"gitSha256": "da15765a3c7f2e01ee3ecb0b6393cf2757392dd12546393d65e1776f9fc18b18",
"attributes": [
"scripts/verify-workers-ui.mjs: text: auto",
"scripts/verify-workers-ui.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "src/client/ModelCenter.css",
"bytes": 10343,
"sha256": "13a8be33b8806275830eabc1f1c1a8bff0335850df7f0b1c8b28b1e5e7beb8a5",
"gitBytes": 10342,
"gitSha256": "4e7737b23113cbbdab2a94082e3644942210ef9e6bd5b957f9384d4b48187244",
"attributes": [
"src/client/ModelCenter.css: text: auto",
"src/client/ModelCenter.css: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "src/client/SkillCenter.tsx",
"bytes": 9303,
"sha256": "b227ffdccf1d90ac969f1e4e2a56c592b7c9eac13746bf0c90ee9d9e702e9f10",
"gitBytes": 9240,
"gitSha256": "de7a4426aef468a73a9522da9883592ab0cc29a1f2b3786b764cd51a304347b2",
"attributes": [
"src/client/SkillCenter.tsx: text: auto",
"src/client/SkillCenter.tsx: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "src/client/mobile/mobile.css",
"bytes": 11473,
"sha256": "a9968352cb6c7429f23fe8ec0c72119cdcec7bbd502631315b7b2125bfe38175",
"gitBytes": 11465,
"gitSha256": "41747da0d7e9f813639b364182ac66b58ccaf3cba71696077a138f61b9a68691",
"attributes": [
"src/client/mobile/mobile.css: text: auto",
"src/client/mobile/mobile.css: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "src/client/styles.css",
"bytes": 40574,
"sha256": "f982b39cdc159ac45ad70e1cfc0df7e46ee11502a8613bc5cc70babf92ea2325",
"gitBytes": 40545,
"gitSha256": "44a12c0ce323a0e82da55056043ca70b2af78c7a343fe4668380eabf11502c9d",
"attributes": [
"src/client/styles.css: text: auto",
"src/client/styles.css: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "src/server/diagnostics.mjs",
"bytes": 8093,
"sha256": "c8664c81473690ae5800674a3f68694c815435880653c7a1ebba57ca39bf5ee4",
"gitBytes": 7965,
"gitSha256": "020d3e949602471e76bfbe849078309c146bf537a3131ee54024fa9396e7d7d3",
"attributes": [
"src/server/diagnostics.mjs: text: auto",
"src/server/diagnostics.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
},
{
"path": "tests/skills.test.mjs",
"bytes": 17628,
"sha256": "c7228c190337bb5f2b9ab5a85ad34b78fd03541854e83197d7c4d2506bb3c6c6",
"gitBytes": 17438,
"gitSha256": "977e987ccdeab5bf07fd3dd1788d8ba866a3dc3715c9dd947982a9abfbab7e06",
"attributes": [
"tests/skills.test.mjs: text: auto",
"tests/skills.test.mjs: eol: lf"
],
"scope": "Only repository-declared CRLF/LF normalization; full normalized buffer matches committed blob."
}
],
"generated": [
{
"path": "examples/delivery-report/dist/cli.mjs",
"bytes": 818,
"sha256": "b8223dd1c0a98b3b7a0cd716db0ae053a2ca17fbd8272b005adcedd3ce5f7a69",
"source": "examples/delivery-report/src/cli.mjs",
"generator": "examples/delivery-report/scripts/build.mjs",
"generatorSha256": "9dbc4bea4940fb0e87e7924a4601166c422149d3e9f5ecf1d22efde3b6c8c302"
},
{
"path": "examples/delivery-report/dist/normalize.mjs",
"bytes": 1254,
"sha256": "57eb47b09913b131982a14c276ba5292aba4c94e0f475c97955221f6a7b7cdcd",
"source": "examples/delivery-report/src/normalize.mjs",
"generator": "examples/delivery-report/scripts/build.mjs",
"generatorSha256": "9dbc4bea4940fb0e87e7924a4601166c422149d3e9f5ecf1d22efde3b6c8c302"
},
{
"path": "examples/delivery-report/dist/package.json",
"bytes": 18,
"sha256": "1239d4d885dcad42201a27ed9324f8f0f760b78700d8db9ced39a511cffe7eae",
"source": null,
"generator": "examples/delivery-report/scripts/build.mjs",
"generatorSha256": "9dbc4bea4940fb0e87e7924a4601166c422149d3e9f5ecf1d22efde3b6c8c302"
},
{
"path": "examples/delivery-report/dist/render.mjs",
"bytes": 375,
"sha256": "d449524d731a55ef51d0143f38ee8b73db56363c78babc385fa58df7f2a542f0",
"source": "examples/delivery-report/src/render.mjs",
"generator": "examples/delivery-report/scripts/build.mjs",
"generatorSha256": "9dbc4bea4940fb0e87e7924a4601166c422149d3e9f5ecf1d22efde3b6c8c302"
},
{
"path": "examples/delivery-report/dist/summary.mjs",
"bytes": 427,
"sha256": "bf7ada9e5dbca167df2b9a5306c1d9399b4a658fb0777c73812856687ad0c5ef",
"source": "examples/delivery-report/src/summary.mjs",
"generator": "examples/delivery-report/scripts/build.mjs",
"generatorSha256": "9dbc4bea4940fb0e87e7924a4601166c422149d3e9f5ecf1d22efde3b6c8c302"
}
],
"scope": "Every frozen source/QA/build input tied to the committed source or exact generated example derivation; unchanged build bytes independently recorded."
