feat: connect controlled engines to owned Tasks, approvals and budget ledger

This commit is contained in:
Agent Workbench Dev
2026-10-31 05:04:55 +08:00
parent a01c41811b
commit f0eecbda47
27 changed files with 1653 additions and 33 deletions
+13 -1
View File
@@ -12,6 +12,18 @@
实际Harness加本地固定模型夹具通过:完整启动目录核对、两次代理模型请求、一次原工具调用、prompt入队收据、所属原活动输出与正常退出;0真实模型请求、0stderr。原输入/协议/通知/桥数据与退出保存在controlled-engine-native-first.json及对应-log.txt。本轮只读取独立夹具README,没有开放用户工程或现有真实连接。
## B:执行账本与权限接入
原引擎工具进入所属Runtime的校验/审批/原结果账本,调用ID以引擎Run命名空间保存原sourceCallId。下一轮直接核对savedToolResult并封存真正反馈的工件;读取/提案/检查与工作台既有范围一致。引擎来源、所属叠加输入在接收工具、开始工具和批准后执行前复核,等待审批期间的变化不能使用旧许可。模型Key留在工作台Provider,原Native输入与实际兼容Chat请求分别保存。
首轮17项15通过/2断言失败,原spec输出保持;按既有check_status/message契约修正。第二轮相关兼容67/67通过;随后最新内核25/25通过,包括等待审批来源变化、损坏报告拒绝交付、原模型预算恢复单次记账、未知核对和完整HTTP原记录分页/下载/归属。各轮分别保存,不拼接统计。界面代码和原生三工程尚待独立验收。
关闭原SDK进程后的桥请求也拒绝进入模型/工具回调,追加7/7桥专项通过。原记录分别为controlled-engine-worker-first-spec-log.txt、controlled-engine-worker-compatible.tap、controlled-engine-worker-kernel.tap和controlled-engine-live-bridge.tap;来源字节和当前内核输入索引为controlled-engine-kernel-evidence.json。
本轮模型实际尝试由引擎桥记账,外层合成工具步骤不再重复记账。重启对原reservedPerAttempt/maxAttempts工件复核后保守记录未封存尝试,第二次恢复不重复计账;损坏预算不能减少累计用量。正常输出前关闭并封存引擎,worker成果仍需原报告、实际提案/检查和父审查。
已加入启动/运行中原创建时间和后代身份观察,关闭后复核完整所属树;活后代或观察失败保留unknown。原SDK会话不自动resume,人工核对独立保存review,原输出不改写为成功。下一步实际Harness+HTTP固定模型三工程及桌面/390px交付验证。
## 尚待验收
B:受限桥与自有Task/Run审批/账本接入;C:独立工程、累计预算、进程树和父审查;D:恢复/桌面/390px;E:Purge来源/兼容/同题对照;F:完整回归、独立发行及保留状态升级。正式4318继续使用不可变1.13.0。
C:实际Harness独立工程/父审查;D:恢复/桌面/390px;E:Purge来源/兼容/同题对照;F:完整回归、独立发行及保留状态升级。正式4318继续使用不可变1.13.0。
+1 -1
View File
@@ -5,7 +5,7 @@
| 切片 | 当前实现/验证 | 遗留与下一步 |
| --- | --- | --- |
| A:版本/协议 | 新受控安装身份、七份关键运行器、独立profile叠加/桥;7/7边界专项和实际Harness固定模型2请求/1工具/收据/wait/正常退出通过 | 原SDK没有单会话取消或反向审批;桥不能单凭启动成功宣称工程交付 |
| B:工具桥 | 原生项目/终端/模型/重试插件禁用,appReady核对完整工具schema;原进程/Run/SDK/来源、输入上限和重复ID守卫 | 接入自有执行账本、文件审查及固定检查,保持原工具身份和未知状态 |
| B:工具桥 | 原目录/身份桥接自有Runtime/账本,命名空间保留sourceCallId,反馈核对真实结果;等待审批的来源复核、单次尝试计账、报告损坏/预算恢复/HTTP分页已通过最新25项;原67项兼容保持 | 实际Harness三工程及最终冻结发行单独验收;已关闭原进程的桥请求追加拒绝 |
| C–F | 尚未完成 | 独立工程/停止恢复/父审查、真实界面、Purge同题对照、完整发行与状态保持升级 |
前置研究确认sdk-minimal仍含danger-full-access策略和本机工具,不能把profile名字当权限边界;采用宿主受限桥和实际目录检查。Harness现有DeepSeek provider使用Messages API,本工作台现有连接是兼容Chat层;自有代理provider保持两层分离,使引擎不取得模型Key,也不把一种协议冒充另一种协议。
+110
View File
@@ -0,0 +1,110 @@
{
"OriginalLogs": [
{
"Passed": 15,
"Tests": 17,
"Sha256": "30a580a96ccf133f1dc7eb854555f3083bfab1fcb18ba0dbd19699d6f90a3871",
"Target": "docs\\controlled-engine-worker-first-spec-log.txt",
"Source": ".qa\\controlled-worker-first.tap",
"Bytes": 3808
},
{
"Passed": 67,
"Tests": 67,
"Sha256": "aab191b90b1afeba5a04e6d1b791cb6ba88e7d9fbe27846b5197329abcd2fa72",
"Target": "docs\\controlled-engine-worker-compatible.tap",
"Source": ".qa\\controlled-worker-second.tap",
"Bytes": 18885
},
{
"Passed": 25,
"Tests": 25,
"Sha256": "79ddc25b37d7fbdf644e5610b7dc265a742d98443ce9bb51ddc2462a704f80c3",
