@fitlab-ai/agent-infra 0.9.13 → 0.9.15

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (428) hide show
  1. package/README.md +10 -0
  2. package/README.zh-CN.md +10 -0
  3. package/bin/cli.ts +26 -9
  4. package/bin/internal-cli.sh +45 -0
  5. package/dist/bin/cli.js +173 -157
  6. package/dist/bin/internal-cli.js +126 -42
  7. package/dist/lib/agent-client.js +5 -0
  8. package/dist/lib/agent-clients/adapters/codex-lifecycle/capability-store.js +165 -27
  9. package/dist/lib/agent-clients/adapters/codex-lifecycle/sandbox-controller.js +3 -2
  10. package/dist/lib/defaults.json +103 -103
  11. package/dist/lib/fs/durable-write.js +37 -0
  12. package/dist/lib/fs/file-lock.js +111 -0
  13. package/dist/lib/host-control/audit.js +19 -0
  14. package/dist/lib/host-control/client.js +142 -0
  15. package/dist/lib/host-control/command.js +49 -0
  16. package/dist/lib/host-control/path.js +160 -0
  17. package/dist/lib/host-control/secure-fs.js +140 -0
  18. package/dist/lib/host-control/server.js +159 -0
  19. package/dist/lib/host-control/service.js +112 -0
  20. package/dist/lib/init.js +11 -2
  21. package/dist/lib/internal/agent-client.js +5 -2
  22. package/dist/lib/internal/cli-route-inventory.js +154 -0
  23. package/dist/lib/internal/codex-lifecycle.js +15 -12
  24. package/dist/lib/internal/codex-sandbox-controller.js +8 -2
  25. package/dist/lib/internal/demo-promotion.js +322 -0
  26. package/dist/lib/internal/demo-transcript.js +454 -0
  27. package/dist/lib/internal/git-workflow.js +10 -6
  28. package/dist/lib/internal/host-control.js +28 -0
  29. package/dist/lib/internal/manual-validation.js +322 -0
  30. package/dist/lib/internal/platform-checks.js +16 -4
  31. package/dist/lib/internal/platform-comment.js +16 -4
  32. package/dist/lib/internal/platform-context.js +4 -1
  33. package/dist/lib/internal/platform-issue.js +12 -4
  34. package/dist/lib/internal/platform-metadata.js +88 -0
  35. package/dist/lib/internal/platform-pr-review.js +16 -3
  36. package/dist/lib/internal/platform-pr.js +42 -14
  37. package/dist/lib/internal/platform-release-notes.js +6 -3
  38. package/dist/lib/internal/platform-security.js +93 -0
  39. package/dist/lib/internal/pr-review-grade.js +14 -3
  40. package/dist/lib/internal/release-workflow.js +82 -89
  41. package/dist/lib/internal/sandbox-control.js +171 -7
  42. package/dist/lib/internal/task-activity.js +13 -4
  43. package/dist/lib/internal/task-artifact.js +19 -107
  44. package/dist/lib/internal/task-context.js +4 -1
  45. package/dist/lib/internal/task-create.js +48 -14
  46. package/dist/lib/internal/task-delivery.js +4 -1
  47. package/dist/lib/internal/task-event.js +38 -35
  48. package/dist/lib/internal/task-finalization.js +4 -1
  49. package/dist/lib/internal/task-invalidation.js +4 -1
  50. package/dist/lib/internal/task-ledger.js +12 -1
  51. package/dist/lib/internal/task-lifecycle.js +4 -1
  52. package/dist/lib/internal/task-operation-registry.js +519 -0
  53. package/dist/lib/internal/task-orchestration.js +12 -0
  54. package/dist/lib/internal/task-override.js +10 -3
  55. package/dist/lib/internal/task-qualification.js +90 -0
  56. package/dist/lib/internal/task-review.js +10 -52
  57. package/dist/lib/internal/task-short-id.js +11 -1
  58. package/dist/lib/internal/task-snapshot.js +4 -1
  59. package/dist/lib/internal/task-validate.js +37 -12
  60. package/dist/lib/internal/task-verify.js +5 -1
  61. package/dist/lib/internal/task-warning.js +8 -1
  62. package/dist/lib/owned-path.js +33 -0
  63. package/dist/lib/platform/check-status.js +11 -0
  64. package/dist/lib/platform/comment-safety.js +229 -0
  65. package/dist/lib/platform/completion-backfill.js +1 -1
  66. package/dist/lib/platform/github-data.js +241 -0
  67. package/dist/lib/platform/github-provider.js +248 -26
  68. package/dist/lib/platform/issue-comments.js +146 -34