}
@@ -0,0 +1,46 @@
✔ native text and call IDs preserve original result pairing while unsupported content fails closed (1.1259ms)
✔ one engine child freezes real installation identity before creating tasks and rejects a changed source (13098.4525ms)
✔ owned native tools return saved results, use real approvals/checks and account each actual request once across new engine Runs (40251.2521ms)
✔ forged native scope cannot change a parent file and a final completion claim cannot replace worker proof (20308.5425ms)
✔ duplicate native IDs are rejected without replaying a saved tool (16562.3545ms)
✔ native input exceeding the context budget is retained and sends no model request (15352.2075ms)
✔ stop retains a late reported model usage once and executes no tool from that output (18454.3389ms)
✔ a source change after the original model intent stops before native tools execute (11914.2429ms)
✔ an engine source change while approval is pending invalidates the old permission before any check starts (20706.4131ms)
✔ missing or noncompleted turn end cannot be adopted even with an owned receipt and assistant text (0.6475ms)
✔ closed root with an original live descendant remains unknown and retains its process tree (0.2834ms)
✔ host recovery reserves original pending attempts exactly once and explicit review preserves unknown outcomes (19007.1945ms)
✔ corrupted model reservation cannot reduce recovered budget or permit a new engine request (16877.8397ms)
✔ owned CC Switch SQLite imports the selected TOML Responses wire_api and leaves unknown clients untouched (3234.3774ms)
✔ Responses executes one original tool through Runtime, carries encrypted Items, reports raw usage and seals exact wire bytes (6488.5999ms)
✔ mixed-protocol pool keeps both actual variants and changes route on 429 without re-executing completed tools (4370.0093ms)
▶ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool
✔ Reject {"body":{"id":"resp_owned","status":"incomplete","output":[{"type":"function_call","id":"fc_partial" (3630.0169ms)
✔ Reject {"body":{"id":"resp_owned","status":"failed","output":[{"type":"function_call","id":"fc_failed","cal (2948.9085ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"function_call","id":"fc_bad","cal (3278.7186ms)
✔ Reject {"body":{"id":"resp_owned","status":"completed","output":[{"type":"computer_call","id":"not-supporte (3321.5086ms)
✔ Responses incomplete, failed, missing call identity and unsupported hosted outputs never dispatch a tool (13180.1108ms)
▶ invalid JSON in a nonstream Chat or Responses tool group prevents every tool and pool replay
✔ chat-completions (3471.3136ms)
✔ responses (2630.7028ms)
✔ invalid JSON in a nonstream Chat or Responses tool group prevents every tool and pool replay (6102.4523ms)
✔ wire adapter preserves tool IDs but never passes Responses-private Items to Chat or another endpoint (0.187ms)
✔ connection JSON import accepts explicit Responses protocol and continues to reject unknown protocols without reading client sessions (1186.2766ms)
✔ a stop from the event loop interrupts immediate model replies before later tools (3385.4454ms)
✔ a stop after the first tool preserves a complete interrupted multi-call protocol group (4073.0285ms)
✔ the compatible HTTP loop preserves call IDs, validates errors and waits before executing commands (10411.9691ms)
✔ denial prevents command execution and produces a paired denied result (5347.8428ms)
✔ stopping during approval interrupts pending calls without executing or replaying them (3917.6167ms)
✔ stopping a running fixture command terminates its own process tree and retains uncertain state (3065.8461ms)
✔ command output decodes split UTF-8 and excludes the workbench key from child environment (1458.0358ms)
✔ model-list testing uses a tentative key without changing the saved connection and redacts service errors (6.6736ms)
✔ truncated or repeated model calls fail before execution; tool input validation rejects unknown capabilities (4921.9888ms)
✔ request projection keeps exact tool pairs and complete recent user instructions while preserving disk history (2627.5789ms)
ℹ tests 36
ℹ suites 0
ℹ pass 36
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 192697.2863
@@ -0,0 +1,20 @@
> local-agent-workbench@1.17.0 build
> tsc --noEmit && vite build && node scripts/build-mobile.mjs && node scripts/build-server.mjs
vite v7.3.7 building client environment for production...
transforming...
✓ 1725 modules transformed.
rendering chunks...
computing gzip size...
dist-client/.vite/manifest.json 0.55 kB │ gzip: 0.22 kB
dist-client/index.html 0.58 kB │ gzip: 0.40 kB
dist-client/mobile.html 0.82 kB │ gzip: 0.46 kB
dist-client/assets/mobile-BOUNcCmV.css 11.42 kB │ gzip: 3.04 kB
dist-client/assets/desktop-5MRuZeKa.css 135.46 kB │ gzip: 25.93 kB
dist-client/assets/mobile-Dz6Bo2B9.js 21.74 kB │ gzip: 7.75 kB
dist-client/assets/x-Cc3T3VuL.js 225.51 kB │ gzip: 70.77 kB
dist-client/assets/desktop-BSNXF-id.js 282.30 kB │ gzip: 79.56 kB
✓ built in 1.16s
Mobile shell built: 3 allowed assets; API and project data are never cached.
Server bundle built.