"Target": "docs\\controlled-engine-worker-kernel.tap",
"Source": ".qa\\controlled-worker-final-kernel.tap",
"Bytes": 7231
},
{
"Passed": 7,
"Tests": 7,
"Sha256": "66f50b9d67268a39b31ca71049682443715385aedc698fc0349450610d9c2a7f",
"Target": "docs\\controlled-engine-live-bridge.tap",
"Source": ".qa\\controlled-bridge-live-guards.tap",
"Bytes": 2169
}
],
"FullReleaseAccepted": false,
"Stage": "1.14 B kernel",
"RealModelRequests": 0,
"CurrentKernelInputs": [
{
"Sha256": "ced3adcea9badc6542e7cd123dc8ea1034f6010b594f058bed5a5e110dfa8c0b",
"Path": "src/core/engines/engine-runner.mjs"
},
{
"Sha256": "5fae9b63d4ddea31b523fb529d8954766123785aeffe31e067b16325f200f9e9",
"Path": "src/core/engines/controlled-session.mjs"
},
{
"Sha256": "ffe6bae64285f2e47128d5227f091f2346d923ff9a170b39728f64350b1a47e3",
"Path": "src/core/engines/controlled-plugin.mjs"
},
{
"Sha256": "6a65e9c34889e093fe64772199582b115ba7f6b1c0965a5c3507f1098c9019be",
"Path": "src/core/engines/controlled-installation.mjs"
},
{
"Sha256": "3d4e35659422d63805b90502e80a95cd4a2367c22e8758251479b5a8895db605",
"Path": "src/shared/contracts/engine-execution.mjs"
},
{
"Sha256": "cd0a6378682182b8c29884e725a284a7b4f32c1355067883321de071ff113bfb",
"Path": "src/shared/contracts/engines.mjs"
},
{
"Sha256": "7f5f50db8996d0449601dc6f219e4b241895745ce52acaa835636d5435ace0f7",
"Path": "src/shared/contracts/workers.mjs"
},
{
"Sha256": "6d25f6048a5fc42358a067d64cd4715e35c19ef99b28c85d739b50e73bdbb544",
"Path": "src/shared/contracts/execution.mjs"
},
{
"Sha256": "3a1284e55b6fa8f3cf9a2c42ab448673c8aff874577fab209db7629473614461",
"Path": "src/core/tasks/worker-manager.mjs"
},
{
"Sha256": "855da237e17877212ca909967cd79d818ed38c7a00251f6733e4c727ed72d505",
"Path": "src/server/app.mjs"
},
{
"Sha256": "4697e0bd0266d305e202c796cf3436a97701fae1bba2434d3663c9b89e29be1b",
"Path": "src/server/store.mjs"
},
{
"Sha256": "9b362c96e5c78d77afd43ac79b28c1edaae5b4f753f3cf079fef6605914507d9",
"Path": "src/server/runtime.mjs"
},
{
"Sha256": "2d22386c679b666de63efcc43d8a5703b166358667d48b17cc544c0cfddc6393",
"Path": "src/server/provider.mjs"
},
{
"Sha256": "c2d344fe8b795789073a16848c910a96fdd95103caf1a6092fcd7a3f0e5f5fbb",
"Path": "tests/controlled-engine.test.mjs"
},
{
"Sha256": "2af6091fb1de61c93d79a305a7c6fdb6ae7d5c09d9e677319cab8e5fe9337ba7",
"Path": "tests/engine-worker.test.mjs"
},
{
"Sha256": "ed3880a1becaa99d836f2c85a76967a6d443b7ec2c71beec8e26360c1040266b",
"Path": "tests/engine-worker-http.test.mjs"
},
{
"Sha256": "a9454456d6747bdff786089d2b7992b873a35dd2bdb54735f1dced4f43cc610e",
"Path": "tests/fixtures/controlled-worker.mjs"
}
],
"CreatedAt": "2026-10-30T21:04:53.2660093Z"
}
+52
View File
@@ -0,0 +1,52 @@
TAP version 13
# Subtest: controlled installation pins the native kernel and bridge independently of the existing connection-probe identity
ok 1 - controlled installation pins the native kernel and bridge independently of the existing connection-probe identity
---
duration_ms: 924.3909
type: 'test'
...
