Compare commits

...

1 Commits

Author SHA1 Message Date
pptg
43e95cb9eb 独立核验最终控制面候选及A01至A12证据适用范围 2026-09-12 08:08:47 +08:00
63 changed files with 25619 additions and 0 deletions

View File

@ -0,0 +1,74 @@
# Issue 26 最终控制面候选独立审查与 QA
**结论:固定 `8d374084e109bbabc9e4bcec2a12c4088099db7d` 的控制面候选,在下列明确的源码和证据范围内通过。整体 Goal 未完成:扑克 #102 / PR #104 尚未合并及合入后 QA,最终 main 未发布。** 不替代扑克产品角色的 CR/QA,不宣布整个项目最终交付。
依据 Issue26 评论19523执行,仅在本clone;未修改产品/冻结验收、不操作4498或扑克、不重复昂贵A02/A03/600秒。业务及交接仅经Git/Gitea。本轮真实模型调用0,没有创建新产品业务任务。
## 当前实际验证
- 固定完整SHA:`8d374084e109bbabc9e4bcec2a12c4088099db7d`。后端树 `d606bea600991e60cd8a13138704054b5087c22b` 与独立通过的 `007bf0988f9bb333e765ed67a0d701ece2ab169e` 完全相同;前端相对411零diff。
- build/typecheck通过,后端30/30单测,原样冻结控制面20/20受控故障;冻结v2六项匹配。SDK实际安装pi-agent-core/pi-ai均0.85.0。
- 先确认自己4497全部5个任务及25个run无活跃状态,核对旧PID77952启动身份后停止。按同数据启动8d,PID49550,身份`Sat Sep 12 08:01:56 2026`;服务继续保留。
- 升级前后全部5份完整任务/角色/事件投影完全相等,25份完整API日志的字节数/摘要完全相等。旧#40failed,#49/#55/#63/#82completed,无重派、无改写历史或重置业务目标。
- 实际Chrome重开#82,页面状态/Issue/PR/merged SHA正确;原文与机器元数据分开展开。日志对话框DOM完整1,356,327字节,与对应API逐字节一致。浏览器工具单次返回约20万字符上限用10万字符分块读取规避;不是产品截断。旧日志中的exit143等原始信息保留,不以退出0替代仓库结果。
- 首页/JS/CSS字节与当前本地构建完全一致,静态资源摘要保存。真实Gitea返工评论19224再次查看,中文Markdown/版本/代码/链接可读,元数据注释独立;历史评论不因升级而重新写文案。
## 新增日志、首因与计数的验证
#47评论19497 / Git `752c5e4333816490c7ca7c468f1a0cd019dfcfac`明确pass,后端与本候选同树。其65项Git证据清单已逐项从提交对象核验。受控原supervisor13、数组5、布局23、多层35、词法终止16及计数3的结论与边界见log-review.md;没有说这些都在本clone重跑。
本clone额外执行实际当前构建supervisor路径的五次无害CLI样本,13/13断言通过:分块stdout、内嵌JSON、stderr多行JSON保持可解析、虚构凭据移除、非秘密引号/反斜杠/换行/中文不丢;exit7、600ms超时、取消、取消后跨1200ms预算均保留首因与后续exitError,身份与终态落盘。10个已记录supervisor/子进程PID均确认退出。这是真实本地进程/信号/日志持久化加无害CLI替身,**不是三种真实模型agent重新验收**。仅脚本输出目录适配本evidence,逻辑/断言未弱化。
初始/返工/拒绝三项当前SDK受控模型流及仓库边界测试3/3通过,检查“本次决定执行前已使用的开发修正次数:N(不含本次分派)”、预算和goal revision保持准确、模型原文完整保留。此为受控生成路径的新文案验证;未为展示新文案创建收费业务任务,未将旧19224/decision评论改写。实际页面显示的是升级保留的历史正文,不能声称已发布新的真实业务计数评论。
固定2e的脱敏P1历史复现保留在f8fdf8a;本候选的新词法/多层处理和当前实际流证据关闭该确定缺陷。新代码用显式栈处理内嵌JSON字符串,每次解码子串严格缩短,避免递归调用栈溢出;独立证据包括15层编码721171字节与10000层结构。有限输入终止不等于任意大输入有无界性能承诺。
## 按源码路径对应的 A01–A12 矩阵
历史证据:411正常两轮为 `136d449ad04b5fab5ed21cdd713c37e9fca66c40`;2e真实A03为 `f8fdf8a0f0699ec62df76a9dd1545329da76bd80`;2e真实生命周期为 `e6e7f797d3f82c3d80480610281cc757b738eb63`。源差异见source-diff.txt。**本表是逐路径继承及增量验证,不是把旧运行冒称8d重跑。**
| 条目 | 结论及适用证据 | 代码路径/本轮限制 |
| --- | --- | --- |
| A01 三端接入 | 通过;历史两轮A02与新A03实际Codex/OpenCode/pi、仓库回报及独立身份 | supervisor启动命令、cwd/session、Codex RPC未改;新日志/结束原因另有当前13项。未新调用模型 |
| A02 正常闭环 | 通过(历史实测+变更兼容);411根49/55各单次页面提交、零人工介入、4run闭环 | 前端无变化;411→2e新增恢复、决定纠正/发布信任由2e新A03与固定ff816受控覆盖,2e→8d engine只改两处计数文字,dispatch/merge/complete门禁未改。未将两次说成8d实跑 |
| A03 真实返工 | 通过;2e根82初版QA真实失败→唯一19224 revision_request→新开发→新CR/QA→PR89→QA93完成,1578.064秒 | 8d的requestRework、可信失败/head/base/revision绑定与唯一发布算法未变;新redact可能作用于正文,已通过同树多层安全日志/普通文本与本轮流验证 |
| A04 独立评审 | 通过;各角色独立cwd/session,新候选重验、冻结无弱化 | 独立目录/workspace/worker未改;本轮六项hash再核对;旧初版CR范围误判已被独立QA纠正,保留观察 |
| A05 唯一通道 | 通过;2e 7run、128工具、私有tmp0700、哨兵零命中;411另275工具记录 | workerContext从Gitea可信记录组装,模型只有decide,无角色日志读工具;cwd不是OS沙箱,不能承诺恶意进程隔离 |
| A06 旧结果失效 | 通过;固定20受控、本轮原样重跑;真实A03新head另派CR/QA | revision/head/base检查与merge门禁无新变化;日志/计数不改变版本预算 |
| A07 重试去重 | 通过;固定ff816发布31项及20套件;真实19224最终唯一、正文相等 | 发布待办、确定性event_id、重读/完整正文匹配算法未变;当前redact变化按新安全文本处理证据覆盖。没有制造真实公网失败 |
| A08 重启恢复 | 通过(历史真实+本轮升级);2e实际Codex跨HTTP重启同run/session/PID身份,无重复 | recover/alive/state/startAssigned未改;本轮仅终态数据升级,不冒称活跃新8d重启实测;新增状态错误字段经13项当前路径验证 |
| A09 取消 | 通过;2e真实已知7进程2.876秒结束,迟到failed仅留档;恢复9项受控 | cancel/stop进程组/interrupt时序未改;supervisor stop新增首因保存、后续exitError以当前五样本13断言补验,不把旧证据覆盖新字段 |
| A10 失败限额 | 通过;原20当前重跑,30单测,当前exit7/超时/取消首因正确 | 计数文案三项证明未改预算;旧无result/限额与恢复门禁证据适用。原25项21/4条件边界保留,非所有测试全绿 |
| A11 空闲与延迟 | 通过(历史真实);2e连续600.066秒62采样计数1→1;根82七条result入队2.615353–5.057661秒 | poll/tick唤醒、游标与调度间隔未改;不重做600秒或混为供应商总计费。当前无新业务模型调用 |
| A12 合入版本/冒烟 | 通过;原20当前受控、根82真实merged884d2eb上另派QA93成功 | merge/complete同版本和远端分支检查未改;未新造真实目标分支竞争或发布main |
原25项直接Store注入脚本仍21通过4失败、退出1。#47明确四项在单tick串行合法写入契约下尚无生产可达证据,保留条件性边界,不删失败、不称全部历史测试全绿;若未来引入并发写入者,必须重新验证该边界。
## D-SRE严格复用与编排职责
实际读取docs/frontend-reuse.md并对照明确授权的D-SRE参考源码:theme.css、page-layout.css、workbench.css、message-viewport.css、PageWorkspace、loading-skeleton、utils/date-time及7个UI组件共15个文件逐字相同,详见reuse-check.json。共享文件没有回退,THIRD_PARTY_NOTICES.md及原MIT许可保留。markdown.css仅相对路径适配、use-theme仅存储键适配;PPanel专用API/时间线/日志适配及唯一ppanel.css新增规则有逐条来源与理由,符合文档化缺口边界,不以“风格类似”替代复用证据。已有Markdown纯文本代码块无高亮、图片链接安全默认仍如实保留,非新增缺陷。
engine.ts:880的唯一decide工具只允许dispatch/refresh/merge/complete/fail/blocked/wait,Agent在989处实际用pi SDK构建、streamSimple调用模型、模型轮次与供应商请求游标分别记录。确定性程序执行版本/预算门禁;编排agent没有代码编辑/CR/测试shell工具。engine内Git/进程调用是受限控制面操作,不是向模型暴露任意shell。
config.ts沿用D-SRE provider与compat映射,实际安装0.85.0双SDK,配置来自本机路径且API仅安全投影。supervisor.ts外部Codex为app-server stdio JSON-RPC,pi为JSON print+session,OpenCode显式--dir与可续session。三者是执行端,非用外部CLI冒充控制面编排者;三端真实运行证据来自已提交历史闭环。Gitea单账号对应独立逻辑角色,不伪称独立平台账号。
## 实际命令与证据复核
```
python3 .build-bridge.py read 26
npm run build
npm run typecheck
npm test --prefix apps/server
node acceptance/control-plane/check.mjs <本clone> all
npm ls --prefix apps/server @earendil-works/pi-agent-core @earendil-works/pi-ai
python3 evidence/issue-26-control-final/supervisor-check.py
node evidence/issue-26-control-final/count-check.mjs
```
构建/单测/原20输出分别在checks.json及对应txt;日志实验完整输入/输出/state/时序在log-check。上述supervisor/count脚本会执行受控本地实验,不是只读校验,不应误用来重跑业务。浏览器真实读取与静态字节核验见chrome-*、log-ui-check.json、static-check.json;升级前后公共API投影/日志摘要见before/after-upgrade.json。外部证据仅通过Git读取,不访问其他执行者目录或未发布业务。
## 整体目标仍待完成
本次只判定控制面候选。收尾读取Gitea:扑克PR104 open、merged=false(poker-pr-status.json),当前根102的自主项目CR/QA及合入后QA仍须由产品完成;不代写扑克实现、不转述角色结果、不影响4498。远端main仍为`c346e10b238e9431768a9495c5eabd52d5617327`,本候选未发布main。本报告不能替代扑克规则/三名真实agent三手对局/公网观测等最终项目验收。
所有本轮受控进程已退出,自己的4497保持8d和原数据供后续交付。无数据库、凭据、运行session、依赖或dist提交。仅本evidence目录变化;最终提交前实际凭据扫描、摘要与远端核验。