@@ -0,0 +1,18 @@
node:internal/modules/cjs/loader:1568
throw err;
^
Error: Cannot find module 'D:\Ai-project\REA逆向\Agent-Workbench\node_modules\npm\bin\npm-cli.js'
at Module._resolveFilename (node:internal/modules/cjs/loader:1564:15)
at wrapResolveFilename (node:internal/modules/cjs/loader:1118:27)
at defaultResolveImplForCJSLoading (node:internal/modules/cjs/loader:1142:10)
at resolveForCJSWithHooks (node:internal/modules/cjs/loader:1169:12)
at Module._load (node:internal/modules/cjs/loader:1341:5)
at wrapModuleLoad (node:internal/modules/cjs/loader:261:19)
at Module.executeUserEntryPoint [as runMain] (node:internal/modules/run_main:154:5)
at node:internal/main/run_main_module:33:47 {
code: 'MODULE_NOT_FOUND',
requireStack: []
}
Node.js v24.22.1
@@ -0,0 +1,28 @@
import fs from 'node:fs';
import path from 'node:path';
import assert from 'node:assert/strict';
import { execFileSync, spawnSync } from 'node:child_process';
import { createHash } from 'node:crypto';
const hash=bytes=>createHash('sha256').update(bytes).digest('hex');
const identity=JSON.parse(fs.readFileSync('docs/model-protocols-release-identity.json'));
assert.equal(identity.success,true);
const manifest=JSON.parse(fs.readFileSync(identity.evidence.manifest));assert.equal(manifest.files.length,identity.evidence.files);
let bytes=0,jsonFiles=0;
for(const row of manifest.files){
const value=fs.readFileSync(row.archived);assert.equal(value.length,row.bytes);assert.equal(hash(value),row.sha256,row.archived);bytes+=value.length;
const indexed=execFileSync('git',['show',':'+row.archived],{windowsHide:true,maxBuffer:64*1024*1024});assert.deepEqual(indexed,value,'Git index raw evidence bytes: '+row.archived);
if(row.archived.endsWith('.json')){JSON.parse(value);jsonFiles++;}
}
assert.equal(bytes,manifest.bytes);
const docs=['README.md','CHANGELOG.md','AGENTS.md','task_plan.md','findings.md','progress.md','docs/QA-1.17.md','docs/RETROSPECTIVE-1.17.md','docs/PLAN-1.17.md','docs/PLAN-1.18.md','docs/PRODUCT-PLAN.md','docs/CAPABILITIES.md','docs/MILESTONE-AUDIT.md','docs/NEXT-STEPS.md','docs/VALIDATION.md'];
let links=0;
for(const name of docs){const text=fs.readFileSync(name,'utf8');for(const match of text.matchAll(/\]\(([^)]+)\)/g)){
const target=match[1].replace(/^<|>$/g,'').split('#')[0];if(!target||/^(?:https?:|codex:|app:|data:|mailto:)/.test(target))continue;
assert.equal(fs.existsSync(path.resolve(path.dirname(name),decodeURIComponent(target))),true,name+': '+target);links++;
}}
const wrappers=fs.readdirSync('.qa').filter(n=>/^(?:protocol-.*|(?:prepare|freeze|package|run|shorten|deploy|verify|audit|seal|record|write|finalize)-protocol-.*)\.mjs$/.test(n));
for(const name of wrappers){const result=spawnSync(process.execPath,['--check','.qa/'+name],{windowsHide:true,encoding:'utf8'});assert.equal(result.status,0,name+': '+result.stderr);}
const baseline=JSON.parse(fs.readFileSync('docs/model-protocols-release-code-baseline.json'));
for(const row of [...baseline.inputs,...baseline.builds,...baseline.benchmarks]){const value=fs.readFileSync(row.path);assert.equal(value.length,row.bytes);assert.equal(hash(value),row.sha256,row.path);}
const result={success:true,version:'1.17.0',rawEvidenceFiles:manifest.files.length,rawEvidenceBytes:bytes,jsonFiles,docs:docs.length,localLinks:links,wrappers:wrappers.length,frozenInputs:baseline.inputs.length,builds:baseline.builds.length,benchmarks:baseline.benchmarks.length,checkedAt:new Date().toISOString()};
fs.writeFileSync('.qa/protocol-f-final-check.json',JSON.stringify(result,null,2)+'\n',{flag:'wx'});console.log(JSON.stringify(result));
@@ -0,0 +1,111 @@
import fs from 'node:fs';
import path from 'node:path';
import assert from 'node:assert/strict';
import { createHash } from 'node:crypto';
import { publicSnapshot, compareOriginalState } from './protocol-f-state.mjs';
const read=file=>JSON.parse(fs.readFileSync(file,'utf8'));
const hash=bytes=>createHash('sha256').update(bytes).digest('hex');
const baseline=read('docs/model-protocols-release-code-baseline.json'), release=read('releases/latest.json');
assert.equal(baseline.version,'1.17.0'); assert.equal(release.installer.version,'1.17.0');
const phases=['archives','source','portable','portable-companion','portable-protocols','prepare-install','install','formal-stage','formal-upgrade','formal-activate','formal-ui','formal-verify','formal-artifacts'];
const receipts=phases.map(phase=>{const file='.qa/protocol-f-'+phase+'-result.json',value=read(file);assert.equal(value.success,true,phase);assert.equal(value.unchangedInputs,true);assert.equal(value.release.setupSha256,release.installer.sha256);return {file,value};});
const source=read('.qa/protocol-f-source-original-validation.json'), portable=read('.qa/protocol-f-portable-original-validation.json');
assert.equal(source.success,true); assert.equal(source.tests.tests,558); assert.equal(source.tests.pass,558);
for(const key of ['fail','cancelled','skipped','todo'])assert.equal(source.tests[key],0);
const sorted=rows=>rows.slice().sort((a,b)=>a.path.localeCompare(b.path));
assert.deepEqual(sorted(source.inputs),sorted(baseline.inputs));
assert.deepEqual(sorted(source.buildFiles),sorted(baseline.builds.map(({path,sha256})=>({path,sha256}))));
assert.equal(source.actualDesktop.window.results.length,7); assert.equal(source.actualDesktop.ui.assertions.length,9); assert.equal(source.actualDesktop.projects.cases.length,16);
assert.deepEqual(source.actualDesktop.projects.results.map(x=>x.parentCheck.results[0].tests.passed),[8,9,10]);
assert.equal(source.actualCompanion.passed.length,7); assert.equal(source.actualProtocols.passed.length,5);
assert.equal(source.actualProtocolEngineering.success,true); assert.equal(source.actualProtocolEngineering.cases.length,13);
assert.deepEqual(source.actualProtocolEngineering.results.map(x=>x.parentCheck.results[0].tests.passed),[8,9,10]);
assert.equal(portable.appRoot,release.portableRoot); assert.equal(portable.actualControlled.success,true);