# Subtest: controlled profiles disable every native project/terminal provider and native model/retry route using the actual include patch contract
ok 2 - controlled profiles disable every native project/terminal provider and native model/retry route using the actual include patch contract
---
duration_ms: 0.5062
type: 'test'
...
# Subtest: owned tool catalogues seal full schemas canonically and reject duplicate or malformed declarations
ok 3 - owned tool catalogues seal full schemas canonically and reject duplicate or malformed declarations
---
duration_ms: 5.4906
type: 'test'
...
# Subtest: bridge authentication, current Run, process, SDK session and browser-origin checks reject before dispatch
ok 4 - bridge authentication, current Run, process, SDK session and browser-origin checks reject before dispatch
---
duration_ms: 773.2805
type: 'test'
...
# Subtest: bridge duplicates, unknown tools, malformed arguments and source changes are refused before a second side effect
ok 5 - bridge duplicates, unknown tools, malformed arguments and source changes are refused before a second side effect
---
duration_ms: 694.599
type: 'test'
...
# Subtest: bridge model requests bind the exact catalogue, route and output bound and preserve raw request/result identity
ok 6 - bridge model requests bind the exact catalogue, route and output bound and preserve raw request/result identity
---
duration_ms: 652.8104
type: 'test'
...
# Subtest: stopped and oversized bridge requests never reach a local model or tool callback
ok 7 - stopped and oversized bridge requests never reach a local model or tool callback
---
duration_ms: 656.527
type: 'test'
...
1..7
# tests 7
# suites 0
# pass 7
# fail 0
# cancelled 0
# skipped 0
# todo 0
# duration_ms 3799.7591
@@ -0,0 +1,412 @@
TAP version 13
# Subtest: controlled installation pins the native kernel and bridge independently of the existing connection-probe identity
ok 1 - controlled installation pins the native kernel and bridge independently of the existing connection-probe identity
---
duration_ms: 816.2291
type: 'test'
...
# Subtest: controlled profiles disable every native project/terminal provider and native model/retry route using the actual include patch contract
ok 2 - controlled profiles disable every native project/terminal provider and native model/retry route using the actual include patch contract
---
duration_ms: 0.5492
type: 'test'
...
# Subtest: owned tool catalogues seal full schemas canonically and reject duplicate or malformed declarations
ok 3 - owned tool catalogues seal full schemas canonically and reject duplicate or malformed declarations
---
duration_ms: 5.4234
type: 'test'
...
# Subtest: bridge authentication, current Run, process, SDK session and browser-origin checks reject before dispatch
ok 4 - bridge authentication, current Run, process, SDK session and browser-origin checks reject before dispatch
---
duration_ms: 950.2421
type: 'test'
...
# Subtest: bridge duplicates, unknown tools, malformed arguments and source changes are refused before a second side effect
ok 5 - bridge duplicates, unknown tools, malformed arguments and source changes are refused before a second side effect
---
duration_ms: 679.3481
type: 'test'
...
# Subtest: bridge model requests bind the exact catalogue, route and output bound and preserve raw request/result identity
ok 6 - bridge model requests bind the exact catalogue, route and output bound and preserve raw request/result identity
---
duration_ms: 697.0703
type: 'test'
...
# Subtest: stopped and oversized bridge requests never reach a local model or tool callback
ok 7 - stopped and oversized bridge requests never reach a local model or tool callback
---
duration_ms: 590.4988
type: 'test'
...
# Subtest: engine HTTP checks bind the detected installation and current task, page original bytes and preserve them across host restart
ok 8 - engine HTTP checks bind the detected installation and current task, page original bytes and preserve them across host restart
---
duration_ms: 2573.4285
type: 'test'
...
# Subtest: HTTP stop seals a pending SDK call, rejects automatic replay and keeps failures attached to the original session
ok 9 - HTTP stop seals a pending SDK call, rejects automatic replay and keeps failures attached to the original session
---
duration_ms: 2958.6899
type: 'test'
...