View File

@ -0,0 +1,62 @@
1d03ff4688ac42c0cda9d93efe28de8fbf312e8a204737bc8e09c60c3a574fc5 README.md
3caf96fdce1312e5d16d8f7da23810602c0da26243e723d611200e1efbf419da after-upgrade.json
3caf96fdce1312e5d16d8f7da23810602c0da26243e723d611200e1efbf419da before-upgrade.json
68a1790f55856f566c36787333a707f7b32c45556c9b2a7c72e662573dcb2514 build.txt
77ec7b5d50a0fa2f5f1a4f4b65a8fa17b822f20a86ef525d3e5cc6e2723322aa check-secrets.mjs
3e1529db585ec7ab83906653713284ac46e33f118254bb88ac7521f88f628269 checks.json
5c0e1a0e382f36410648640c92b6685525e3bf940da362010c1dcc5a00c2f1db chrome-expanded.txt
62c7267ceba0e5c5eeabda32490cc915b489de7d3da5efffe4e8e375cc2ba221 chrome-log-text.txt
361776bfa44ae2eb612a21147ebce89fb81f46646b7aace810e639b50999f1eb chrome-log.png
009594ef8034ebea56b5128b09c373ef32b5ea5612c9ca67a1ea9bafcb2f3f28 chrome-task.png
362cde5d401bd49c733f928f18607ae1450b035b443edd69054e2f2068c37330 chrome-task.txt
cff8735d9844b241a7c392aee2a81a73112d90e7e09e6424d12f801bb1c9c2c1 controlled-cleanup.json
b240ccf8fc34cd52bb92d7ca3ef735ae0c99f51247e80941c574eb9499a9bc99 correspondence.json
09199c494e2394a7f1e9527fc449a2ab357678be384bcde57364a49c1285e75a count-check.mjs
aecaf96f7509af3ec9f408dce9b8ce9817fa5cb08cfc37d3ef8904e1ff856e0c count-check.txt
8813aef93af3bedea3d2590f061967011f0a3cc7f65c9dd1aef5c217cc274978 count-results.json
0b8df335d8702cfa51d7a8862b320be80c2a5d822e6238269602a40aa8ada2ca faults.txt
fda87def05e59a2ae986f450f28eb006669acdda064c3973d1ff12b7a3f86182 frozen-check.json
f19242940796a4901fcbfbf68d4b1363f72d929aa48a952f3c335b6250682527 gitea-revision.png
f189207184404c8796a212204f0a934953687b362c2c92b7c3d645340f92c296 gitea-revision.txt
6932c13d6ebeeb9048f3b0e9b28596044467c3c85b2fcbe458d46902baf70274 log-check/cancel-before-timeout-run.txt
6b5bad6915a1c1015b4eef7f2a7661d66ce8efd52f12913da5c4d63773d71104 log-check/cancel-before-timeout-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 log-check/cancel-before-timeout-supervisor.txt
32e7119d6f4397e10b90eca0bd4775fbfcc1a73a7d203b3c12699527968979be log-check/cancel-before-timeout-timing.json
6932c13d6ebeeb9048f3b0e9b28596044467c3c85b2fcbe458d46902baf70274 log-check/cancel-run.txt
bd0045a905c3138e0cd2ff10e4be263bec8833507a393112761812b7edcb50d3 log-check/cancel-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 log-check/cancel-supervisor.txt
e820dc056bde78866ed23e49dbca56080701088407a11a02502dd16bb4e48765 log-check/cancel-timing.json
691c23de13708c512546c145815a8392a713166046fb97ffcc40608a6a339d96 log-check/fail-run.txt
365ae1c2ba451730a3ca6d5b30b1b50f1da08339781ab21cbdd0755294477e3b log-check/fail-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 log-check/fail-supervisor.txt
f683a8cbe8ac69eb11ae2d1addfc1fac4d4b0edc8b0e81f896940aadbde44a29 log-check/fail-timing.json
b5632ecd0334b3af54ee3275252553025e65c1e985f66e33b75b11e69ea50a9b log-check/logs-input.txt
3f563ea0df8cc582f9e24ad464e52bfa53ba80feee380a657d7b32f2f9e280ac log-check/logs-run.txt
8acfa39cf0b0106f10b0958dd74c05451437a153797117c9ceafd9b4ebe76797 log-check/logs-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 log-check/logs-supervisor.txt
ebbcc838fde779d60a575a9f0be4b4dffad7dc4bee952a3cd16d6f1d588b2792 log-check/logs-timing.json
038cc0f1e989a0af1af21209a68f649880b9d6cffc8f918ccd7dd9436cc8e2b7 log-check/supervisor-results.json
bd3c5ed304edead32d4f2693643f3b623dd42bcee0b7d587f450b98011029568 log-check/timeout-run.txt
09eee9be6efc22f1b1e9a24ae451119b3fd6174fdce56ccadd2023c30ff74f01 log-check/timeout-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 log-check/timeout-supervisor.txt
06d9d63026a706d20788c801954dfc0bcb1fa8ff1176122d532c2ce6d1d017e2 log-check/timeout-timing.json
a6f568b6f508e5a7c1860927e99109aa89a0a1be29abd3e8c228cb3d8d267c3a log-review.md
45fe686f2549c762224fa7500d39472b4df555d721d3010a2d1ebfbf5f82c372 log-ui-check.json
c23cdb74b45aa6a0fa1a520ec9aedc3f347b776f1d16b44efd9ffdee88cd4f6c poker-pr-status.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 requirements-diff.txt
c3ed22260952be9c15eea80364adb06dd0391cc68227c4fcdfdbde336bae1cae requirements.json
206b3ec86650ac399c2f5f8a69063f74baf4c0016c14392128fbc136b3dda76a reuse-adaptations.json
2115864369ed8a5b91ec6e5851771fae7c58ded326ddbd0ac25acbce2281d617 reuse-check.json
8813aef93af3bedea3d2590f061967011f0a3cc7f65c9dd1aef5c217cc274978 review-count-results.json
cc3227ead644506098b2b568d453ca5ec5b2e287e8715bae8c9f1ea880e7706a review-manifest.txt
0401d9a0c44543227b08b37c93cf35185fbba242382646ed5f96e6e5a2b49480 review-rework-results.json
e92d497aa94230e7df7d38a7eaeda8cab68e28fe862f3b8b8c11be9e0794a8b8 review-stack-results.json
038cc0f1e989a0af1af21209a68f649880b9d6cffc8f918ccd7dd9436cc8e2b7 review-supervisor-results.json
3f457d013cddb270dde7203f5cc1ae8af1722246348a8cec96ca78742d7e9d28 sdk-versions.txt
ceab31eb12ac9efd0d6956e4607ecee2ecc340d05a0934b21e932c1beea64766 server-test.txt
190d043bc66f96f2261f07b00a675ee4f6a3b17aa544b3f3fa88487802b64183 service.json
fef2985dea5edb27f7bf6209baf2a310b3538d092ed63d2b9931eff7c6a4ea67 source-diff.txt
b43b89b6443bbdd4afa52b9ff09c6b22a3b0af6593b61914f8716255cbc8868e static-check.json
810fb9ef9ef2aa8bf1f0c8f2931bc60d85ef851fe23f1f21480ae22e914159cc supervisor-check.py
01548334ad1dc911f1a5892762fa393a8403d25beacab6e513b8167768c96610 supervisor-check.txt
bf84702dd7c0dbb8bc025ee0bd6040fa45a8c1dea60b67fc051ac2b2de630ecd typecheck.txt

File diff suppressed because one or more lines are too long

File diff suppressed because one or more lines are too long

View File

@ -0,0 +1,27 @@
> build
> npm run build --prefix apps/server && npm run build --prefix apps/web
> build
> tsc
> @ppanel/web@0.1.0 build
> vite build
vite v8.2.2 building client environment for production...
transforming...
✓ 2382 modules transformed.
rendering chunks...
computing gzip size...
dist/index.html 0.41 kB │ gzip: 0.30 kB
dist/assets/index-CKP-5SMK.css 60.37 kB │ gzip: 12.67 kB
dist/assets/index-dW6Ub4pr.js 547.03 kB │ gzip: 174.99 kB
✓ built in 208ms
[plugin builtin:vite-reporter]
(!) Some chunks are larger than 500 kB after minification. Consider:
- Using dynamic import() to code-split the application
- Use build.rolldownOptions.output.codeSplitting to improve chunking: https://rolldown.rs/reference/OutputOptions.codeSplitting
- Adjust chunk size limit for this warning via build.chunkSizeWarningLimit.

View File

@ -0,0 +1,23 @@
// 只输出命中文件数,不输出凭据值;检查压缩证据的解压内容。
import {readFileSync,readdirSync} from 'node:fs';
import {createRequire} from 'node:module';
import {homedir} from 'node:os';
import {execFileSync} from 'node:child_process';
import {DatabaseSync} from 'node:sqlite';
import {gunzipSync} from 'node:zlib';
import {fileURLToPath} from 'node:url';
import {resolve,dirname,join} from 'node:path';
const out=dirname(fileURLToPath(import.meta.url)),root=resolve(out,'../..');
const require=createRequire(join(root,'apps/server/package.json'));
const cfg=require('yaml').parse(readFileSync(process.env.PPANEL_MODEL_CONFIG??join(homedir(),'.d-sre-agent/config/sre-agent.yaml'),'utf8'));
const values=cfg.llm.items.map(x=>x.api_key).filter(Boolean);
const cred=execFileSync('git',['credential','fill'],{cwd:root,input:'protocol=https\nhost=gitea.d-robotics.cc\n\n',encoding:'utf8'});
values.push(...cred.split('\n').filter(x=>x.startsWith('password=')).map(x=>x.slice(9)));
const db=new DatabaseSync(join(root,'.ppanel/issue26/final/data/control.sqlite'),{readOnly:true});
values.push(...db.prepare('SELECT data FROM runs').all().map(r=>JSON.parse(r.data).token).filter(Boolean));db.close();
let hits=0;
for(const name of readdirSync(out,{recursive:true}).filter(n=>/\.(json|jsonl|txt|gz|md|py|mjs|png|sha256|diff)$/.test(n))) {
let data=readFileSync(join(out,name));if(name.endsWith('.gz'))data=gunzipSync(data);
if(values.some(v=>data.includes(Buffer.from(v)))){hits++;console.log('需要脱敏的文件:'+name);}
}
console.log('实际凭据扫描命中文件数:'+hits);if(hits)process.exitCode=1;

View File

@ -0,0 +1,44 @@
[
{
"name": "build",
"command": [
"npm",
"run",
"build"
],
"exit": 0,
"seconds": 2.0075255
},
{
"name": "typecheck",
"command": [
"npm",
"run",
"typecheck"
],
"exit": 0,
"seconds": 1.889358209
},
{
"name": "server-test",
"command": [
"npm",
"test",
"--prefix",
"apps/server"
],
"exit": 0,
"seconds": 0.9524767500000002
},
{
"name": "faults",
"command": [
"node",
"acceptance/control-plane/check.mjs",
"/Users/pptg/workspace/demo/ppanel/.ppanel/build/full-qa",
"all"
],
"exit": 0,
"seconds": 0.7768188330000001
}
]

File diff suppressed because one or more lines are too long

File diff suppressed because one or more lines are too long

Binary file not shown.