const ui=read('docs/model-protocols-all-ui-validation.json'); assert.equal(ui.success,true); assert.equal(ui.results.length,25); assert.equal(ui.groups,191);
const installLocation=read('.qa/protocol-f-install-location.json'), install=read(installLocation.result);
assert.equal(install.success,true); assert.equal(install.cases.length,7); assert.equal(install.installer.productSetup.sha256,release.installer.sha256);
assert.equal(install.installer.samePayloadManifestSha256,hash(fs.readFileSync(path.join(release.portableRoot,'MANIFEST.json'))));
const formal=read('.qa/protocol-f-formal-verify.json'), before=read('.qa/protocol-f-formal-before.json');
assert.equal(formal.success,true); assert.equal(formal.identity.pid,36552); assert.equal(formal.currentProviderPreserved,true); assert.equal(formal.inspectorClosed,true);
const current=await publicSnapshot('http://127.0.0.1:4318',formal.dataRoot),retained=compareOriginalState(before,current);
assert.equal(current.version,'1.17.0'); assert.equal(current.runtimeRoot,formal.candidate);
const smoke=read('.qa/protocol-f-formal-ui-validation.json'), artifacts=read('.qa/protocol-f-formal-artifacts.json'), native=read('.qa/protocol-f-formal-native.json');
assert.equal(smoke.success,true); assert.equal(smoke.results.length,20); assert.equal(artifacts.success,true); assert.equal(artifacts.checked,artifacts.total);
assert.equal(native.success,true); assert.equal(native.version,'1.17.0'); assert.equal(native.hostPid,36552);
for(const key of ['pageErrors','requestFailures','httpErrors','blockedWrites'])assert.deepEqual(smoke[key],[]);
for(const item of [...baseline.inputs,...baseline.builds,...baseline.benchmarks]){const bytes=fs.readFileSync(item.path);assert.equal(bytes.length,item.bytes);assert.equal(hash(bytes),item.sha256,item.path);}
for(const item of [release.portable,release.source,release.installer])assert.equal(hash(fs.readFileSync(item.path)),item.sha256);
const gitAudit=read('.qa/protocol-f-code-git-audit.json');assert.equal(gitAudit.success,true);assert.equal(gitAudit.codeCommit,baseline.codeCommit);assert.equal(gitAudit.inputFiles,baseline.inputs.length);
const license=read('.qa/protocol-f-license-audit.json');assert.equal(license.success,true);
const completed=read('.qa/protocol-f-log-completions/index.json');assert.equal(completed.success,true);
const directory='docs/model-protocols-f-release-evidence';assert.equal(fs.existsSync(directory),false);
const queue=[],seen=new Map();
function bytesCopy(original,bytes,expected={}){
const absolute=path.resolve(original),exists=seen.get(absolute);if(exists){if(expected.sha256)assert.equal(exists.sha256,expected.sha256);return exists;}
const sha256=hash(bytes);if(expected.sha256)assert.equal(sha256,expected.sha256,original);if(expected.bytes!==undefined)assert.equal(bytes.length,expected.bytes,original);
const name=String(queue.length+1).padStart(3,'0')+'-'+path.basename(original).replace(/\.log$/,'-log.txt');
const row={original:absolute,archived:directory+'/'+name,bytes:bytes.length,sha256};queue.push({row,bytes});seen.set(absolute,row);return row;
}
function copy(file,expected={}){
assert.doesNotMatch(file,/(?:\.sqlite(?:-(?:wal|shm))?|\.(?:db|key|pem|pfx|p12))$/i,'No private database or certificate material');
const stat=fs.lstatSync(file);assert.equal(stat.isSymbolicLink(),false);assert.equal(stat.isFile(),true);return bytesCopy(file,fs.readFileSync(file),expected);
}
function installerLog(file,sha256){
const bytes=fs.readFileSync(file);if(hash(bytes)===sha256)return copy(file,{sha256});
const item=completed.completions.find(x=>x.source===path.resolve(file));assert.ok(item);assert.equal(item.original.sha256,sha256);
copy(item.original.path,item.original);return copy(file,item.completed);
}
function surface(root){if(!root||!fs.existsSync(root))return;for(const e of fs.readdirSync(root,{withFileTypes:true}))if(e.isFile()&&/\.(?:json|png|jpg|html|txt|log|tap|mjs|cjs|sse)$/.test(e.name))copy(path.join(root,e.name));}
function screenshots(value){
if(!value||typeof value!=='object')return;if(Array.isArray(value)){for(const x of value)screenshots(x);return;}
if(typeof value.file==='string'&&/\.(png|jpg)$/.test(value.file))copy(value.file,value.sha256?{sha256:value.sha256}:{});
for(const [key,x]of Object.entries(value)){
if(typeof x==='string'&&/screenshot$/i.test(key)&&/\.(png|jpg)$/.test(x))copy(x,value.screenshotSha256?{sha256:value.screenshotSha256}:{});
if(key==='screenshots'&&Array.isArray(x))for(const file of x)if(typeof file==='string')copy(file);
if(x&&typeof x==='object')screenshots(x);
}
}
function projectTree(root){if(!fs.existsSync(root))return;for(const e of fs.readdirSync(root,{withFileTypes:true})){if(['.git','node_modules','__pycache__','.pytest_cache','.venv'].includes(e.name))continue;const f=path.join(root,e.name);if(e.isDirectory())projectTree(f);else copy(f);}}
for(const {file,value}of receipts){copy(file);copy(value.logPath,{bytes:value.logBytes,sha256:value.logSha256});}
for(const name of fs.readdirSync('.qa'))if(/^(?:protocol-.*\.(?:json|log|tap|mjs)|(?:prepare|freeze|package|run|shorten|deploy|verify|audit|seal|record|write|finalize)-protocol-.*\.mjs)$/.test(name))copy(path.join('.qa',name));
for(const name of ['docs/model-protocols-release-code-baseline.json','docs/model-protocols-all-ui-validation.json','releases/latest.json'])copy(name);
for(const name of fs.readdirSync('docs'))if(/^model-protocols-.*\.(?:json|txt)$/.test(name))copy(path.join('docs',name));
for(const name of fs.readdirSync('.qa'))if(/^model-protocols-.*\.(?:json|log)$/.test(name))copy(path.join('.qa',name));
for(const row of source.logs)copy(row.path,row);copy(path.join(source.directory,'inputs.json'));copy(path.join(source.directory,'result.json'));
for(const row of source.logs.filter(x=>['contained companion gateway and UI','contained model protocols and mobile streams'].includes(x.stage))){const root=fs.readFileSync(row.path,'utf8').match(/EVIDENCE ([^\r\n]+)/)?.[1];assert.ok(root);surface(root);}