# Subtest: native text and call IDs preserve original result pairing while unsupported content fails closed
ok 10 - native text and call IDs preserve original result pairing while unsupported content fails closed
---
duration_ms: 0.7762
type: 'test'
...
# Subtest: one engine child freezes real installation identity before creating tasks and rejects a changed source
ok 11 - one engine child freezes real installation identity before creating tasks and rejects a changed source
---
duration_ms: 5899.1837
type: 'test'
...
# Subtest: owned native tools return saved results, use real approvals/checks and account each actual request once across new engine Runs
ok 12 - owned native tools return saved results, use real approvals/checks and account each actual request once across new engine Runs
---
duration_ms: 17601.082
type: 'test'
...
# Subtest: forged native scope cannot change a parent file and a final completion claim cannot replace worker proof
ok 13 - forged native scope cannot change a parent file and a final completion claim cannot replace worker proof
---
duration_ms: 9063.3297
type: 'test'
...
# Subtest: duplicate native IDs are rejected without replaying a saved tool
ok 14 - duplicate native IDs are rejected without replaying a saved tool
---
duration_ms: 8481.8966
type: 'test'
...
# Subtest: native input exceeding the context budget is retained and sends no model request
ok 15 - native input exceeding the context budget is retained and sends no model request
---
duration_ms: 7140.3817
type: 'test'
...
# Subtest: stop retains a late reported model usage once and executes no tool from that output
ok 16 - stop retains a late reported model usage once and executes no tool from that output
---
duration_ms: 7565.0223
type: 'test'
...
# Subtest: a source change after the original model intent stops before native tools execute
ok 17 - a source change after the original model intent stops before native tools execute
---
duration_ms: 7435.7032
type: 'test'
...
# Subtest: missing or noncompleted turn end cannot be adopted even with an owned receipt and assistant text
ok 18 - missing or noncompleted turn end cannot be adopted even with an owned receipt and assistant text
---
duration_ms: 0.5515
type: 'test'
...
# Subtest: closed root with an original live descendant remains unknown and retains its process tree
ok 19 - closed root with an original live descendant remains unknown and retains its process tree
---
duration_ms: 0.2928
type: 'test'
...
# Subtest: host recovery reserves original pending attempts exactly once and explicit review preserves unknown outcomes
ok 20 - host recovery reserves original pending attempts exactly once and explicit review preserves unknown outcomes
---
duration_ms: 10865.7804
type: 'test'
...
# Subtest: corrupted model reservation cannot reduce recovered budget or permit a new engine request
ok 21 - corrupted model reservation cannot reduce recovered budget or permit a new engine request
---
duration_ms: 8811.645
type: 'test'
...
# Subtest: engine detection seals the actual CLI, SDK, profile and Node and rejects mixed versions without launch
ok 22 - engine detection seals the actual CLI, SDK, profile and Node and rejects mixed versions without launch
---
duration_ms: 731.2372
type: 'test'
...
# Subtest: owned engine probe waits for real ready identity, verifies six native responses and saves immutable task evidence
ok 23 - owned engine probe waits for real ready identity, verifies six native responses and saves immutable task evidence
---
duration_ms: 1384.3111
type: 'test'
...
# Subtest: an invalid startup nonce is rejected before initialize and preserves the process close and partial report
ok 24 - an invalid startup nonce is rejected before initialize and preserves the process close and partial report
---
duration_ms: 1236.3923
type: 'test'
...
# Subtest: SDK identity and session directory mismatches retain raw responses and never become verified
ok 25 - SDK identity and session directory mismatches retain raw responses and never become verified
---
duration_ms: 3841.0501
type: 'test'
...
# Subtest: cancel during a pending initialize closes only the owned process without sending an unsupported MCP cancellation
ok 26 - cancel during a pending initialize closes only the owned process without sending an unsupported MCP cancellation
---
duration_ms: 1337.7515
type: 'test'
...
# Subtest: cancel before readiness sends no protocol request and a user retry creates a different process and record
ok 27 - cancel before readiness sends no protocol request and a user retry creates a different process and record
---
duration_ms: 1766.3237
type: 'test'
...
# Subtest: source or scratch changes and unexpected model traffic fail after preserving the actual observed result
ok 28 - source or scratch changes and unexpected model traffic fail after preserving the actual observed result
---
duration_ms: 3876.5429
type: 'test'
...
# Subtest: unowned idle notifications stay quarantined and server requests receive a rejection without granting operations
ok 29 - unowned idle notifications stay quarantined and server requests receive a rejection without granting operations
---
duration_ms: 2487.8291
type: 'test'
...