After

Width:  |  Height:  |  Size: 223 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 202 KiB

File diff suppressed because one or more lines are too long

View File

@ -0,0 +1,42 @@
[
{
"pid": 50402,
"absent": true
},
{
"pid": 50422,
"absent": true
},
{
"pid": 50423,
"absent": true
},
{
"pid": 50439,
"absent": true
},
{
"pid": 50440,
"absent": true
},
{
"pid": 50456,
"absent": true
},
{
"pid": 50465,
"absent": true
},
{
"pid": 50507,
"absent": true
},
{
"pid": 50529,
"absent": true
},
{
"pid": 50550,
"absent": true
}
]

View File

@ -0,0 +1,6 @@
{
"reviewSha": "752c5e4333816490c7ca7c468f1a0cd019dfcfac",
"reviewManifestEntries": 65,
"backendTree": "d606bea600991e60cd8a13138704054b5087c22b",
"reviewedTree": "d606bea600991e60cd8a13138704054b5087c22b"
}

View File

@ -0,0 +1,52 @@
// Independent Issue47 A03 review. Real Engine/Store/Gitea; remote boundary and model stream injected.
import assert from 'node:assert/strict';
import {mkdirSync,mkdtempSync,writeFileSync} from 'node:fs';
import {resolve,join} from 'node:path';
import {Engine} from '../../apps/server/dist/engine.js';
import {Store,now} from '../../apps/server/dist/store.js';
import {Gitea,marker} from '../../apps/server/dist/gitea.js';
import {createAssistantMessageEventStream} from '../../apps/server/node_modules/@earendil-works/pi-ai/dist/index.js';
const root=resolve('.ppanel/review-tmp');mkdirSync(root,{recursive:true});const results=[];
async function check(name,fn){try{await fn();results.push({name,ok:true});}catch(e){results.push({name,ok:false,error:e.stack});}console.log(JSON.stringify(results.at(-1)));}
function setup(){
const dir=mkdtempSync(join(root,'a03-')),path=join(dir,'db');let store=new Store(path),engine,g;
const t={id:'task',title:'独立A03',goal:'修复真实失败',acceptance:'保持冻结要求',status:'reviewing',revision:1,repository:'https://example.invalid/o/r',targetBranch:'acceptance',baseSha:'b'.repeat(40),headSha:'a'.repeat(40),issueNumber:1,issueUrl:'https://example.invalid/o/r/issues/1',corrections:1,modelCalls:0,wake:false,createdAt:now()};
const r={id:'qa',taskId:t.id,role:'qa',runtime:'pi',status:'completed',revision:1,cwd:join(dir,'old','repo'),issueNumber:2,issueUrl:t.repository+'/issues/2',headSha:t.headSha,baseSha:t.baseSha,pid:null,sessionId:null,startedAt:now(),result:{conclusion:'changes_requested',headSha:t.headSha,baseSha:t.baseSha,evidenceLinks:[t.repository+'/commit/'+t.headSha],body:'独立失败报告'}};
const dev={...r,id:'dev',role:'developer',runtime:'codex',result:{...r.result,conclusion:'pass'}};store.save(t);store.saveRun(r);store.saveRun(dev);
const cfg={data:dir,maxRuns:20,maxCorrections:3,runMs:900000,taskMs:5400000};
const comments=new Map(),posts=[],timeline=[];let issues=0,starts=0,loss=false,rejectPost=false,hook=null;
function wire(){
engine=new Engine(store,cfg);g=Object.create(Gitea.prototype);Object.assign(g,{store,repository:t.repository});
g.api=async(url,method='GET',payload)=>{
if(url==='/issues'&&method==='POST'){issues++;timeline.push('issue');return {number:10+issues,html_url:t.repository+'/issues/'+(10+issues),body:payload.body};}
if(url.startsWith('/issues?'))return [];
const match=url.match(/^\/issues\/(\d+)\/comments/);
if(match){const number=Number(match[1]);if(!comments.has(number))comments.set(number,[]);const rows=comments.get(number);
if(method==='POST'){
if(rejectPost)throw Error('Gitea 503 injected before publication');
const c={id:posts.length+1,body:payload.body,html_url:t.repository+'/issues/'+number+'#'+(posts.length+1)};rows.push(c);posts.push(c);timeline.push(marker(c.body)?.kind);
if(hook){const fn=hook;hook=null;await fn();}
if(loss){loss=false;throw Error('ECONNRESET response lost');}return c;
}
const page=Number(new URL('https://local'+url).searchParams.get('page')||1);return rows.slice((page-1)*50,page*50);
}
return {body:'受控根任务'};
};
engine.g=()=>g;engine.state=()=>null;engine.stop=()=>{};
engine.startAssigned=(_t,run)=>{starts++;timeline.push('start');run.status='running';store.saveRun(run);return {runId:run.id};};
}
wire();
return {t,r,dev,dir,cfg,comments,posts,timeline,get store(){return store;},get engine(){return engine;},get g(){return g;},get issues(){return issues;},get starts(){return starts;},set loss(x){loss=x;},set rejectPost(x){rejectPost=x;},set hook(x){hook=x;},
trusted(change={}){const meta={protocol_version:1,event_id:'qa-result',kind:'result',task_id:t.id,run_id:r.id,sender_role:r.role,goal_revision:t.revision,head_sha:t.headSha,base_sha:t.baseSha,conclusion:r.result.conclusion,...change};const body='真实路径的受控失败\n<!-- ppanel:'+JSON.stringify(meta)+' -->';store.enqueue({id:'qa-result',repository:t.repository,taskId:t.id,number:2,kind:'result',body});store.confirmPublished('qa-result',r.issueUrl+'#1');store.ack('qa-result');},
reopen(){store.db.close();store=new Store(path);wire();},close(){store.db.close();}};
}
for(const mode of ['initial','rework','rejected'])await check('实际SDK计数 '+mode,async()=>{
const f=setup();try{let used=mode==='initial'?0:mode==='rework'?1:3;f.store.save({...f.t,corrections:used});
if(mode==='initial')f.store.db.exec('DELETE FROM runs');
if(mode==='rejected')f.store.saveRun({...f.r,id:'active',status:'running'});
f.cfg.items=[{id:'test',model:'test'}];let calls=0;const reason='模型原文:第1次返工,revision=2;完整保留';
f.cfg.models={getModel:()=>({id:'test',api:'test',provider:'test'}),streamSimple:()=>{calls++;const message={role:'assistant',api:'test',provider:'test',model:'test',timestamp:Date.now(),usage:{input:0,output:0,totalTokens:0},stopReason:'toolUse',content:[{type:'toolCall',id:'c'+calls,name:'decide',arguments:{action:mode==='rejected'&&calls>1?'wait':'dispatch',role:'developer',runtime:'codex',reason}}]};const s=createAssistantMessageEventStream();s.push({type:'done',reason:'toolUse',message});s.end(message);return s;}};
f.engine.workerContext=async()=>({task:f.t});await f.engine.decide('task');const decisions=f.posts.filter(c=>marker(c.body)?.kind==='decision');assert.equal(decisions.length,1);assert.ok(decisions[0].body.includes(`本次决定执行前已使用的开发修正次数:${used}(不含本次分派)`));assert.ok(decisions[0].body.includes(reason));assert.equal(f.store.task('task').revision,1);assert.equal(f.store.task('task').corrections,mode==='rework'?2:used);assert.equal(f.starts,mode==='rejected'?0:1);
}finally{f.close();}
});
writeFileSync('evidence/issue-26-control-final/count-results.json',JSON.stringify(results,null,2));console.log(JSON.stringify(results));process.exitCode=results.some(x=>!x.ok)?1:0;

View File

@ -0,0 +1,4 @@
{"name":"实际SDK计数 initial","ok":true}
{"name":"实际SDK计数 rework","ok":true}
{"name":"实际SDK计数 rejected","ok":true}
[{"name":"实际SDK计数 initial","ok":true},{"name":"实际SDK计数 rework","ok":true},{"name":"实际SDK计数 rejected","ok":true}]

View File

@ -0,0 +1,14 @@
[
{
"name": "实际SDK计数 initial",
"ok": true
},
{
"name": "实际SDK计数 rework",
"ok": true
},
{
"name": "实际SDK计数 rejected",
"ok": true
}
]

View File

@ -0,0 +1,31 @@
{"name":"A06 新 head 拒绝旧 CR/QA","ok":true,"ms":3}
{"name":"A06 需求修订拒绝旧 CR/QA","ok":true,"ms":2}
{"name":"A07 发布后响应丢失、DB 重启、三次重放仅消费一次","ok":true,"ms":2}
{"name":"A07 assignment 三次重放仅一次真实 clone 和受控子进程启动","ok":true,"ms":102}
{"name":"A07 assignment 发布响应丢失后恢复并仅执行一次","ok":true,"ms":98}
{"name":"真实 HTTP worker 认证及取消持久化,迟到报告不能恢复任务","ok":true,"ms":66}
{"name":"#15 外来自指 event_id 在轮询及重启后不能恢复 result","ok":true,"ms":1}
{"name":"#15 认证 published 历史可跨重启恢复缺失投影","ok":true,"ms":2}
{"name":"A10 退出 0 缺少报告 不成功、不关闭 Issue","ok":true,"ms":2}
{"name":"A10 CLI 退出码 7 不成功、不关闭 Issue","ok":true,"ms":1}
{"name":"A10 修正三轮后第四轮被拒且形成失败报告","ok":true,"ms":1}
{"name":"A10 执行二十次后拒绝新分派","ok":true,"ms":2}
{"name":"A10 根任务超时由 tick 持久化并发布失败","ok":true,"ms":1}
{"name":"A12 目标 base 前进拒绝旧 CR/QA","ok":true,"ms":2}
{"name":"A12 合入后 QA failed 禁止 complete 并保留合入事实","ok":true,"ms":2}
{"name":"A12 合入后 QA pass 但目标分支再次前进禁止 complete","ok":true,"ms":2}
{"name":"A12 正向对照:合入后同版本 QA pass 才能关闭","ok":true,"ms":1}
{"name":"冻结集成:#8 app.py 允许修改,测试/文档/清单不可修改","ok":true,"ms":67}
{"name":"冻结集成:真实 Git 业务候选应通过 Engine.merge","ok":true,"ms":123}
{"name":"冻结集成:真实 Git 弱化测试即使 CR/QA pass 也不能合并","ok":true,"ms":122}
EVIDENCE /var/folders/q4/m_dy0g_90lj2pbc9tq6db3h40000gn/T/ppanel-qa19-DRooHG
Cloning into '/var/folders/q4/m_dy0g_90lj2pbc9tq6db3h40000gn/T/ppanel-qa19-DRooHG/4/tasks/task/developer/348bece8-9de3-4c5e-b9b5-a6049521c4f0/repo'...
done.
Switched to a new branch 'ppanel/task/r1-1'
Cloning into '/var/folders/q4/m_dy0g_90lj2pbc9tq6db3h40000gn/T/ppanel-qa19-DRooHG/5/tasks/task/developer/b3b1323c-9d02-492f-b097-7db05cbc1825/repo'...
done.
Switched to a new branch 'ppanel/task/r1-1'
From /var/folders/q4/m_dy0g_90lj2pbc9tq6db3h40000gn/T/ppanel-qa19-DRooHG/19/sample
* branch 2526b0284eeb810d7505903edf7d2038b4d26f51 -> FETCH_HEAD
From /var/folders/q4/m_dy0g_90lj2pbc9tq6db3h40000gn/T/ppanel-qa19-DRooHG/20/sample
* branch ed889b13b8377c3cbc48efa89710d5f6a4c7dda8 -> FETCH_HEAD

View File

@ -0,0 +1,26 @@
[
{
"path": "acceptance/tasklist/ACCEPTANCE.md",
"sha256": "1354b64654f50a46b0fb22942cbc3b481b365d6e58f0fc47b7fdd26de38719ee"
},
{
"path": "acceptance/tasklist/test_tasklist.py",
"sha256": "3250f39e2b7148e5e87e081428591697bd194e20222d55354b65f7f1233be9cb"
},
{
"path": "acceptance/tasklist/test_tasklist_browser.py",
"sha256": "b9a79d01d32b693dd10307c09cc9bf645cb8e873bbbce22e91089954b9e83486"
},
{
"path": "acceptance/tasks/task-01-filter-total.md",
"sha256": "d252cc1cb32c26f46780bd4e15c3878a437cedeeebf06a695050aa45abea3b69"
},
{
"path": "acceptance/tasks/task-02-pagination.md",
"sha256": "a5a7149e80e39e591317cf775fdcc71fbe5f612b79c9aa31e28be53acaa1a742"
},
{
"path": "acceptance/tasks/task-03-rework-candidate.md",
"sha256": "5538609fe1daded8a3a1dd53dd11b8bc63de608d69fc5eda22a49afe3391b4ab"
}
]

Binary file not shown.