for(const row of install.cases){if(row.processLog)copy(row.processLog,{sha256:row.processLogSha256});if(row.setupLog)installerLog(row.setupLog,row.setupLogSha256);if(row.uninstallLog)installerLog(row.uninstallLog,row.uninstallLogSha256);}
copy(installLocation.result);copy(release.installer.source);copy(path.join(path.dirname(release.installer.source),'compile.txt'));copy(install.installer.source);copy(path.join(path.dirname(install.installer.source),'compile.txt'));
copy('.qa/protocol-f-formal-setup.log',{sha256:read('.qa/protocol-f-formal-stage.json').setupLogSha256});
surface('.qa/protocol-native-before');surface('.qa/protocol-native-after');surface('.qa/protocol-first-ui-environment-failure');surface('.qa/protocol-f-log-completions');
for(const name of fs.readdirSync('.qa'))if(name.startsWith('model-protocols-ui-'))surface(path.join('.qa',name));
surface(source.actualDesktop.window.root);projectTree(path.join(source.actualDesktop.window.root,'owned-project'));
for(const name of ['desktop-controlled-ui','desktop-projects','responses-engineering']){
const file=path.join(source.sourceRoot,'.qa',name+'-location.json'),location=read(file);copy(file);copy(location.result);surface(location.directory);
if(name==='desktop-controlled-ui')projectTree(path.join(location.directory,'original-fixture','project'));
else for(const kind of ['node','react','python']){const root=path.join(location.directory,kind);surface(root);projectTree(path.join(root,'original-fixture','project'));}
}
for(const name of ['.qa/controlled-projects-cf832704-bb40-4563-b086-23b93ec6aeb8','.qa/controlled-projects-c0121328-f10f-4c33-8960-e4c1db7df1f7']){
surface(name);for(const kind of ['node','react','python']){surface(path.join(name,kind));projectTree(path.join(name,kind,'original-fixture','project'));}
}
for(const value of [source.actualDesktop,source.actualCompanion,source.actualProtocols,smoke,native])screenshots(value);
for(const phase of ['source','portable-companion','portable-protocols']){
const log=phase==='source'?null:fs.readFileSync('.qa/protocol-f-'+phase+'.log','utf8');
if(log){const root=log.match(/EVIDENCE ([^\r\n]+)/)?.[1];assert.ok(root);surface(root);}
}
for(const value of [source.actualCompanion,source.actualProtocols])surface(value.directory);
const manifest={success:true,version:'1.17.0',codeCommit:baseline.codeCommit,sealedAt:new Date().toISOString(),files:queue.map(x=>x.row),
bytes:queue.reduce((sum,x)=>sum+x.bytes.length,0),sourceTests:source.tests,ui:{classes:25,groups:191},scope:'Original bytes of release and original failure evidence. Formal snapshots are public hashes only; no private SQLite, model Key, certificate, or QA device identity is sealed.'};
fs.mkdirSync(directory);fs.writeFileSync(directory+'/.gitattributes','* -text -whitespace\n',{flag:'wx'});
for(const {row,bytes}of queue)fs.writeFileSync(row.archived,bytes,{flag:'wx'});
fs.writeFileSync(directory+'/manifest.json',JSON.stringify(manifest,null,2)+'\n',{flag:'wx'});
const identity={success:true,version:'1.17.0',codeCommit:baseline.codeCommit,immutableRelease:path.basename(release.portableRoot),release,
frozen:{inputs:baseline.inputs.length,builds:baseline.builds.length,benchmarks:baseline.benchmarks.length,gitAudit},
independentSource:{sourceRoot:source.sourceRoot,tests:source.tests,inputs:source.inputFiles,buildMatchedFiles:source.buildMatchedFiles,
desktopWindow:7,desktopUi:9,desktopEngineering:16,responsesEngineering:13,parentTests:[8,9,10],containedCompanion:7,containedProtocols:5},
ui:{classes:25,groups:191,formalPanels:20},samePayloadInstall:{cases:7,setupSha256:install.installer.sha256,productSetupSha256:release.installer.sha256},
formal:{pid:formal.identity.pid,runtimeRoot:formal.candidate,dataRoot:formal.dataRoot,counts:formal.counts,artifactRecords:artifacts.artifactRecords,blobFiles:artifacts.blobFiles,
currentProviderPreserved:true,retained,inspectorClosed:true,companion:formal.companion,native},
evidence:{manifest:directory+'/manifest.json',files:queue.length,bytes:manifest.bytes},realModelRequests:0,
limitations:['Real configured-model engineering and Purge benefit remain separately deferred.','No new actual iPhone 1.17 stream acceptance; paired Edge/TLS viewports are separate from 1.16 user-confirmed iPhone evidence.','Vision/audio are capability declarations only.','Unknown WorkBuddy formats and model-independent computer control remain planned.'],sealedAt:manifest.sealedAt};
fs.writeFileSync('docs/model-protocols-release-identity.json',JSON.stringify(identity,null,2)+'\n',{flag:'wx'});
console.log(JSON.stringify({success:true,files:queue.length,bytes:manifest.bytes,identity:'docs/model-protocols-release-identity.json'}));
@@ -0,0 +1,101 @@
{
"success": true,
"phase": "activated",
"version": "1.17.0",
"activated": {
"phase": "activated",
"active": {
"directory": "Agent-Workbench-1.17.0-win-x64-20261031150548103",
"version": "1.17.0",
"manifestSha256": "79dc4ef4023ed5596f24e32ac9de8153a9f7d42f90e74917514c1dc2e19fffe7",
"runtimeIdentity": "a37c85e60d13624e8929586aaab798da5e8df63a4c68a7bf79eb2987304e0d80"
},
"receiptFile": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\.installation\\operation-a2ea60ed-9783-4c63-b027-810b3290f2ac.json",
"hostRetained": false,
"dataPreserved": true
},
"identity": {
"schema": 1,
"protocol": "agent-workbench-host-v1",
"app": "agent-workbench",
"version": "1.17.0",
"instanceId": "4c910dd5-82cb-4c8a-92f6-328338ea0662",
"pid": 36552,
"createdAt": "2026-10-31T06:30:34.359Z",
"userSid": "S-1-5-21-3906741337-925321518-3767092861-1001",
"ownerId": "b3ae90d3-9aa1-4b40-bd96-a53f01dbb502",
"executable": {
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.14.0-win-x64-20261030235440461\\runtime\\node.exe",
"bytes": 93915464,
"sha256": "d702daa815eabbff763c7d3808e235cf031cf2ca70211b56847bc72f51e240e5"
},
"appRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.14.0-win-x64-20261030235440461",
"runtimeRoot": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\versions\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"runtimeIdentity": "a37c85e60d13624e8929586aaab798da5e8df63a4c68a7bf79eb2987304e0d80",