# Subtest: engine probe environment excludes parent secrets and NODE_OPTIONS while preserving operating system paths
ok 30 - engine probe environment excludes parent secrets and NODE_OPTIONS while preserving operating system paths
---
duration_ms: 1244.7198
type: 'test'
...
# Subtest: foreign sessions and altered reports or task bindings cannot reuse verified engine evidence
ok 31 - foreign sessions and altered reports or task bindings cannot reuse verified engine evidence
---
duration_ms: 1117.1807
type: 'test'
...
# Subtest: restart never replays an active engine check or stops a historical PID, and keeps prior task state
ok 32 - restart never replays an active engine check or stops a historical PID, and keeps prior task state
---
duration_ms: 1359.5664
type: 'test'
...
# Subtest: model argument validation and runtime activity do not grant external engine prompt or project execution
ok 33 - model argument validation and runtime activity do not grant external engine prompt or project execution
---
duration_ms: 661.7343
type: 'test'
...
# Subtest: a follow-up question receives the actual failed engine check from its original Task and keeps its report unchanged
ok 34 - a follow-up question receives the actual failed engine check from its original Task and keeps its report unchanged
---
duration_ms: 1925.1735
type: 'test'
...
# Subtest: HTTP worker identities require a current sealed plan, reject overrides and close unused plans without touching source
ok 35 - HTTP worker identities require a current sealed plan, reject overrides and close unused plans without touching source
---
duration_ms: 10680.8387
type: 'test'
...
# Subtest: HTTP unknown recovery only accepts the original evidence SHA and preserves changed source for explicit retention
ok 36 - HTTP unknown recovery only accepts the original evidence SHA and preserves changed source for explicit retention
---
duration_ms: 13802.8412
type: 'test'
...
# Subtest: verified isolated outputs form a new parent batch, rerun real checks and commit only reviewed source
ok 37 - verified isolated outputs form a new parent batch, rerun real checks and commit only reviewed source
---
duration_ms: 34905.4242
type: 'test'
...
# Subtest: overlapping worker paths are explicit conflicts even if each independent task checks successfully
ok 38 - overlapping worker paths are explicit conflicts even if each independent task checks successfully
---
duration_ms: 23076.2678
type: 'test'
...
# Subtest: source edits after isolated work preserve the owner file and invalidate collection without proposing a merge
ok 39 - source edits after isolated work preserve the owner file and invalidate collection without proposing a merge
---
duration_ms: 21509.2095
type: 'test'
...
# Subtest: changed child bytes after parent preview block the entire batch before any parent write
ok 40 - changed child bytes after parent preview block the entire batch before any parent write
---
duration_ms: 28009.8619
type: 'test'
...
# Subtest: corrupt original check logs cannot become collected outputs or a parent review
ok 41 - corrupt original check logs cannot become collected outputs or a parent review
---
duration_ms: 21518.3373
type: 'test'
...
# Subtest: stopping a parent accounts late responses and executes none of their proposed tools
ok 42 - stopping a parent accounts late responses and executes none of their proposed tools
---
duration_ms: 10380.3694
type: 'test'
...
# Subtest: unreported interrupted requests reserve estimated usage and the total deadline cannot reset on resume
ok 43 - unreported interrupted requests reserve estimated usage and the total deadline cannot reset on resume
---
duration_ms: 10919.4728
type: 'test'
...
# Subtest: a response exceeding the cumulative token allocation stops before registering or executing its tools
ok 44 - a response exceeding the cumulative token allocation stops before registering or executing its tools
---
duration_ms: 10330.442
type: 'test'
...
# Subtest: cumulative tool allocation rejects the entire returned batch before any tool side effect
ok 45 - cumulative tool allocation rejects the entire returned batch before any tool side effect
---
duration_ms: 10120.0901
type: 'test'
...
# Subtest: single response 16 tool ceiling rejects the entire returned batch before any tool side effect
ok 46 - single response 16 tool ceiling rejects the entire returned batch before any tool side effect
---
duration_ms: 10288.9937
type: 'test'
...
# Subtest: consumed request steps remain consumed across explicit worker resumes
ok 47 - consumed request steps remain consumed across explicit worker resumes
---
duration_ms: 11051.708
type: 'test'
...
# Subtest: a saved completed response with missing usage is conservatively accounted exactly once during recovery
ok 48 - a saved completed response with missing usage is conservatively accounted exactly once during recovery
---
duration_ms: 8958.148
type: 'test'
...
# Subtest: an unexecuted second workspace plan resumes its original identity without recreating the first checkout
ok 49 - an unexecuted second workspace plan resumes its original identity without recreating the first checkout
---
duration_ms: 9815.1841
type: 'test'
...