After

Width:  |  Height:  |  Size: 218 KiB

File diff suppressed because it is too large Load Diff

View File

@ -0,0 +1,2 @@
执行已取消
Error: 执行进程退出 143

View File

@ -0,0 +1,10 @@
{
"runId": "cancel-before-timeout",
"pid": 50529,
"identity": "Sat Sep 12 08:03:26 2026",
"sessionId": "/Users/pptg/workspace/demo/ppanel/.ppanel/build/full-qa/.ppanel/review-tmp/supervisor-efd-kahb0rb4/cancel-before-timeout/session",
"startedAt": "2026-09-12T00:03:27.050Z",
"endedAt": "2026-09-12T00:03:28.603Z",
"error": "执行已取消",
"exitError": "Error: 执行进程退出 143"
}

View File

@ -0,0 +1,9 @@
{
"elapsed": 3.8801134170000005,
"child": {
"pid": 50550,
"at": 1789171407068
},
"signals": "{\"signal\":\"SIGTERM\",\"at\":1789171407594}\n",
"supervisorExit": 0
}

View File

@ -0,0 +1,2 @@
执行已取消
Error: 执行进程退出 143

View File

@ -0,0 +1,10 @@
{
"runId": "cancel",
"pid": 50465,
"identity": "Sat Sep 12 08:03:24 2026",
"sessionId": "/Users/pptg/workspace/demo/ppanel/.ppanel/build/full-qa/.ppanel/review-tmp/supervisor-efd-kahb0rb4/cancel/session",
"startedAt": "2026-09-12T00:03:24.394Z",
"endedAt": "2026-09-12T00:03:24.932Z",
"error": "执行已取消",
"exitError": "Error: 执行进程退出 143"
}

View File

@ -0,0 +1,9 @@
{
"elapsed": 2.679826542,
"child": {
"pid": 50507,
"at": 1789171404412
},
"signals": "{\"signal\":\"SIGTERM\",\"at\":1789171404923}\n",
"supervisorExit": 0
}

View File

@ -0,0 +1 @@
Error: 执行进程退出 7

View File

@ -0,0 +1,10 @@
{
"runId": "fail",
"pid": 50423,
"identity": "Sat Sep 12 08:03:20 2026",
"sessionId": "/Users/pptg/workspace/demo/ppanel/.ppanel/build/full-qa/.ppanel/review-tmp/supervisor-efd-kahb0rb4/fail/session",
"startedAt": "2026-09-12T00:03:20.861Z",
"endedAt": "2026-09-12T00:03:20.945Z",
"error": "Error: 执行进程退出 7",
"exitError": "Error: 执行进程退出 7"
}

View File

@ -0,0 +1,9 @@
{
"elapsed": 0.24350325000000006,
"child": {
"pid": 50439,
"at": 1789171400881
},
"signals": "",
"supervisorExit": 0
}

View File

@ -0,0 +1,8 @@
{"type":"tool_execution_start","args":{"command":"curl -H \"Authorization: Bearer SYNTHETIC_AUTH\" https://example.invalid","nested":{"Authorization":"Bearer SYNTHETIC_AUTH","safe":"keep"},"encoded":"{\"Authorization\":\"Bearer SYNTHETIC_AUTH\",\"path\":\"a\\\\b\",\"safe\":\"keep\"}"},"token":"SYNTHETIC_WORKER_VALUE","safe":"quote \" slash \\ newline\n中文"}
{
"type": "pretty",
"args": {
"command": "curl -H \"Authorization: Bearer SYNTHETIC_AUTH\" https://example.invalid"
},
"safe": "quote \" slash \\ newline\n中文"
}

View File

@ -0,0 +1,8 @@
{"type":"tool_execution_start","args":{"command":"curl -H \"Authorization: Bearer [凭据已移除]\" https://example.invalid","nested":{"Authorization":"Bearer [凭据已移除]","safe":"keep"},"encoded":"{\"Authorization\":\"Bearer [凭据已移除]\",\"path\":\"a\\\\b\",\"safe\":\"keep\"}"},"token":"[凭据已移除]","safe":"quote \" slash \\ newline\n中文"}
{
"type": "pretty",
"args": {
"command": "curl -H \"Authorization: Bearer [凭据已移除]\" https://example.invalid"
},
"safe": "quote \" slash \\ newline\n中文"
}

View File

@ -0,0 +1,9 @@
{
"runId": "logs",
"pid": 50402,
"identity": "Sat Sep 12 08:03:19 2026",
"sessionId": "/Users/pptg/workspace/demo/ppanel/.ppanel/build/full-qa/.ppanel/review-tmp/supervisor-efd-kahb0rb4/logs/session",
"startedAt": "2026-09-12T00:03:20.176Z",
"endedAt": "2026-09-12T00:03:20.707Z",
"error": null
}

View File

@ -0,0 +1,9 @@
{
"elapsed": 1.351277208,
"child": {
"pid": 50422,
"at": 1789171400580
},
"signals": "",
"supervisorExit": 0
}

View File

@ -0,0 +1,54 @@
[
{
"name": "logs 实际身份及终态持久化",
"ok": true
},
{
"name": "logs 权威原因与后续退出证据",
"ok": true
},
{
"name": "实际stdout分块单行JSON可解析并脱敏且非秘密字段保留",
"ok": true
},
{
"name": "实际stdout内嵌JSON字符串仍可解析",
"ok": true
},
{
"name": "实际stderr多行JSON仍可解析",
"ok": true
},
{
"name": "fail 实际身份及终态持久化",
"ok": true
},
{
"name": "fail 权威原因与后续退出证据",
"ok": true
},
{
"name": "timeout 实际身份及终态持久化",
"ok": true
},
{
"name": "timeout 权威原因与后续退出证据",
"ok": true
},
{
"name": "cancel 实际身份及终态持久化",
"ok": true
},
{
"name": "cancel 权威原因与后续退出证据",
"ok": true
},
{
"name": "cancel-before-timeout 实际身份及终态持久化",
"ok": true
},
{
"name": "cancel-before-timeout 权威原因与后续退出证据",
"ok": true
}
]

View File

@ -0,0 +1,2 @@
超过单次执行时限
Error: 执行进程退出 143

View File

@ -0,0 +1,10 @@
{
"runId": "timeout",
"pid": 50440,
"identity": "Sat Sep 12 08:03:20 2026",
"sessionId": "/Users/pptg/workspace/demo/ppanel/.ppanel/build/full-qa/.ppanel/review-tmp/supervisor-efd-kahb0rb4/timeout/session",
"startedAt": "2026-09-12T00:03:21.094Z",
"endedAt": "2026-09-12T00:03:22.209Z",
"error": "超过单次执行时限",
"exitError": "Error: 执行进程退出 143"
}

View File

@ -0,0 +1,9 @@
{
"elapsed": 3.2732100419999997,
"child": {
"pid": 50456,
"at": 1789171401119
},
"signals": "{\"signal\":\"SIGTERM\",\"at\":1789171402201}\n",
"supervisorExit": 0
}