"dataRoot": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench",
"port": 4318
},
"pointer": {
"schema": 1,
"app": "agent-workbench-installation",
"installId": "74c11bdd-4895-459b-8403-097f79980b41",
"revision": 4,
"dataRoot": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench",
"port": 4318,
"active": {
"directory": "Agent-Workbench-1.17.0-win-x64-20261031150548103",
"version": "1.17.0",
"manifestSha256": "79dc4ef4023ed5596f24e32ac9de8153a9f7d42f90e74917514c1dc2e19fffe7",
"runtimeIdentity": "a37c85e60d13624e8929586aaab798da5e8df63a4c68a7bf79eb2987304e0d80"
},
"previous": {
"directory": "Agent-Workbench-1.16.0-win-x64-20261031114604110",
"version": "1.16.0",
"manifestSha256": "5ee8b7f83e4d42ac8499c684d2a38b64ccce1b7d5936aca0cb455a9e23688629",
"runtimeIdentity": "6a81e7fec198be881e2110a14004cb1ab32a916a784391c28dcf6d9b6ee86270"
},
"activatedAt": "2026-10-31T16:07:35.060Z"
},
"retained": {
"originalRows": "exact",
"originalExports": "exact",
"originalArtifactBytes": "exact",
"originalModelCenter": "exact-with-validated-1.17-read-defaults",
"schemaDefaults": [
"streaming:false",
"capabilitySchema:1",
"explicit declarations",
"text/streaming capability state"
],
"originalStoredRows": "exact-no-migration"
},
"companion": {
"enabled": false,
"devices": 0,
"certificate": null
},
"currentProviderPreserved": true,
"counts": {
"projects": 17,
"sessions": 30,
"records": 2590,
"events": 4653,
"artifacts": 1366
},
"inspectorClosed": true,
"startedAt": "2026-10-31T15:57:34.616Z",
"finishedAt": "2026-10-31T16:07:40.489Z",
"installRoot": "C:/Users/wangy/AppData/Local/Programs/AgentWorkbench",
"dataRoot": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench",
"candidate": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\versions\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16",
"originalProcess": {
"pid": 36552,
"parentPid": 32968,
"createdAt": "2026-10-31T06:30:34.359Z",
"executable": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.14.0-win-x64-20261030235440461\\runtime\\node.exe",
"userSid": "S-1-5-21-3906741337-925321518-3767092861-1001"
},
"realModelRequests": 0,
"scope": "Accepted product setup stages the existing install, one in-memory handler upgrade preserves the original PID/Provider, pointer activation and final read-only validation. Every original table/row, all session exports and artifact bytes compared; no model Key extraction or QA certificate/device migration."
}
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
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,10 @@
import fs from 'node:fs';
import assert from 'node:assert/strict';
import { spawn } from 'node:child_process';
const source=JSON.parse(fs.readFileSync('.qa/protocol-f-source-result.json'));assert.equal(source.success,true);
const phases=['formal-stage','formal-upgrade','formal-activate','formal-ui','formal-verify','formal-artifacts'];
for(const phase of phases){
const child=spawn(process.execPath,['.qa/run-protocol-phase.mjs',phase],{windowsHide:true,stdio:['ignore','inherit','inherit']});
const code=await new Promise((resolve,reject)=>{child.once('error',reject);child.once('close',resolve);});
if(code!==0){process.exitCode=1;break;}
}
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,21 @@
{
"schema": 1,
"app": "agent-workbench-installation",
"installId": "74c11bdd-4895-459b-8403-097f79980b41",
"revision": 3,
"dataRoot": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench",
"port": 4318,
"active": {
"directory": "Agent-Workbench-1.16.0-win-x64-20261031114604110",
"version": "1.16.0",
"manifestSha256": "5ee8b7f83e4d42ac8499c684d2a38b64ccce1b7d5936aca0cb455a9e23688629",
"runtimeIdentity": "6a81e7fec198be881e2110a14004cb1ab32a916a784391c28dcf6d9b6ee86270"
},
"previous": {
"directory": "Agent-Workbench-1.15.0-win-x64-20261031071202956",
"version": "1.15.0",
"manifestSha256": "bac1a84429dea1b9f03c8fe5a538cd831b94469c888d64e23328b539551f6227",
"runtimeIdentity": "f3595cafc771df3ea938767fc4b1125c235f43cb9d16d2a3d8f5bffe12e82607"
},
"activatedAt": "2026-10-31T12:50:24.152Z"
}
@@ -0,0 +1,7 @@
{
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-original.sqlite",
"bytes": 13926400,
"sha256": "2a93faadcbf9693ef09ccf5f9e22b2a7e08eafd868995b102655729e2bbfff05",
"credentialsExtracted": false,
"committed": false
}
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,31 @@
{
"success": true,
"phase": "staged",
"version": "1.17.0",
"candidateIdentity": "a37c85e60d13624e8929586aaab798da5e8df63a4c68a7bf79eb2987304e0d80",
"manifestSha256": "79dc4ef4023ed5596f24e32ac9de8153a9f7d42f90e74917514c1dc2e19fffe7",
"retained": {
"originalRows": "exact",
"originalExports": "exact",
"originalArtifactBytes": "exact",
"originalModelCenter": "exact"
},
"setupLog": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-setup.log",
"setupLogSha256": "e0741a8f82c3e202a1bb49e06e331a61fb9e1c7ecd77adc16c8b4d76687d7a36",
"setupOutputEmpty": true,
"startedAt": "2026-10-31T15:44:48.657Z",
"finishedAt": "2026-10-31T15:55:27.450Z",
"installRoot": "C:/Users/wangy/AppData/Local/Programs/AgentWorkbench",
"dataRoot": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench",
"candidate": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\versions\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16",
"originalProcess": {
"pid": 36552,
"parentPid": 32968,
"createdAt": "2026-10-31T06:30:34.359Z",
"executable": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.14.0-win-x64-20261030235440461\\runtime\\node.exe",
"userSid": "S-1-5-21-3906741337-925321518-3767092861-1001"
},
"realModelRequests": 0,
"scope": "Accepted product setup stages the existing install, one in-memory handler upgrade preserves the original PID/Provider, pointer activation and final read-only validation. Every original table/row, all session exports and artifact bytes compared; no model Key extraction or QA certificate/device migration."