  69. package/dist/lib/platform/issues.js +1 -179
  70. package/dist/lib/platform/pr-change-report.js +6 -12
  71. package/dist/lib/platform/pr-checks.js +12 -65
  72. package/dist/lib/platform/pr-summary.js +83 -10
  73. package/dist/lib/platform/provider-contract.d.ts +67 -1
  74. package/dist/lib/platform/provider-contract.js +6 -2
  75. package/dist/lib/platform/provider-validation.js +76 -3
  76. package/dist/lib/platform/pull-requests.js +117 -275
  77. package/dist/lib/platform/repository-metadata.js +377 -0
  78. package/dist/lib/platform/security-alerts.js +155 -0
  79. package/dist/lib/platform/snapshots.js +2 -0
  80. package/dist/lib/platform/verification-sync.js +154 -162
  81. package/dist/lib/process-data/index.js +7 -0
  82. package/dist/lib/release/workflow.js +1 -1
  83. package/dist/lib/run/index.js +5 -0
  84. package/dist/lib/sandbox/commands/control-status.js +22 -0
  85. package/dist/lib/sandbox/commands/create.js +39 -8
  86. package/dist/lib/sandbox/commands/ls.js +5 -0
  87. package/dist/lib/sandbox/commands/rm.js +3 -807
  88. package/dist/lib/sandbox/commands/show.js +16 -0
  89. package/dist/lib/sandbox/control/audit.js +203 -0
  90. package/dist/lib/sandbox/control/client.js +125 -12
  91. package/dist/lib/sandbox/control/completed-reentry.js +85 -0
  92. package/dist/lib/sandbox/control/container-identity.js +28 -31
  93. package/dist/lib/sandbox/control/executor.js +233 -34
  94. package/dist/lib/sandbox/control/finalization-response.js +27 -0
  95. package/dist/lib/sandbox/control/identity-sentinel.js +107 -0
  96. package/dist/lib/sandbox/control/lifecycle.js +656 -128
  97. package/dist/lib/sandbox/control/native-file-lock.js +85 -0
  98. package/dist/lib/sandbox/control/protocol.js +55 -4
  99. package/dist/lib/sandbox/control/server.js +720 -79
  100. package/dist/lib/sandbox/control/state.js +63 -3
  101. package/dist/lib/sandbox/control/task-view.js +183 -0
  102. package/dist/lib/sandbox/control/task-workflow.js +74 -0
  103. package/dist/lib/sandbox/control/workflow-executor.js +99 -0
  104. package/dist/lib/sandbox/engines/authority.js +289 -0
  105. package/dist/lib/sandbox/engines/docker-exact-discovery.js +97 -0
  106. package/dist/lib/sandbox/index.js +7 -2
  107. package/dist/lib/sandbox/recovery.js +49 -5
  108. package/dist/lib/sandbox/removal.js +1717 -0
  109. package/dist/lib/sandbox/task-cleanup.js +217 -0
  110. package/dist/lib/sandbox/workspace-view.js +15 -15
  111. package/dist/lib/sandbox/worktree-safety.js +4 -1
  112. package/dist/lib/server/index.js +8 -0
  113. package/dist/lib/task/activity-intent.js +2 -1
  114. package/dist/lib/task/activity-log.js +30 -24
  115. package/dist/lib/task/artifact-command.js +101 -0
  116. package/dist/lib/task/artifact-lifecycle.js +71 -40
  117. package/dist/lib/task/artifact-name.js +36 -0
  118. package/dist/lib/task/artifact-operations.js +474 -0
  119. package/dist/lib/task/artifact-receipts.js +3 -15
  120. package/dist/lib/task/artifact-repair-intent.js +130 -0
  121. package/dist/lib/task/artifact-schema.js +127 -0
  122. package/dist/lib/task/capabilities.js +83 -19
  123. package/dist/lib/task/codex-orchestration.js +71 -25
  124. package/dist/lib/task/commit-identity.js +28 -0
  125. package/dist/lib/task/commit-operation.js +2 -16
  126. package/dist/lib/task/control-authority.js +375 -5
  127. package/dist/lib/task/control-recovery.js +128 -0
  128. package/dist/lib/task/create-service.js +103 -4
  129. package/dist/lib/task/create.js +80 -2
  130. package/dist/lib/task/decision-details.js +2 -46
  131. package/dist/lib/task/events.js +306 -76
  132. package/dist/lib/task/finalization-state.js +9 -0
  133. package/dist/lib/task/human-override.js +3 -6
  134. package/dist/lib/task/implementation-inputs.js +3 -3
  135. package/dist/lib/task/index.js +11 -3
  136. package/dist/lib/task/intermediate-cleanup.js +383 -0