# Subtest: a verified interrupted parent proposal resumes sealing without duplicating or replacing its original change
ok 50 - a verified interrupted parent proposal resumes sealing without duplicating or replacing its original change
---
duration_ms: 32012.3475
type: 'test'
...
# Subtest: recovery attaches a sealed parent batch saved before its group reference without creating another batch
ok 51 - recovery attaches a sealed parent batch saved before its group reference without creating another batch
---
duration_ms: 28124.7263
type: 'test'
...
# Subtest: explicitly retaining an unexecuted parent merge cancels its batch and frees the next parent goal
ok 52 - explicitly retaining an unexecuted parent merge cancels its batch and frees the next parent goal
---
duration_ms: 28818.7589
type: 'test'
...
# Subtest: unknown collaboration blocks ordinary proposals but permits only original batch recovery and rollback
ok 53 - unknown collaboration blocks ordinary proposals but permits only original batch recovery and rollback
---
duration_ms: 31832.0692
type: 'test'
...
# Subtest: after a restart the batch application lazily reinstates the worker result guard before any parent write
ok 54 - after a restart the batch application lazily reinstates the worker result guard before any parent write
---
duration_ms: 27876.2984
type: 'test'
...
# Subtest: unknown original tool inputs must match before acknowledgement, and reviewed checkpoint IDs preserve the unknown fact
ok 55 - unknown original tool inputs must match before acknowledgement, and reviewed checkpoint IDs preserve the unknown fact
---
duration_ms: 11211.8706
type: 'test'
...
# Subtest: a parent stop observes and waits for a manually approved child check and its real descendant process
ok 56 - a parent stop observes and waits for a manually approved child check and its real descendant process
---
duration_ms: 10807.5113
type: 'test'
...
# Subtest: a real host crash preserves unknown checks, refuses a still living owned PID and never replays a model or tool
ok 57 - a real host crash preserves unknown checks, refuses a still living owned PID and never replays a model or tool
---
duration_ms: 19019.2047
type: 'test'
...
# Subtest: a persisted child initialization resumes the original control run after a failure before its worker record
ok 58 - a persisted child initialization resumes the original control run after a failure before its worker record
---
duration_ms: 9889.5335
type: 'test'
...
# Subtest: failed request intent persistence never dispatches a worker request or changes source
ok 59 - failed request intent persistence never dispatches a worker request or changes source
---
duration_ms: 10482.6794
type: 'test'
...
# Subtest: failed usage persistence rolls back the uncommitted usage and conservatively reserves one unknown request without tools
ok 60 - failed usage persistence rolls back the uncommitted usage and conservatively reserves one unknown request without tools
---
duration_ms: 10040.0269
type: 'test'
...
# Subtest: a child file change after observation prevents acknowledgement even when the tool inputs did not change
ok 61 - a child file change after observation prevents acknowledgement even when the tool inputs did not change
---
duration_ms: 10021.0669
type: 'test'
...
# Subtest: worker scopes and allocated budgets reject aliases, extra permissions and aggregate overflow
ok 62 - worker scopes and allocated budgets reject aliases, extra permissions and aggregate overflow
---
duration_ms: 653.4276
type: 'test'
...
# Subtest: two sequential Git workspaces freeze a new parent goal, separate owned tasks and preserve all source bytes
ok 63 - two sequential Git workspaces freeze a new parent goal, separate owned tasks and preserve all source bytes
---
duration_ms: 8956.1164
type: 'test'
...
# Subtest: real isolated changes, approvals and npm checks produce original owned outputs, not model completion claims
ok 64 - real isolated changes, approvals and npm checks produce original owned outputs, not model completion claims
---
duration_ms: 22042.2963
type: 'test'
...
# Subtest: forged worker tools cannot execute terminal or Git even when absent from the advertised tool list
ok 65 - forged worker tools cannot execute terminal or Git even when absent from the advertised tool list
---
duration_ms: 9099.6576
type: 'test'
...
# Subtest: a completion-only model pauses each worker without fabricated changes or checks
ok 66 - a completion-only model pauses each worker without fabricated changes or checks
---
duration_ms: 10901.1151
type: 'test'
...
# Subtest: restart retains frozen scope and marks interrupted parent and workers unknown without sending requests
ok 67 - restart retains frozen scope and marks interrupted parent and workers unknown without sending requests
---
duration_ms: 8780.2684
type: 'test'
...