}
@@ -0,0 +1,180 @@
{
"success": true,
"version": "1.17.0",
"pid": 36552,
"directory": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460",
"selectedHistory": {
"projectId": "46d9ba13-b48c-43a2-a726-fc8ad47badb1",
"sessionId": "3fffee63-edc6-4538-b1dd-941354535e85",
"selection": "Actual current-project selector; no task created."
},
"results": [
{
"width": 1280,
"height": 900,
"panel": "执行计划",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\plan-1280.png",
"screenshotSha256": "d21db500bafc718aec3b9a700da5ad567c24c5b5ada8f7596d87e9c1dec68c1a"
},
{
"width": 1280,
"height": 900,
"panel": "模型中心",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\model-center-1280.png",
"screenshotSha256": "97b28034d7db7e8488e71c9330cd37663b60c6304b6a59cf83e93e1efab832b9"
},
{
"width": 1280,
"height": 900,
"panel": "本机加密凭据只读状态",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\credentials-1280.png",
"screenshotSha256": "d4d300dbf0fc53a4e83e4b7f7e4e66a629202ae1b4683ed318d9d7748ae10214"
},
{
"width": 1280,
"height": 900,
"panel": "运行巡检及实际覆盖范围",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\operations-1280.png",
"screenshotSha256": "4002b9ba116ef58824877c65af6f01dc2c1e72fe70c75a9dbff47074ae72d0c6"
},
{
"width": 1280,
"height": 900,
"panel": "项目技能",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\skills-1280.png",
"screenshotSha256": "9c05bb4944d3f0013a47ffe44aef5ca1f3d23335d0416575e3894d3ced17b3d1"
},
{
"width": 1280,
"height": 900,
"panel": "项目 MCP 工具",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\mcp-1280.png",
"screenshotSha256": "345af1ddf5b0081887c73fa42ea2f87b2d747302673ca77eb19254667c3bd69c"
},
{
"width": 1280,
"height": 900,
"panel": "隔离协作",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\controlled-engine-1280.png",
"screenshotSha256": "4ab2c3b16c00d800352a8b4ba667da0239ce36fb32c710e57755cc754f0deb5b"
},
{
"width": 1280,
"height": 900,
"panel": "受控引擎安装与默认关闭策略",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\controlled-native-1280.png",
"screenshotSha256": "3b35addb6c86c6a56723e169a1c9f0eeae6549c5e0b208b717546affd0f9a4b6"
},
{
"width": 1280,
"height": 900,
"panel": "Purge来源只读核对",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\controlled-purge-1280.png",
"screenshotSha256": "247511a5a6476d94bb851673e41163647a9fe3bb5e37d7ee0a1d00c205bad1f2"
},
{
"width": 1280,
"height": 900,
"panel": "默认关闭的手机网关与设备管理",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\companion-1280.png",
"screenshotSha256": "a4963159ae1946384a3532de157d65e86d9a72b0d1fcc3d0a82ac0d654f231d9"
},
{
"width": 390,
"height": 844,
"panel": "执行计划",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\plan-390.png",
"screenshotSha256": "47381f6434a803d18b98ff49cc92417d37274c759bfa9665b1475419239c9517"
},
{
"width": 390,
"height": 844,
"panel": "模型中心",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\model-center-390.png",
"screenshotSha256": "866d712640b84ed12c6c6fa47857347a48d1561aaf362e9c81a4445974915b9b"
},
{
"width": 390,
"height": 844,
"panel": "本机加密凭据只读状态",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\credentials-390.png",
"screenshotSha256": "94ff79882211c6898128c5d967a242f7a9ed2103faaa3fbc38a9b52838b4893b"
},
{
"width": 390,
"height": 844,
"panel": "运行巡检及实际覆盖范围",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\operations-390.png",
"screenshotSha256": "f30b92df3f6f979499797f7d464d569ca03e5386872b9a0f25b65b3c7939e378"
},
{
"width": 390,
"height": 844,
"panel": "项目技能",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\skills-390.png",
"screenshotSha256": "2019ba395495625ceb76bfc329d8ec5c49278cb254685db0c73c6aaf575468d6"
},
{
"width": 390,
"height": 844,
"panel": "项目 MCP 工具",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\mcp-390.png",
"screenshotSha256": "e17a464a68ef0cb2dd374b7abc72c42047e45cb158934899269b38ce2bf5977b"
},
{
"width": 390,
"height": 844,
"panel": "隔离协作",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\controlled-engine-390.png",
"screenshotSha256": "232c4cc91d6ac3c212db990b7dbae4fb5360a784a0a91634be3ed911d18adf81"
},
{
"width": 390,
"height": 844,
"panel": "受控引擎安装与默认关闭策略",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\controlled-native-390.png",
"screenshotSha256": "125993490aeea27f5dd6b9ecc72c7cd90b1f131da0f34ca9dfba7f1f25f86386"
},
{
"width": 390,
"height": 844,
"panel": "Purge来源只读核对",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\controlled-purge-390.png",
"screenshotSha256": "2deca05c1b43f8f582c40e621eabb8503a78e1ae9d23eb1b2b33a87f0926e6a9"
},
{
"width": 390,
"height": 844,
"panel": "默认关闭的手机网关与设备管理",
"readOnly": true,
"screenshot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\.qa\\protocol-f-formal-ui-f9e8b687-5ca7-4fc0-955e-31680e249460\\companion-390.png",
"screenshotSha256": "0bc56fffc81eb08822f441723b6dfc446d292fece33393b5d27cb06a2e1b3428"
}
],
"testedAt": "2026-10-31T16:07:45.990Z",
"pageErrors": [],
"requestFailures": [],
"httpErrors": [],
"blockedWrites": [],
"modelRequests": 0,
"scope": "Read-only smoke of deployed host in actual Edge desktop/390px; API mutations are blocked, no task/tool/model/configuration/project writes."