  137. package/dist/lib/task/invalidation.js +18 -65
  138. package/dist/lib/task/ledger-intents.js +7 -22
  139. package/dist/lib/task/ledger.js +1 -3
  140. package/dist/lib/task/lifecycle.js +19 -1
  141. package/dist/lib/task/local-artifact-finalization.js +190 -290
  142. package/dist/lib/task/manual-validation-completion.js +26 -0
  143. package/dist/lib/task/manual-validation-evidence.js +130 -0
  144. package/dist/lib/task/manual-validation-receipt.js +101 -0
  145. package/dist/lib/task/manual-validation-shared.js +40 -0
  146. package/dist/lib/task/manual-validation-transaction.js +243 -0
  147. package/dist/lib/task/markdown.js +47 -0
  148. package/dist/lib/task/orchestration.js +206 -69
  149. package/dist/lib/task/qualification-audit.js +476 -0
  150. package/dist/lib/task/qualification-confirmation.js +109 -0
  151. package/dist/lib/task/qualification-intents.js +148 -0
  152. package/dist/lib/task/review-artifacts.js +23 -40
  153. package/dist/lib/task/review-command.js +25 -0
  154. package/dist/lib/task/review-finalization.js +146 -97
  155. package/dist/lib/task/review-fingerprint.js +2 -2
  156. package/dist/lib/task/sections.js +34 -54
  157. package/dist/lib/task/verification-config.js +11 -0
  158. package/dist/lib/task/verification-engine.js +170 -327
  159. package/dist/lib/task/verification.js +1 -1
  160. package/dist/lib/task/workflow-command.js +17 -0
  161. package/dist/lib/task/workflow-warning-intents.js +0 -16
  162. package/dist/lib/task/workflow-warnings.js +28 -2
  163. package/dist/lib/update.js +2 -0
  164. package/dist/lifecycle-build-manifest.json +3 -3
  165. package/lib/agent-client.ts +9 -0
  166. package/lib/agent-clients/adapters/codex-lifecycle/capability-store.ts +223 -33
  167. package/lib/agent-clients/adapters/codex-lifecycle/sandbox-controller.ts +3 -2
  168. package/lib/fs/durable-write.ts +29 -0
  169. package/lib/fs/file-lock.ts +113 -0
  170. package/lib/host-control/audit.ts +22 -0
  171. package/lib/host-control/client.ts +172 -0
  172. package/lib/host-control/command.ts +60 -0
  173. package/lib/host-control/path.ts +167 -0
  174. package/lib/host-control/secure-fs.ts +135 -0
  175. package/lib/host-control/server.ts +180 -0
  176. package/lib/host-control/service.ts +133 -0
  177. package/lib/init.ts +15 -2
  178. package/lib/internal/agent-client.ts +4 -2
  179. package/lib/internal/cli-route-inventory.ts +142 -0
  180. package/lib/internal/codex-lifecycle.ts +14 -12
  181. package/lib/internal/codex-sandbox-controller.ts +7 -2
  182. package/lib/internal/demo-promotion.ts +354 -0
  183. package/lib/internal/demo-transcript.ts +419 -0
  184. package/lib/internal/git-workflow.ts +9 -6
  185. package/lib/internal/host-control.ts +28 -0
  186. package/lib/internal/manual-validation.ts +306 -0
  187. package/lib/internal/platform-checks.ts +12 -4
  188. package/lib/internal/platform-comment.ts +12 -4
  189. package/lib/internal/platform-context.ts +3 -1
  190. package/lib/internal/platform-issue.ts +11 -4
  191. package/lib/internal/platform-metadata.ts +70 -0
  192. package/lib/internal/platform-pr-review.ts +14 -3
  193. package/lib/internal/platform-pr.ts +38 -14
  194. package/lib/internal/platform-release-notes.ts +5 -3
  195. package/lib/internal/platform-security.ts +83 -0
  196. package/lib/internal/pr-review-grade.ts +10 -3
  197. package/lib/internal/release-workflow.ts +91 -87
  198. package/lib/internal/sandbox-control.ts +181 -6
  199. package/lib/internal/task-activity.ts +12 -4
  200. package/lib/internal/task-artifact.ts +15 -79
  201. package/lib/internal/task-context.ts +3 -1
  202. package/lib/internal/task-create.ts +76 -15
  203. package/lib/internal/task-delivery.ts +3 -1
  204. package/lib/internal/task-event.ts +21 -10
  205. package/lib/internal/task-finalization.ts +3 -1
  206. package/lib/internal/task-invalidation.ts +3 -1