1..67
# tests 67
# suites 0
# pass 67
# fail 0
# cancelled 0
# skipped 0
# todo 0
# duration_ms 643018.4941
@@ -0,0 +1,59 @@
✔ controlled installation pins the native kernel and bridge independently of the existing connection-probe identity (870.6639ms)
✔ controlled profiles disable every native project/terminal provider and native model/retry route using the actual include patch contract (0.5375ms)
✔ owned tool catalogues seal full schemas canonically and reject duplicate or malformed declarations (5.2037ms)
✔ bridge authentication, current Run, process, SDK session and browser-origin checks reject before dispatch (639.515ms)
✔ bridge duplicates, unknown tools, malformed arguments and source changes are refused before a second side effect (675.3767ms)
✔ bridge model requests bind the exact catalogue, route and output bound and preserve raw request/result identity (629.76ms)
✔ stopped and oversized bridge requests never reach a local model or tool callback (713.8051ms)
✔ native text and call IDs preserve original result pairing while unsupported content fails closed (0.8087ms)
✔ one engine child freezes real installation identity before creating tasks and rejects a changed source (6165.6107ms)
✖ owned native tools return saved results, use real approvals/checks and account each actual request once across new engine Runs (17171.0179ms)
✖ forged native scope cannot change a parent file and a final completion claim cannot replace worker proof (8831.888ms)
✔ duplicate native IDs are rejected without replaying a saved tool (8616.6411ms)
✔ native input exceeding the context budget is retained and sends no model request (7015.7346ms)
✔ stop retains a late reported model usage once and executes no tool from that output (7397.2983ms)
✔ a source change after the original model intent stops before native tools execute (7301.4165ms)
✔ missing or noncompleted turn end cannot be adopted even with an owned receipt and assistant text (0.6823ms)
✔ closed root with an original live descendant remains unknown and retains its process tree (0.3306ms)
ℹ tests 17
ℹ suites 0
ℹ pass 15
ℹ fail 2
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 66242.3459
✖ failing tests:
test at tests\engine-worker.test.mjs:41:1
✖ owned native tools return saved results, use real approvals/checks and account each actual request once across new engine Runs (17171.0179ms)
AssertionError [ERR_ASSERTION]: The expression evaluated to a falsy value:
assert.ok(context.requests.some(request => request.messages.some(message => message.role === 'tool' && JSON.parse(message.content).status === 'passed')))
at TestContext.<anonymous> (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/tests/engine-worker.test.mjs:63:10)
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: false,
expected: true,
operator: '==',
diff: 'simple'
}
test at tests\engine-worker.test.mjs:68:1
✖ forged native scope cannot change a parent file and a final completion claim cannot replace worker proof (8831.888ms)
AssertionError [ERR_ASSERTION]: The "string" argument must be of type string. Received type undefined (undefined)
at TestContext.<anonymous> (file:///D:/Ai-project/REA%E9%80%86%E5%90%91/Agent-Workbench/tests/engine-worker.test.mjs:76:78)
at process.processTicksAndRejections (node:internal/process/task_queues:104:5)
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: /范围/,
operator: 'match',
diff: 'simple'
}
+160
View File
@@ -0,0 +1,160 @@
TAP version 13
# Subtest: controlled installation pins the native kernel and bridge independently of the existing connection-probe identity
ok 1 - controlled installation pins the native kernel and bridge independently of the existing connection-probe identity
---
duration_ms: 823.5923
type: 'test'
...
# Subtest: controlled profiles disable every native project/terminal provider and native model/retry route using the actual include patch contract
ok 2 - controlled profiles disable every native project/terminal provider and native model/retry route using the actual include patch contract
---
duration_ms: 0.5387
type: 'test'
...
# Subtest: owned tool catalogues seal full schemas canonically and reject duplicate or malformed declarations
ok 3 - owned tool catalogues seal full schemas canonically and reject duplicate or malformed declarations
---
duration_ms: 5.7335
type: 'test'
...
# Subtest: bridge authentication, current Run, process, SDK session and browser-origin checks reject before dispatch
ok 4 - bridge authentication, current Run, process, SDK session and browser-origin checks reject before dispatch
---
duration_ms: 769.9066
type: 'test'
...
# Subtest: bridge duplicates, unknown tools, malformed arguments and source changes are refused before a second side effect
ok 5 - bridge duplicates, unknown tools, malformed arguments and source changes are refused before a second side effect
---
duration_ms: 654.6547
type: 'test'
...
# Subtest: bridge model requests bind the exact catalogue, route and output bound and preserve raw request/result identity
ok 6 - bridge model requests bind the exact catalogue, route and output bound and preserve raw request/result identity
---
duration_ms: 699.4788
type: 'test'
...