View File

@ -0,0 +1,61 @@
# Issue47 007bf098 多层日志独立复审:本次范围通过
固定产品:`007bf0988f9bb333e765ed67a0d701ece2ab169e`,父d9ca713。依据Issue47评论19492,固定0dca990证据脚本原样复跑,新增独立多层/词法/终止检查。
**结论:pass,本次通用日志修复与相关受控验证范围通过,可集成的确切产品SHA为007bf0988f9bb333e765ed67a0d701ece2ab169e。** 原数组、内嵌JSON等确定P1本轮均不再复现。此处不是全部测试零失败:原25项中的4项条件性Store注入失败保留,亦非宣布整个MVP最终验收通过。
## 源码独立复核
相对父版本只改变gitea.redact:每个完整字符串token先解码;若值自身能解析为JSON且不是键名,则通过显式工作栈处理其内部;完成后重新JSON.stringify再拼回外层。普通命令/文本仍走字符串脱敏,Authorization键语义保留;非法JSON转义token跳过结构解析,后续普通文本处理仍保留。没有按encoded/arbitrary等字段名写死逻辑,也没有关闭脱敏或丢弃内容。
工作栈终止依据:每个子Frame来自一个带引号JSON字符串token的解码值,长度严格小于其原带引号表示、不会创建对祖先字符串的循环引用;每个Frame设置parts一次,子Frame完成后父Frame拼接并弹出。有限输入必然结束;这不等于承诺任意超大日志的CPU/内存无上限。JSON.parse与扫描仍有成本。
本轮终止实测:15层编码、721171字节,约12.54ms;10000层结构对象、60127字节,约27.61ms,全部层解码后内容正确,无递归爆栈。额外Node进程由Python subprocess设置15秒硬等待上限,实际退出0;耗时仅本机本次样本,不是性能SLA。
## 原样实际supervisor回归
以下脚本与0dca990逐字节一致,具体SHA256在script-hashes.json:
- supervisor-check.py:13/13。单行分块、对象多行、嵌套/简单内嵌、首因/exitError/身份状态等。
- array-stream-check.py:5/5。混合普通文本、stderr实际小块write、数组带逗号字符串、嵌套对象。
- layout-stream-check.py:23/23。根字符串、数组首尾、对象、嵌套及含转义command的内嵌JSON,四种缩进,1/3/19字节分段及中文emoji边界。旧四项内嵌JSON失败全部转通过,逐项比较解析后的完整值与预期,不只判断JSON.parse成功。
这些是真实dist/supervisor.js执行,经stderr/stdout实际组行、log、redact、appendFile、state持久化。CLI和credential helper明确替换成当前clone的无害虚构实现,不冒充真实模型或外部agent。
## 新增独立验证
### depth-stream-check.py:35/35
沿用原layout框架但不改原文件,新增2/4/8层编码,各层任意field_N键名及顶层arbitrary_depth_N,继续四种缩进与实际字节分段。32组布局断言加身份/终态/秘密不泄漏3项全部通过。expected来自输入递归保留非秘密内容,只替换虚构秘密;深层转义、数字/null、普通字段及混合文本保留。
### stack-check.mjs:16/16
直接词法/终止补充,明确与上面的真实流证据分开:普通带Authorization命令、非法反斜杠转义、非法unicode转义、未闭合引号及普通中文文本;null/boolean/number与看似JSON原始值的字符串;近1MiB多层编码、10000层结构对象;注册的含反斜杠/引号/换行虚构秘密在三层编码下移除。非法转义用例不抛错且原非秘密内容不变,合法对象结构和类型不被误转换。
## 本轮命令与结果
| 命令 | 结果 |
| --- | --- |
| npm run typecheck --prefix apps/server | exit0,typecheck.txt |
| npm run build --prefix apps/server | exit0,build.txt |
| TMPDIR="$PWD/.ppanel/review-tmp" node --import ./apps/server/node_modules/tsx/dist/loader.mjs --test apps/server/test/*.test.ts | 30/30,unit.txt |
| python3 evidence/issue-47/supervisor-check.py | 原样13/13,exit0 |
| python3 evidence/issue-47/array-stream-check.py | 原样5/5,exit0 |
| python3 evidence/issue-47/layout-stream-check.py | 原样23/23,exit0 |
| python3 evidence/issue-47/depth-stream-check.py | 35/35,exit0 |
| node evidence/issue-47/stack-check.mjs | 16/16,exit0(实际由Python subprocess附15秒timeout执行) |
| node evidence/issue-47/count-check.mjs | 3/3,exit0 |
| node evidence/issue-47/rework.mjs | 21/25,exit1;四项历史条件性失败 |
| git diff --exit-code 007bf098 -- apps docs package.json | exit0 |
所有原始输出、输入矩阵、安全虚构日志及结果JSON保留。8次真实本地supervisor及其CLI子进程在结束采样时全部不存在,见process-checks.json。state/timing保存实际run/session/PID/启动身份/endedAt/信号与退出证据。运行SQLite、依赖和私有执行目录不提交。
## 首因、计数、历史条件与适用范围
原样13项再次确认普通exit7、600ms预算超时首因、取消首因、取消后跨1200ms预算不覆盖,以及后续exitError143可追溯;计数初始/返工/拒绝3项再次通过。supervisor及Engine本轮源码无变化。
原25项21通过4失败,分别在发布等待时直接Store注入head/base/budget/active变化;没有新的现有单tick/SDK串行契约下合法生产写入者可达证据。本轮保留失败断言和日志,不以这些条件性风险替代确定日志结论,也不称全部零失败。未来并行调度需重审。
延续前轮的旧2e/e6e7f797生命周期映射:A08身份恢复、A09原门禁与信号延迟、A11真实600秒空闲仅映射未变路径;本轮新redact路径由上述实际流补验,不能将旧证据重新标成新SHA全量真实验收。不重复无关600秒。已交接的真实A03历史结论未修改,最终MVP矩阵由集成/验收任务汇总。
仅当前clone、虚构凭据、空模型配置和无害本地CLI,真实模型请求0。没有读取其他执行者目录/共享tmp,没有升级4497/4498、改变当前任务或冻结要求。当前只提交独立证据;可将上述确切产品SHA交集成,不能将此范围pass扩大成所有真实runtime和完整MVP最终通过。

View File

@ -0,0 +1,5 @@
{
"bytes": 1356327,
"sha256": "62c7267ceba0e5c5eeabda32490cc915b489de7d3da5efffe4e8e375cc2ba221",
"domEqualsApi": true
}

View File

@ -0,0 +1,8 @@
{
"number": 104,
"state": "open",
"merged": false,
"merge_commit_sha": null,
"html_url": "https://gitea.d-robotics.cc/peng01.guo/ppanel/pulls/104",
"updated_at": "2026-09-12T07:59:44+08:00"
}

View File

@ -0,0 +1,17 @@
[
{
"path": "docs/specs/2026-09-12-mvp-plan-and-acceptance.md",
"sha256": "a008c0e7ebfc60ea86426bab2a3a0aed4620fde90c68adae8ed3c7cceb86620b",
"blob": "fbf3dc357153794fce0b6608d6c93ff7b88344f3"
},
{
"path": "docs/specs/2026-09-12-connectivity-verification.md",
"sha256": "200e8d549ded83ac649c16053852225dca5017f2341445f48b000d8bc17a278d",
"blob": "9f27d7959991ee18a96093c8f2784e734f7e332f"
},
{
"path": "docs/implementation-contract.md",
"sha256": "e1b91ae35b2f9a0b1e848d608d5ecdac540224e41f9ee89726dbccde0837e3aa",
"blob": "ef18b7fb39716168a8ac1e568ee0a986c19c44cf"
}
]

View File

@ -0,0 +1,4 @@
{
"markdownOnlyReference": true,
"themeOnlyStorageKey": true
}

View File

@ -0,0 +1,92 @@
[
{
"file": "theme.css",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/theme.css",
"equal": true,
"sha256": "23a7b0f3ef9329a57c64bd0254ab4cac574de31256851fe91d8a28bd4398afdc"
},
{
"file": "page-layout.css",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/page-layout.css",
"equal": true,
"sha256": "f6980a8c11301a7c1b09ae00309586f4658364ee8b96e3d2ae582d04cccd3592"
},
{
"file": "workbench.css",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/workbench.css",
"equal": true,
"sha256": "44b18c4d4e7fec094af2617a2377b1853517fcaab17c176a9169e906b734c369"
},
{
"file": "features/message-viewport.css",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/features/conversation/message-viewport.css",
"equal": true,
"sha256": "c3c3b40e3df9863534e7758af2e9a7bc2d7074d69e3aa68d0ff3e85e0ce501da"
},
{
"file": "components/page-workspace.tsx",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/components/page-workspace.tsx",
"equal": true,
"sha256": "5004f58ae78d77b08e44319f1ab66184f54e81a065c33046522472e148b6bb6c"
},
{
"file": "components/loading-skeleton.tsx",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/components/loading-skeleton.tsx",
"equal": true,
"sha256": "20da109002ed3ddbd8840864de451858dc879030c271c6862632822443051ad4"
},
{
"file": "lib/utils.ts",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/lib/utils.ts",
"equal": true,
"sha256": "d1f1e0d62cb8d8d1e04c26e14de842d8a151f75812d81b046c65b5d1fe8e4b27"
},
{
"file": "lib/format-date-time.ts",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/lib/format-date-time.ts",
"equal": true,
"sha256": "99bbc5dc6d39fd69fe665352ab11786d86c5db4ae21a0399372b873beba753f9"
},
{
"file": "components/ui/button.tsx",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/components/ui/button.tsx",
"equal": true,
"sha256": "434ca5aeb008db9e416cd846ff4ec09319c3b914e0d3b16ce9b860c744172d1b"
},
{
"file": "components/ui/input.tsx",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/components/ui/input.tsx",
"equal": true,
"sha256": "b459e459e31f842624eca4f605fab6e669165865f25e05f2631cd4f6328d7aba"
},
{
"file": "components/ui/textarea.tsx",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/components/ui/textarea.tsx",
"equal": true,
"sha256": "c9127763f39d4089b32aae1c834115c8bdaebb9482969d29ff3a3269a262bd27"
},
{
"file": "components/ui/dialog.tsx",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/components/ui/dialog.tsx",
"equal": true,
"sha256": "722846bb51d188a50e346426df730f1d48dc59c706a08fb9f21d48e2e02bd65b"
},
{
"file": "components/ui/select.tsx",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/components/ui/select.tsx",
"equal": true,
"sha256": "f6c850ceb4090e003bc13e99edeca670751e0ca35ede0d971211d8ce645bfdcc"
},
{
"file": "components/ui/skeleton.tsx",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/components/ui/skeleton.tsx",
"equal": true,
"sha256": "f21f692b93fc2b76602d60fc3131d19efdd466141b7576a981abfb766b894b5e"
},
{
"file": "components/ui/select-field.tsx",
"source": "/Users/pptg/workspace/project/d-sre-agent/apps/web/src/components/ui/select-field.tsx",
"equal": true,
"sha256": "06e1d7be112baaff061d5a8581869a6a5e8da1473dbd9987e41a5e853db03b87"
}
]

View File

@ -0,0 +1,14 @@
[
{
"name": "实际SDK计数 initial",
"ok": true
},
{
"name": "实际SDK计数 rework",
"ok": true
},
{
"name": "实际SDK计数 rejected",
"ok": true
}
]