}
@@ -0,0 +1,45 @@
{
"success": true,
"phase": "upgraded",
"version": "1.17.0",
"upgraded": {
"version": "1.17.0",
"pid": 36552,
"hasKey": true,
"projects": 17,
"sessions": 30,
"appDirectory": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\versions\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"verified": true,
"connectionPreserved": true,
"url": "http://127.0.0.1:4318"
},
"retained": {
"originalRows": "exact",
"originalExports": "exact",
"originalArtifactBytes": "exact",
"originalModelCenter": "exact-with-validated-1.17-read-defaults",
"schemaDefaults": [
"streaming:false",
"capabilitySchema:1",
"explicit declarations",
"text/streaming capability state"
],
"originalStoredRows": "exact-no-migration"
},
"currentProviderPreserved": true,
"startedAt": "2026-10-31T15:55:28.512Z",
"finishedAt": "2026-10-31T15:57:33.657Z",
"installRoot": "C:/Users/wangy/AppData/Local/Programs/AgentWorkbench",
"dataRoot": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench",
"candidate": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\versions\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16",
"originalProcess": {
"pid": 36552,
"parentPid": 32968,
"createdAt": "2026-10-31T06:30:34.359Z",
"executable": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.14.0-win-x64-20261030235440461\\runtime\\node.exe",
"userSid": "S-1-5-21-3906741337-925321518-3767092861-1001"
},
"realModelRequests": 0,
"scope": "Accepted product setup stages the existing install, one in-memory handler upgrade preserves the original PID/Provider, pointer activation and final read-only validation. Every original table/row, all session exports and artifact bytes compared; no model Key extraction or QA certificate/device migration."
}
@@ -0,0 +1,90 @@
{
"success": true,
"phase": "verified",
"version": "1.17.0",
"activated": null,
"identity": {
"schema": 1,
"protocol": "agent-workbench-host-v1",
"app": "agent-workbench",
"version": "1.17.0",
"instanceId": "4c910dd5-82cb-4c8a-92f6-328338ea0662",
"pid": 36552,
"createdAt": "2026-10-31T06:30:34.359Z",
"userSid": "S-1-5-21-3906741337-925321518-3767092861-1001",
"ownerId": "b3ae90d3-9aa1-4b40-bd96-a53f01dbb502",
"executable": {
"path": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.14.0-win-x64-20261030235440461\\runtime\\node.exe",
"bytes": 93915464,
"sha256": "d702daa815eabbff763c7d3808e235cf031cf2ca70211b56847bc72f51e240e5"
},
"appRoot": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.14.0-win-x64-20261030235440461",
"runtimeRoot": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\versions\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"runtimeIdentity": "a37c85e60d13624e8929586aaab798da5e8df63a4c68a7bf79eb2987304e0d80",
"dataRoot": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench",
"port": 4318
},
"pointer": {
"schema": 1,
"app": "agent-workbench-installation",
"installId": "74c11bdd-4895-459b-8403-097f79980b41",
"revision": 4,
"dataRoot": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench",
"port": 4318,
"active": {
"directory": "Agent-Workbench-1.17.0-win-x64-20261031150548103",
"version": "1.17.0",
"manifestSha256": "79dc4ef4023ed5596f24e32ac9de8153a9f7d42f90e74917514c1dc2e19fffe7",
"runtimeIdentity": "a37c85e60d13624e8929586aaab798da5e8df63a4c68a7bf79eb2987304e0d80"
},
"previous": {
"directory": "Agent-Workbench-1.16.0-win-x64-20261031114604110",
"version": "1.16.0",
"manifestSha256": "5ee8b7f83e4d42ac8499c684d2a38b64ccce1b7d5936aca0cb455a9e23688629",
"runtimeIdentity": "6a81e7fec198be881e2110a14004cb1ab32a916a784391c28dcf6d9b6ee86270"
},
"activatedAt": "2026-10-31T16:07:35.060Z"
},
"retained": {
"originalRows": "exact",
"originalExports": "exact",
"originalArtifactBytes": "exact",
"originalModelCenter": "exact-with-validated-1.17-read-defaults",
"schemaDefaults": [
"streaming:false",
"capabilitySchema:1",
"explicit declarations",
"text/streaming capability state"
],
"originalStoredRows": "exact-no-migration"
},
"companion": {
"enabled": false,
"devices": 0,
"certificate": null
},
"currentProviderPreserved": true,
"counts": {
"projects": 17,
"sessions": 30,
"records": 2590,
"events": 4653,
"artifacts": 1366
},
"inspectorClosed": true,
"startedAt": "2026-10-31T16:07:46.388Z",
"finishedAt": "2026-10-31T16:09:56.718Z",
"installRoot": "C:/Users/wangy/AppData/Local/Programs/AgentWorkbench",
"dataRoot": "C:\\Users\\wangy\\AppData\\Local\\Packages\\OpenAI.Codex_2p2nqsd0c76g0\\LocalCache\\Local\\AgentWorkbench",
"candidate": "C:\\Users\\wangy\\AppData\\Local\\Programs\\AgentWorkbench\\versions\\Agent-Workbench-1.17.0-win-x64-20261031150548103",
"setupSha256": "3d60a3aab877268826dba8bf0d346eb154e4e2ed39aa034ee847b25985e8be16",
"originalProcess": {
"pid": 36552,
"parentPid": 32968,
"createdAt": "2026-10-31T06:30:34.359Z",
"executable": "D:\\Ai-project\\REA逆向\\Agent-Workbench\\releases\\Agent-Workbench-1.14.0-win-x64-20261030235440461\\runtime\\node.exe",
"userSid": "S-1-5-21-3906741337-925321518-3767092861-1001"
},
"realModelRequests": 0,
"scope": "Accepted product setup stages the existing install, one in-memory handler upgrade preserves the original PID/Provider, pointer activation and final read-only validation. Every original table/row, all session exports and artifact bytes compared; no model Key extraction or QA certificate/device migration."
}
File diff suppressed because it is too large Load Diff

Some files were not shown because too many files have changed in this diff Show More