# Subtest: stopped and oversized bridge requests never reach a local model or tool callback
ok 7 - stopped and oversized bridge requests never reach a local model or tool callback
---
duration_ms: 665.6117
type: 'test'
...
# Subtest: engine HTTP checks bind the detected installation and current task, page original bytes and preserve them across host restart
ok 8 - engine HTTP checks bind the detected installation and current task, page original bytes and preserve them across host restart
---
duration_ms: 2732.0424
type: 'test'
...
# Subtest: HTTP stop seals a pending SDK call, rejects automatic replay and keeps failures attached to the original session
ok 9 - HTTP stop seals a pending SDK call, rejects automatic replay and keeps failures attached to the original session
---
duration_ms: 3160.8457
type: 'test'
...
# Subtest: controlled engine HTTP freezes one child, pages original owned input/report, rejects foreign reads and prevents changed-source reuse
ok 10 - controlled engine HTTP freezes one child, pages original owned input/report, rejects foreign reads and prevents changed-source reuse
---
duration_ms: 11219.5604
type: 'test'
...
# Subtest: native text and call IDs preserve original result pairing while unsupported content fails closed
ok 11 - native text and call IDs preserve original result pairing while unsupported content fails closed
---
duration_ms: 0.8848
type: 'test'
...
# Subtest: one engine child freezes real installation identity before creating tasks and rejects a changed source
ok 12 - one engine child freezes real installation identity before creating tasks and rejects a changed source
---
duration_ms: 5971.1886
type: 'test'
...
# Subtest: owned native tools return saved results, use real approvals/checks and account each actual request once across new engine Runs
ok 13 - owned native tools return saved results, use real approvals/checks and account each actual request once across new engine Runs
---
duration_ms: 18042.8416
type: 'test'
...
# Subtest: forged native scope cannot change a parent file and a final completion claim cannot replace worker proof
ok 14 - forged native scope cannot change a parent file and a final completion claim cannot replace worker proof
---
duration_ms: 9566.7856
type: 'test'
...
# Subtest: duplicate native IDs are rejected without replaying a saved tool
ok 15 - duplicate native IDs are rejected without replaying a saved tool
---
duration_ms: 8715.6494
type: 'test'
...
# Subtest: native input exceeding the context budget is retained and sends no model request
ok 16 - native input exceeding the context budget is retained and sends no model request
---
duration_ms: 7280.4503
type: 'test'
...
# Subtest: stop retains a late reported model usage once and executes no tool from that output
ok 17 - stop retains a late reported model usage once and executes no tool from that output
---
duration_ms: 7614.7364
type: 'test'
...
# Subtest: a source change after the original model intent stops before native tools execute
ok 18 - a source change after the original model intent stops before native tools execute
---
duration_ms: 7676.3322
type: 'test'
...
# Subtest: an engine source change while approval is pending invalidates the old permission before any check starts
ok 19 - an engine source change while approval is pending invalidates the old permission before any check starts
---
duration_ms: 9807.1356
type: 'test'
...
# Subtest: missing or noncompleted turn end cannot be adopted even with an owned receipt and assistant text
ok 20 - missing or noncompleted turn end cannot be adopted even with an owned receipt and assistant text
---
duration_ms: 0.6841
type: 'test'
...
# Subtest: closed root with an original live descendant remains unknown and retains its process tree
ok 21 - closed root with an original live descendant remains unknown and retains its process tree
---
duration_ms: 0.2756
type: 'test'
...
# Subtest: host recovery reserves original pending attempts exactly once and explicit review preserves unknown outcomes
ok 22 - host recovery reserves original pending attempts exactly once and explicit review preserves unknown outcomes
---
duration_ms: 11081.577
type: 'test'
...
# Subtest: corrupted model reservation cannot reduce recovered budget or permit a new engine request
ok 23 - corrupted model reservation cannot reduce recovered budget or permit a new engine request
---
duration_ms: 9010.1322
type: 'test'
...
# Subtest: HTTP worker identities require a current sealed plan, reject overrides and close unused plans without touching source
ok 24 - HTTP worker identities require a current sealed plan, reject overrides and close unused plans without touching source
---
duration_ms: 10348.3848
type: 'test'
...
# Subtest: HTTP unknown recovery only accepts the original evidence SHA and preserves changed source for explicit retention
ok 25 - HTTP unknown recovery only accepts the original evidence SHA and preserves changed source for explicit retention
---
duration_ms: 14049.5917
type: 'test'
...
1..25
# tests 25
# suites 0
# pass 25
# fail 0
# cancelled 0
# skipped 0
# todo 0
# duration_ms 140419.1483