View File

@ -0,0 +1,65 @@
a6f568b6f508e5a7c1860927e99109aa89a0a1be29abd3e8c228cb3d8d267c3a README.md
805f991a6000db1034d464a22053a3e26b41cdf07511bc2b56a4640e635da9ef array-stream/logs-input.txt
5fa255e8ecfd947617ccd24ee72803b1c452db5cdf0ae0d04487b54be216d462 array-stream/logs-run.txt
4e3aa7c4fa6dfc88c18b1abc8a43c2c4f5a25c9fa26ef79a462dcb997b827d1c array-stream/logs-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 array-stream/logs-supervisor.txt
b29d6e885648ae661b22a35bfb046ab4b919ba14d390e7ae2b5f7abeeeb4d44d array-stream/logs-timing.json
ec40e6518e60461fb8c4bb726b770dcf4f066db9e647d6275ea36685afe580eb array-stream/supervisor-results.json
46f1df270340cf0601605854660bc2c69c7d13927480f0b7665a6811cca833c4 array-stream-check.py
b63805084a5718597975c672f5dc4b355867b063ecfda90476b158fe1a19b335 array-stream-check.txt
8fe4571ad4e1c8d99b7bc7e6576a20f57e5d161cc24c4d631421068c0431e2c1 build.txt
6932c13d6ebeeb9048f3b0e9b28596044467c3c85b2fcbe458d46902baf70274 cancel-before-timeout-run.txt
0c9825aa844138a6400a3bb2cac81e658d600509add27f1634846926e111f443 cancel-before-timeout-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 cancel-before-timeout-supervisor.txt
a1ba8e3f64eb2facca0f09831c4e7c1f4286add627e52ed9b50c8e3eceeac664 cancel-before-timeout-timing.json
6932c13d6ebeeb9048f3b0e9b28596044467c3c85b2fcbe458d46902baf70274 cancel-run.txt
b548e2b73e932001c992c224678a8ab7bfd67877f17c6e4d2bebd2f4d299e277 cancel-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 cancel-supervisor.txt
0b4ce9538e0b50a0bc6d96eb3293527d4720dd98cd416f26ea7f150dfecb5885 cancel-timing.json
8fd80f201d271e7dae4d5a9141cc1061f9477c7f4f0d2ff4c80486af1459f822 count-check.mjs
8813aef93af3bedea3d2590f061967011f0a3cc7f65c9dd1aef5c217cc274978 count-results.json
aecaf96f7509af3ec9f408dce9b8ce9817fa5cb08cfc37d3ef8904e1ff856e0c count.txt
fee3d3bf789ba672604ac3a8ef27b5f67998ed9929d95d4ac8f19fdbf6feac6f depth-stream/inputs.json
0e0593b6b3538268db49896ff5913418bc24696d29e52d9b4fda2fb6b00144ef depth-stream/logs-run.txt
eadcc1695791d7c46f94aec18cec1918155fb92e4ed917c34d6c15a761c7517c depth-stream/logs-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 depth-stream/logs-supervisor.txt
d8a36aaeb627db68a5bf685fd18594a647deea5f5394cf247ad0e07da9eb1c53 depth-stream/logs-timing.json
d651bfd0e83dafa9f29bdb7f373a735dc74ec19461eaff035cf8d17eee012c4b depth-stream/supervisor-results.json
d7cce8d8b6f010006a64af3ede1701ae34334abf44e4ddcd3dfef91bc74f3a68 depth-stream-check.py
7bfd6be52aa25ca6c4ee56099c1a1917dfc1ceb286ca28a640a8753dfecbf919 depth-stream-check.txt
691c23de13708c512546c145815a8392a713166046fb97ffcc40608a6a339d96 fail-run.txt
f7b0651d0e63383d15175d0024499f4282255b66b01406f6eedd7e5ba4d3d551 fail-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 fail-supervisor.txt
a08adac95ae16c3c95126565c0b473ea0a9d2fe21b0d143b937b9d7b8062dbf6 fail-timing.json
976471555a48aa4892e54de62cdd35a273ec6f4109341def5378032261839fdd layout-stream/inputs.json
3d8cca4456c3313b7df41c1adb5a70f734cb8466f2e703dd3c8ce7ce4d4ddfd0 layout-stream/logs-run.txt
87433418fcfe7cd089027c6a715c6ba61eb61df01459e321ac32dde0a4b10cfb layout-stream/logs-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 layout-stream/logs-supervisor.txt
36d107f19bf1569770120e7953fdb5c49b68998c64301c5afe6e7ccc46959bc2 layout-stream/logs-timing.json
64291fbb3f8f873f00300f4b0d609a75044986f5791b00a7b712eaae85eed529 layout-stream/supervisor-results.json
45e014f0a16f5813db16163b4247f3b499880151be0f9add80d25bef72381346 layout-stream-check.py
2e746d67be6912abbc8098c6e27393ba491f9a017a59e0dcb3c1d5a053b8997d layout-stream-check.txt
b5632ecd0334b3af54ee3275252553025e65c1e985f66e33b75b11e69ea50a9b logs-input.txt
3f563ea0df8cc582f9e24ad464e52bfa53ba80feee380a657d7b32f2f9e280ac logs-run.txt
9ff36a2ca1b0c955aced50164b0b5a484ea9beef03447c60b68a31c839c32061 logs-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 logs-supervisor.txt
8d29c100f47e988eb99d0292e0ca9b285b28f66936a4123262c256d00b886f50 logs-timing.json
1b873b4a35d22c0ccd0976bd9e9e337a6fecc869dedcfe4cd977cf69e2cfba3a process-checks.json
56b8c28e710dfe420c302c18d03143411497d3ceb019339f53cca9b082ef488c product-delta.txt
0401d9a0c44543227b08b37c93cf35185fbba242382646ed5f96e6e5a2b49480 rework-results.json
a24f6fcc4b4d9725af7729a2c4c7ef926834175d7a51a1097ccede9764e3951b rework.mjs
fb1a3382270b4df4a2fce022cdf886ad0992e0d24a0d02fac0c8264d6af1c0d3 rework.txt
f2ad3bed9b54ff2c47c7616c051281f5c90ca57f9288856bca168372f7bc9f14 script-hashes.json
6e28394a51c4e4c67066232288cd155b0fd6b912644060d9b42ab865a29a970f spoof-snapshot.json
063b7aab61b6b2fbeec42a2ac7790e021bee9957c6925cee0fcbbedd4cdd1df4 stack-check.mjs
2f4741cd8a16411967d7c595a4f9ec54940a3fa72069cb9ac73de6a452c64dc4 stack-check.txt
e92d497aa94230e7df7d38a7eaeda8cab68e28fe862f3b8b8c11be9e0794a8b8 stack-results.json
49485445763d94fd290b5399fcf26de6bd02ab3f1fedcfeecdbaf20ad9edbc26 supervisor-check.py
01548334ad1dc911f1a5892762fa393a8403d25beacab6e513b8167768c96610 supervisor-check.txt
038cc0f1e989a0af1af21209a68f649880b9d6cffc8f918ccd7dd9436cc8e2b7 supervisor-results.json
bd3c5ed304edead32d4f2693643f3b623dd42bcee0b7d587f450b98011029568 timeout-run.txt
252815deaaf3b5bb57b186f98ea9358915b254c02b1295b83c4f4b81848b9eba timeout-state.json
e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855 timeout-supervisor.txt
74eaeff72fc02b542a8e6c01437ef7d14a12439991c7d1afbdc4b8a180eb8f92 timeout-timing.json
8fa1cf5506304e8abac55868e7f1a136c9b1dde57a3981a382da4c21ea129a6f typecheck.txt
c6800fcb2cfab6e00e02cc6c07270ae7f6a996bfc292e36299e78f00c8ddb0e1 unit.txt

View File

@ -0,0 +1,109 @@
{
"boundary": "受控远端/模型流,实际Engine/Store/Gitea/SDK,无真实模型调用",
"results": [
{
"name": "可信来源 trusted",
"ok": true
},
{
"name": "可信来源 untrusted",
"ok": true
},
{
"name": "可信来源 wrong-event",
"ok": true
},
{
"name": "可信来源 wrong-task",
"ok": true
},
{
"name": "可信来源 wrong-role",
"ok": true
},
{
"name": "可信来源 old-revision",
"ok": true
},
{
"name": "可信来源 old-head",
"ok": true
},
{
"name": "可信来源 old-base",
"ok": true
},
{
"name": "可信来源 pass",
"ok": true
},
{
"name": "可信来源 running",
"ok": true
},
{
"name": "可信来源 pending-only",
"ok": true
},
{
"name": "丢响应→SQLite重开→不同模型说明重试→一次分派→重复tick",
"ok": true
},
{
"name": "分派负例 active",
"ok": true
},
{
"name": "分派负例 runs",
"ok": true
},
{
"name": "分派负例 corrections",
"ok": true
},
{
"name": "分派负例 invalid-body",
"ok": true
},
{
"name": "分派负例 publish-error",
"ok": true
},
{
"name": "分派负例 cancel-during-post",
"ok": true
},
{
"name": "分派负例 revision-during-post",
"ok": true
},
{
"name": "可预测event_id不能让普通评论冒充返工请求",
"ok": true
},
{
"name": "发布await期间变化 head",
"ok": false,
"error": "AssertionError [ERR_ASSERTION]: {\"mode\":\"head\",\"starts\":1,\"task\":{\"id\":\"task\",\"title\":\"独立A03\",\"goal\":\"修复真实失败\",\"acceptance\":\"保持冻结要求\",\"status\":\"reviewing\",\"revision\":1,\"repository\":\"https://example.invalid/o/r\",\"targetBranch\":\"acceptance\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"issueNumber\":1,\"issueUrl\":\"https://example.invalid/o/r/issues/1\",\"corrections\":2,\"modelCalls\":0,\"wake\":false,\"createdAt\":\"2026-09-11T23:50:24.024Z\",\"updatedAt\":\"2026-09-11T23:50:24.025Z\"},\"runs\":[{\"id\":\"qa\",\"status\":\"completed\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"},{\"id\":\"dev\",\"status\":\"completed\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"},{\"id\":\"40f8e048-84ce-4948-a41e-537dc8e53cef\",\"status\":\"running\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"}]}\n\n1 !== 0\n\n at file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:100:9\n at async check (file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:10:35)\n at async file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:94:53"
},
{
"name": "发布await期间变化 base",
"ok": false,
"error": "AssertionError [ERR_ASSERTION]: {\"mode\":\"base\",\"starts\":1,\"task\":{\"id\":\"task\",\"title\":\"独立A03\",\"goal\":\"修复真实失败\",\"acceptance\":\"保持冻结要求\",\"status\":\"reviewing\",\"revision\":1,\"repository\":\"https://example.invalid/o/r\",\"targetBranch\":\"acceptance\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"issueNumber\":1,\"issueUrl\":\"https://example.invalid/o/r/issues/1\",\"corrections\":2,\"modelCalls\":0,\"wake\":false,\"createdAt\":\"2026-09-11T23:50:24.027Z\",\"updatedAt\":\"2026-09-11T23:50:24.028Z\"},\"runs\":[{\"id\":\"qa\",\"status\":\"completed\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"},{\"id\":\"dev\",\"status\":\"completed\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"},{\"id\":\"d90edbf8-ba67-49de-a92c-5829d5426757\",\"status\":\"running\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"}]}\n\n1 !== 0\n\n at file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:100:9\n at async check (file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:10:35)\n at async file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:94:53"
},
{
"name": "发布await期间变化 budget",
"ok": false,
"error": "AssertionError [ERR_ASSERTION]: {\"mode\":\"budget\",\"starts\":1,\"task\":{\"id\":\"task\",\"title\":\"独立A03\",\"goal\":\"修复真实失败\",\"acceptance\":\"保持冻结要求\",\"status\":\"reviewing\",\"revision\":1,\"repository\":\"https://example.invalid/o/r\",\"targetBranch\":\"acceptance\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"issueNumber\":1,\"issueUrl\":\"https://example.invalid/o/r/issues/1\",\"corrections\":2,\"modelCalls\":0,\"wake\":false,\"createdAt\":\"2026-09-11T23:50:24.030Z\",\"updatedAt\":\"2026-09-11T23:50:24.031Z\"},\"runs\":[{\"id\":\"qa\",\"status\":\"completed\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"},{\"id\":\"dev\",\"status\":\"completed\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"},{\"id\":\"538f3750-c825-4e1b-905f-7acc2545d732\",\"status\":\"running\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"}]}\n\n1 !== 0\n\n at file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:100:9\n at async check (file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:10:35)\n at async file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:94:53"
},
{
"name": "发布await期间变化 active",
"ok": false,
"error": "AssertionError [ERR_ASSERTION]: {\"mode\":\"active\",\"starts\":1,\"task\":{\"id\":\"task\",\"title\":\"独立A03\",\"goal\":\"修复真实失败\",\"acceptance\":\"保持冻结要求\",\"status\":\"reviewing\",\"revision\":1,\"repository\":\"https://example.invalid/o/r\",\"targetBranch\":\"acceptance\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"issueNumber\":1,\"issueUrl\":\"https://example.invalid/o/r/issues/1\",\"corrections\":2,\"modelCalls\":0,\"wake\":false,\"createdAt\":\"2026-09-11T23:50:24.033Z\",\"updatedAt\":\"2026-09-11T23:50:24.034Z\"},\"runs\":[{\"id\":\"qa\",\"status\":\"completed\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"},{\"id\":\"dev\",\"status\":\"completed\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"},{\"id\":\"new-active\",\"status\":\"running\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"},{\"id\":\"0ed38ee5-5376-4933-9e27-1a6060b2b4ac\",\"status\":\"running\",\"headSha\":\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\",\"baseSha\":\"bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\"}]}\n\n1 !== 0\n\n at file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:100:9\n at async check (file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:10:35)\n at async file:///Users/pptg/workspace/demo/ppanel/.ppanel/build/recovery-review-codex/evidence/issue-47/rework.mjs:94:53"
},
{
"name": "真实SDK:决定正文权威版本在前、错误模型原文及原始事件保留",
"ok": true
}
]
}

View File

@ -0,0 +1,86 @@
[
{
"name": "普通/非法转义 \"ordinary curl -H \\\"Authorization: Bearer FAKE_STACK\\\" https://example.invalid\"",
"ok": true,
"ms": 0.1812920000000009
},
{
"name": "普通/非法转义 \"not JSON \\\"bad\\\\escape\\\" remains text\"",
"ok": true,
"ms": 0.011625000000002217
},
{
"name": "普通/非法转义 \"broken \\\"unterminated trailing\\\\\"",
"ok": true,
"ms": 0.008290999999999826
},
{
"name": "普通/非法转义 \"plain 中文 🧪 \\n \\t \\\\\"",
"ok": true,
"ms": 0.06854200000000077
},
{
"name": "普通/非法转义 \"bad \\\"\\\\uZZZZ\\\" and \\\"valid text\\\"\"",
"ok": true,
"ms": 0.15191699999999742
},
{
"name": "JSON原始类型 null",
"ok": true,
"ms": 0.11387500000000017
},
{
"name": "JSON原始类型 true",
"ok": true,
"ms": 0.004374999999999574
},
{
"name": "JSON原始类型 false",
"ok": true,
"ms": 0.0024590000000017653
},
{
"name": "JSON原始类型 123",
"ok": true,
"ms": 0.00837500000000091
},
{
"name": "JSON原始类型 \"true\"",
"ok": true,
"ms": 0.00916700000000148
},
{
"name": "JSON原始类型 \"123\"",
"ok": true,
"ms": 0.004790999999997325
},
{
"name": "JSON原始类型 \"null\"",
"ok": true,
"ms": 0.0027500000000024727
},
{
"name": "JSON原始类型 \"\\\"quoted\\\"\"",
"ok": true,
"ms": 0.014834000000000458
},
{
"name": "多层编码近1MiB终止并完整解码",
"ok": true,
"ms": 12.537208,
"bytes": 721171,
"depth": 15
},
{
"name": "10000层结构对象不递归爆栈且保留全部层",
"ok": true,
"ms": 27.614292,
"bytes": 60127,
"depth": 10000
},
{
"name": "注册的虚构复杂秘密在三层编码里移除",
"ok": true,
"ms": 0.10162499999999852
}
]

View File

@ -0,0 +1,54 @@
[
{
"name": "logs 实际身份及终态持久化",
"ok": true
},
{
"name": "logs 权威原因与后续退出证据",
"ok": true
},
{
"name": "实际stdout分块单行JSON可解析并脱敏且非秘密字段保留",
"ok": true
},
{
"name": "实际stdout内嵌JSON字符串仍可解析",
"ok": true
},
{
"name": "实际stderr多行JSON仍可解析",
"ok": true
},
{
"name": "fail 实际身份及终态持久化",
"ok": true
},
{
"name": "fail 权威原因与后续退出证据",
"ok": true
},
{
"name": "timeout 实际身份及终态持久化",
"ok": true
},
{
"name": "timeout 权威原因与后续退出证据",
"ok": true
},
{
"name": "cancel 实际身份及终态持久化",
"ok": true
},
{
"name": "cancel 权威原因与后续退出证据",
"ok": true
},
{
"name": "cancel-before-timeout 实际身份及终态持久化",
"ok": true
},
{
"name": "cancel-before-timeout 权威原因与后续退出证据",
"ok": true
}
]

View File

@ -0,0 +1,5 @@
ppanel-server@ /Users/pptg/workspace/demo/ppanel/.ppanel/build/full-qa/apps/server
├─┬ @earendil-works/pi-agent-core@0.85.0
│ └── @earendil-works/pi-ai@0.85.0 deduped
└── @earendil-works/pi-ai@0.85.0

View File

@ -0,0 +1,45 @@
> test
> tsx --test test/*.test.ts
✔ HTTP 创建/单活跃/修订/取消持久化,安全投影与跨站限制 (79.92775ms)
fatal: '../main' is not a valid branch name
✔ 合并必须匹配 revision/head/base、独立 CR/QA 且无活跃修正 (0.467208ms)
✔ SQLite 事件去重三次、待发布和游标重启保留 (5.8695ms)
✔ 标记必须独立尾行,凭据清理及 Git 分支验证 (13.673ms)
✔ Codex 连续 reader 保留响应前后的完成事件并路由 RPC (73.158667ms)
✔ 明确故障注入:发布成功但响应丢失,重放不重复评论 (1.429209ms)
✔ 并发发布共享一个串行 outbox,不重复创建同 event_id 评论 (22.786917ms)
✔ 候选必须包含目标基线,分叉基线和循环提交图不能通过 (0.276ms)
✔ redaction preserves JSON escapes and complete non-secret event content (0.706584ms)
✔ timeout and cancellation remain authoritative over exit 143; ordinary failures retain code (0.061125ms)
✔ stderr line slicing preserves pretty JSON amid plain text and chunked transport (0.676459ms)
✔ all JSON string layouts share safe token redaction across stream chunks (2.671833ms)
✔ nested JSON text is redacted at every encoding layer without damaging inner structure (0.614833ms)
✔ 轮询等待期间新报告写入,不能被旧运行快照覆盖 (3.316709ms)
✔ 恢复只接受本地受信 outbox 投影;普通伪造 marker 不成为报告 (0.461375ms)
✔ 取消后迟到的受信报告仅留档,不唤醒任务 (0.294291ms)
✔ 合并读取远端期间需求修订,不能用旧快照发送合并 (0.376584ms)
✔ 空闲重复轮询不调用编排模型,恢复只唤醒无运行的未消费任务 (2.108583ms)
✔ CR P0: gitea:<id> 自指 marker 跨轮询不能恢复为受信报告 (0.377791ms)
✔ CR P0: 外来伪造和普通事件不能跨数据库重启获得来源证明 (4.942125ms)
✔ A07 only proven unstarted assignments resume; uncertain launches remain stopped (1.428625ms)
✔ terminal task observes ended workers without turning process exit into success (0.278791ms)
✔ real SDK corrects text-only wait once; normal wait and repeated text stay distinct (5.704458ms)
✔ invalid assignment has zero issue/run side effects and clean retry dispatches once (0.986083ms)
✔ old unpublished validation orphan retires only with complete negative evidence (3.949375ms)
✔ trusted failure rework survives lost response and restart without duplicate revision_request (5.18975ms)
✔ publication requires exact body and binding; old false confirmation fails closed (0.438208ms)
✔ SDK dispatch decision labels pre-action correction count without changing budget (5.61325ms)
From /var/folders/q4/m_dy0g_90lj2pbc9tq6db3h40000gn/T/ppanel-frozen-SwQtck
* branch b43d10b9c6da9f4e7b26fcfaf2d07db9814cf244 -> FETCH_HEAD
✔ 真实 Git: 基线验收摘要不可被弱化、删除、改名或修订清单覆盖 (327.622583ms)
✔ private temp paths remain stable across restart and reject foreign reports (39.439042ms)
ℹ tests 30
ℹ suites 0
ℹ pass 30
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 799.493625

View File

@ -0,0 +1,9 @@
{
"pid": 49550,
"identity": "Sat Sep 12 08:01:56 2026",
"productSha": "8d374084e109bbabc9e4bcec2a12c4088099db7d",
"data": "/Users/pptg/workspace/demo/ppanel/.ppanel/build/full-qa/.ppanel/issue26/final/data",
"allProjectionsAndLogHashesUnchanged": true,
"taskCount": 5,
"logCount": 25
}

View File

@ -0,0 +1,129 @@
diff --git a/apps/server/src/engine.ts b/apps/server/src/engine.ts
index 02bc640..8dc9613 100644
--- a/apps/server/src/engine.ts
+++ b/apps/server/src/engine.ts
@@ -910,7 +910,7 @@ export class Engine {
t = this.store.task(taskId);
if (t.revision !== revision || terminal(t.status))
throw new Error("需求版本已变化或任务停止");
- Gitea.prototype.validateBody(`## 编排决定:${p.action}\n\n权威需求版本(goal revision):${t.revision};开发修正轮次:${t.corrections}。自动返工不改变需求版本,模型说明中的版本如有冲突,以此为准。\n\n### 模型说明\n${p.reason}`);
+ Gitea.prototype.validateBody(`## 编排决定:${p.action}\n\n权威需求版本(goal revision):${t.revision};本次决定执行前已使用的开发修正次数:${t.corrections}(不含本次分派)。自动返工不改变需求版本,模型说明中的版本如有冲突,以此为准。\n\n### 模型说明\n${p.reason}`);
let result: any;
if (p.action === "dispatch")
result = await this.dispatch(taskId, p.role, p.runtime, p.reason);
@@ -965,7 +965,7 @@ export class Engine {
t.id,
t.issueNumber!,
"decision",
- `## 编排决定:${p.action}\n\n权威需求版本(goal revision):${t.revision};开发修正轮次:${t.corrections}。自动返工不改变需求版本,模型说明中的版本如有冲突,以此为准。\n\n### 模型说明\n${p.reason}`,
+ `## 编排决定:${p.action}\n\n权威需求版本(goal revision):${t.revision};本次决定执行前已使用的开发修正次数:${t.corrections}(不含本次分派)。自动返工不改变需求版本,模型说明中的版本如有冲突,以此为准。\n\n### 模型说明\n${p.reason}`,
);
action = true;
return {
diff --git a/apps/server/src/gitea.ts b/apps/server/src/gitea.ts
index 810f53d..40cebff 100644
--- a/apps/server/src/gitea.ts
+++ b/apps/server/src/gitea.ts
@@ -2,13 +2,52 @@ import https from "node:https";
import { execFileSync } from "node:child_process";
import { Store, id } from "./store.js";
export const secrets = new Set<string>();
-export function redact(s: string) {
+function redactText(s: string) {
for (const v of secrets) if (v) s = s.split(v).join("[凭据已移除]");
return s.replace(
/(Authorization["': ]+(?:Bearer|token)\s+)[^\s"']+/gi,
"$1[凭据已移除]",
);
}
+export function redact(s: string) {
+ // Each decoded JSON string is shorter than its quoted representation. An
+ // explicit postorder stack handles nested encodings without call-stack recursion.
+ type Frame = { input: string; parts?: (string | Frame)[]; output?: string };
+ const root: Frame = { input: s }, stack = [root];
+ while (stack.length) {
+ const frame = stack[stack.length - 1];
+ if (frame.parts) {
+ frame.output = frame.parts.map((part) => typeof part === "string" ? part : JSON.stringify(part.output)).join("");
+ stack.pop();
+ continue;
+ }
+ const parts: (string | Frame)[] = [], children: Frame[] = [];
+ let end = 0, authorization = false;
+ for (const match of frame.input.matchAll(/"(?:[^"\\\r\n]|\\.)*"/g)) {
+ let value: string;
+ try { value = JSON.parse(match[0]); } catch { continue; }
+ parts.push(redactText(frame.input.slice(end, match.index)));
+ const isKey = /^\s*:/.test(frame.input.slice(match.index! + match[0].length));
+ let embedded = false;
+ try { JSON.parse(value); embedded = true; } catch { /* Ordinary command/text. */ }
+ if (embedded && !isKey) {
+ const child: Frame = { input: value };
+ parts.push(child); children.push(child);
+ } else {
+ const clean = authorization && !isKey
+ ? redactText(value).replace(/^(Bearer|token)\s+.+$/is, "$1 [凭据已移除]")
+ : redactText(value);
+ parts.push(JSON.stringify(clean));
+ }
+ authorization = isKey && value.toLowerCase() === "authorization";
+ end = match.index! + match[0].length;
+ }
+ parts.push(redactText(frame.input.slice(end)));
+ frame.parts = parts;
+ for (const child of children) stack.push(child);
+ }
+ return root.output!;
+}
export class Gitea {
base: string;
token: string;
diff --git a/apps/server/src/supervisor-state.ts b/apps/server/src/supervisor-state.ts
new file mode 100644
index 0000000..71a3fb8
--- /dev/null
+++ b/apps/server/src/supervisor-state.ts
@@ -0,0 +1,5 @@
+// The first termination reason is authoritative; a later CLI error is additional evidence.
+export function recordFailure(state: { error: string | null; exitError?: string }, error: string) {
+ state.exitError = error;
+ state.error ??= error;
+}
diff --git a/apps/server/src/supervisor.ts b/apps/server/src/supervisor.ts
index 216f2e1..df09bb5 100644
--- a/apps/server/src/supervisor.ts
+++ b/apps/server/src/supervisor.ts
@@ -1,3 +1,4 @@
+import { recordFailure } from "./supervisor-state.js";
import { workerEnvironment, workspacePrompt } from "./workspace.js";
import { spawn, execFileSync } from "node:child_process";
import {
@@ -45,6 +46,8 @@ let child: ReturnType<typeof spawn> | undefined,
cancelled = false;
const stop = () => {
cancelled = true;
+ state.error ??= "执行已取消";
+ save();
if (rpc && threadId && turnId)
void rpc.request("turn/interrupt", { threadId, turnId }).catch(() => {});
setTimeout(() => {
@@ -63,7 +66,7 @@ const stop = () => {
process.on("SIGTERM", stop);
process.on("SIGINT", stop);
const timer = setTimeout(() => {
- state.error = "超过单次执行时限";
+ state.error ??= "超过单次执行时限";
stop();
}, spec.runMs);
try {
@@ -178,8 +181,9 @@ try {
if (stderr) log(stderr);
if (cancelled) throw new Error(state.error ?? "执行已取消");
} catch (e) {
- state.error = redact(String(e));
+ recordFailure(state, redact(String(e)));
log(state.error + "\n");
+ if (state.exitError !== state.error) log(state.exitError + "\n");
} finally {
clearTimeout(timer);
if (child?.pid)

View File

@ -0,0 +1,12 @@
[
{
"url": "/assets/index-dW6Ub4pr.js",
"bytes": 547038,
"sha256": "4c6849f810655fe4a4750dd7613e03cceaba3d20bf656dd14bdd12b0d4f52d1d"
},
{
"url": "/assets/index-CKP-5SMK.css",
"bytes": 60374,
"sha256": "1ecdae6b517ca9073d640ac61c32329ae7023942bbc8b6cea325da5283a0211c"
}
]

View File

@ -0,0 +1,61 @@
"""Actual supervisor, harmless local CLI fixtures. All credentials synthetic; no network."""
import os,json,pathlib,subprocess,time,signal,shutil,tempfile,sys
root=pathlib.Path.cwd();out=root/'evidence/issue-26-control-final/log-check';private=pathlib.Path(tempfile.mkdtemp(prefix='supervisor-efd-',dir=root/'.ppanel/review-tmp'))
bin=private/'bin';bin.mkdir(exist_ok=True)
node=shutil.which('node');(private/'config.yaml').write_text('llm:\n items: []\n')
(bin/'git').write_text('#!'+node+'\nprocess.stdout.write("username=fixture\\npassword=SYNTHETIC_GIT_VALUE\\n");\n');(bin/'git').chmod(0o755)
cli=r'''const fs=require('node:fs');
const mode=process.env.FIXTURE_MODE;
fs.writeFileSync(process.env.FIXTURE_READY,JSON.stringify({pid:process.pid,at:Date.now()}));
process.on('SIGTERM',()=>{fs.appendFileSync(process.env.FIXTURE_SIGNALS,JSON.stringify({signal:'SIGTERM',at:Date.now()})+'\n');if(mode==='cancel-before-timeout')setTimeout(()=>process.exit(143),1000);else process.exit(143);});
process.stdin.resume();
if(mode==='logs'){
const inner=JSON.stringify({Authorization:'Bearer SYNTHETIC_AUTH',path:'a\\b',safe:'keep'});
const obj={type:'tool_execution_start',args:{command:'curl -H "Authorization: Bearer SYNTHETIC_AUTH" https://example.invalid',nested:{Authorization:'Bearer SYNTHETIC_AUTH',safe:'keep'},encoded:inner},token:process.env.PPANEL_WORKER_TOKEN,safe:'quote " slash \\ newline\n中文'};
const compact=JSON.stringify(obj)+'\n';
fs.writeFileSync(process.env.FIXTURE_INPUT,compact+JSON.stringify({type:'pretty',args:{command:obj.args.command},safe:obj.safe},null,2)+'\n');
process.stdout.write(compact.slice(0,31));setTimeout(()=>{process.stdout.write(compact.slice(31));process.stderr.write(JSON.stringify({type:'pretty',args:{command:obj.args.command},safe:obj.safe},null,2)+'\n');setTimeout(()=>process.exit(0),80);},40);
}else if(mode==='fail')setTimeout(()=>process.exit(7),60);
else setInterval(()=>{},100);
'''
for name in ['pi','opencode']:(bin/name).write_text('#!'+node+'\n'+cli);(bin/name).chmod(0o755)
results=[]
def check(name,fn):
try:fn();results.append({'name':name,'ok':True})
except Exception as e:results.append({'name':name,'ok':False,'error':repr(e)})
for mode in ['logs','fail','timeout','cancel','cancel-before-timeout']:
d=private/mode;d.mkdir(exist_ok=True);spec={'runId':mode,'cwd':str(d),'prompt':'无害受控执行','token':'SYNTHETIC_WORKER_VALUE','database':str(d/'db'),'repository':'https://example.invalid/o/r','runtime':'pi','session':str(d/'session'),'state':str(d/'state.json'),'log':str(d/'run.txt'),'url':'http://127.0.0.1:1','runMs':600 if mode=='timeout' else 1200 if mode=='cancel-before-timeout' else 8000}
(d/'spec.json').write_text(json.dumps(spec));env={**os.environ,'PATH':str(bin)+os.pathsep+os.environ['PATH'],'PPANEL_MODEL_CONFIG':str(private/'config.yaml'),'FIXTURE_MODE':mode,'FIXTURE_INPUT':str(d/'input.txt'),'FIXTURE_READY':str(d/'ready.json'),'FIXTURE_SIGNALS':str(d/'signals.jsonl')}
start=time.monotonic()
with (out/(mode+'-supervisor.txt')).open('w') as stream:
proc=subprocess.Popen([node,str(root/'apps/server/dist/supervisor.js'),str(d/'spec.json')],env=env,stdout=stream,stderr=subprocess.STDOUT,start_new_session=True)
deadline=time.monotonic()+8
while not (d/'ready.json').exists() and proc.poll() is None and time.monotonic()<deadline:time.sleep(.02)
if mode.startswith('cancel'):proc.send_signal(signal.SIGTERM)
try:proc.wait(timeout=12)
except subprocess.TimeoutExpired:os.killpg(proc.pid,signal.SIGKILL);proc.wait();raise
state=json.loads((d/'state.json').read_text());log=(d/'run.txt').read_text();(out/(mode+'-state.json')).write_text(json.dumps(state,ensure_ascii=False,indent=2));(out/(mode+'-run.txt')).write_text(log)
ready=json.loads((d/'ready.json').read_text());signals=(d/'signals.jsonl').read_text() if (d/'signals.jsonl').exists() else ''
(out/(mode+'-timing.json')).write_text(json.dumps({'elapsed':time.monotonic()-start,'child':ready,'signals':signals,'supervisorExit':proc.returncode},indent=2))
def identity():
assert state['runId']==mode and state['pid']==proc.pid and state['identity'] and state['startedAt'] and state['endedAt'];assert state['sessionId']==spec['session']
check(mode+' 实际身份及终态持久化',identity)
def reason():
if mode=='logs':assert state['error'] is None
elif mode=='fail':assert state['error']=='Error: 执行进程退出 7' and state['exitError']==state['error']
else:
assert state['error']==('超过单次执行时限' if mode=='timeout' else '执行已取消');assert '143' in state['exitError'];assert state['error'] in log and state['exitError'] in log;assert 'SIGTERM' in signals
check(mode+' 权威原因与后续退出证据',reason)
if mode=='logs':
(out/'logs-input.txt').write_bytes((d/'input.txt').read_bytes())
lines=log.splitlines();first=json.loads(lines[0]);pretty='\n'.join(lines[1:])
def compact():
assert first['safe']=='quote " slash \\ newline\n中文';assert first['args']['nested']['safe']=='keep';assert 'SYNTHETIC_AUTH' not in log;assert 'SYNTHETIC_WORKER_VALUE' not in log
check('实际stdout分块单行JSON可解析并脱敏且非秘密字段保留',compact)
def embedded():assert json.loads(first['args']['encoded'])['safe']=='keep'
check('实际stdout内嵌JSON字符串仍可解析',embedded)
def multiline():assert json.loads(pretty)['type']=='pretty'
check('实际stderr多行JSON仍可解析',multiline)
(out/'supervisor-results.json').write_text(json.dumps(results,ensure_ascii=False,indent=2));print(json.dumps(results,ensure_ascii=False,indent=2))
sys.exit(1 if any(not r['ok'] for r in results) else 0)

View File

@ -0,0 +1,54 @@
[
{
"name": "logs 实际身份及终态持久化",
"ok": true
},
{
"name": "logs 权威原因与后续退出证据",
"ok": true
},
{
"name": "实际stdout分块单行JSON可解析并脱敏且非秘密字段保留",
"ok": true
},
{
"name": "实际stdout内嵌JSON字符串仍可解析",
"ok": true
},
{
"name": "实际stderr多行JSON仍可解析",
"ok": true
},
{
"name": "fail 实际身份及终态持久化",
"ok": true
},
{
"name": "fail 权威原因与后续退出证据",
"ok": true
},
{
"name": "timeout 实际身份及终态持久化",
"ok": true
},
{
"name": "timeout 权威原因与后续退出证据",
"ok": true
},
{
"name": "cancel 实际身份及终态持久化",
"ok": true
},
{
"name": "cancel 权威原因与后续退出证据",
"ok": true
},
{
"name": "cancel-before-timeout 实际身份及终态持久化",
"ok": true
},
{
"name": "cancel-before-timeout 权威原因与后续退出证据",
"ok": true
}
]

View File

@ -0,0 +1,12 @@
> typecheck
> npm run typecheck --prefix apps/server && npm run typecheck --prefix apps/web
> typecheck
> tsc --noEmit
> @ppanel/web@0.1.0 typecheck
> tsc --noEmit