  207. package/lib/internal/task-ledger.ts +11 -1
  208. package/lib/internal/task-lifecycle.ts +3 -1
  209. package/lib/internal/task-operation-registry.ts +590 -0
  210. package/lib/internal/task-orchestration.ts +12 -0
  211. package/lib/internal/task-override.ts +9 -3
  212. package/lib/internal/task-qualification.ts +61 -0
  213. package/lib/internal/task-review.ts +7 -50
  214. package/lib/internal/task-short-id.ts +10 -1
  215. package/lib/internal/task-snapshot.ts +3 -1
  216. package/lib/internal/task-validate.ts +32 -12
  217. package/lib/internal/task-verify.ts +4 -1
  218. package/lib/internal/task-warning.ts +7 -1
  219. package/lib/owned-path.ts +26 -0
  220. package/lib/platform/check-status.ts +9 -0
  221. package/lib/platform/comment-safety.ts +255 -0
  222. package/lib/platform/completion-backfill.ts +1 -1
  223. package/lib/platform/github-data.ts +322 -0
  224. package/lib/platform/github-provider.ts +244 -27
  225. package/lib/platform/issue-comments.ts +159 -35
  226. package/lib/platform/issues.ts +4 -245
  227. package/lib/platform/pr-change-report.ts +6 -12
  228. package/lib/platform/pr-checks.ts +13 -77
  229. package/lib/platform/pr-summary.ts +88 -11
  230. package/lib/platform/provider-contract.ts +78 -3
  231. package/lib/platform/provider-validation.ts +85 -3
  232. package/lib/platform/pull-requests.ts +129 -348
  233. package/lib/platform/repository-metadata.ts +453 -0
  234. package/lib/platform/security-alerts.ts +191 -0
  235. package/lib/platform/snapshots.ts +53 -0
  236. package/lib/platform/verification-sync.ts +171 -177
  237. package/lib/process-data/index.ts +7 -0
  238. package/lib/release/workflow.ts +3 -3
  239. package/lib/run/index.ts +5 -0
  240. package/lib/sandbox/commands/control-status.ts +26 -0
  241. package/lib/sandbox/commands/create.ts +53 -12
  242. package/lib/sandbox/commands/ls.ts +7 -0
  243. package/lib/sandbox/commands/rm.ts +3 -996
  244. package/lib/sandbox/commands/show.ts +16 -0
  245. package/lib/sandbox/control/audit.ts +253 -0
  246. package/lib/sandbox/control/client.ts +150 -13
  247. package/lib/sandbox/control/completed-reentry.ts +98 -0
  248. package/lib/sandbox/control/container-identity.ts +27 -43
  249. package/lib/sandbox/control/executor.ts +262 -37
  250. package/lib/sandbox/control/finalization-response.ts +29 -0
  251. package/lib/sandbox/control/identity-sentinel.ts +137 -0
  252. package/lib/sandbox/control/lifecycle.ts +672 -18
  253. package/lib/sandbox/control/native-file-lock.ts +108 -0
  254. package/lib/sandbox/control/protocol.ts +80 -8
  255. package/lib/sandbox/control/server.ts +779 -82
  256. package/lib/sandbox/control/state.ts +84 -3
  257. package/lib/sandbox/control/task-view.ts +240 -0
  258. package/lib/sandbox/control/task-workflow.ts +95 -0
  259. package/lib/sandbox/control/workflow-executor.ts +102 -0
  260. package/lib/sandbox/engines/authority.ts +365 -0
  261. package/lib/sandbox/engines/docker-exact-discovery.ts +132 -0
  262. package/lib/sandbox/index.ts +9 -2
  263. package/lib/sandbox/recovery.ts +49 -6
  264. package/lib/sandbox/removal.ts +2144 -0
  265. package/lib/sandbox/task-cleanup.ts +209 -0
  266. package/lib/sandbox/workspace-view.ts +26 -18
  267. package/lib/sandbox/worktree-safety.ts +4 -1
  268. package/lib/server/index.ts +10 -0
  269. package/lib/task/activity-intent.ts +2 -2
  270. package/lib/task/activity-log.ts +27 -19
  271. package/lib/task/artifact-command.ts +104 -0
  272. package/lib/task/artifact-lifecycle.ts +66 -39
  273. package/lib/task/artifact-name.ts +55 -0
  274. package/lib/task/artifact-operations.ts +580 -0
  275. package/lib/task/artifact-receipts.ts +3 -14
  276. package/lib/task/artifact-repair-intent.ts +152 -0
  277. package/lib/task/artifact-schema.ts +163 -0
  278. package/lib/task/capabilities.ts +82 -22
  279. package/lib/task/codex-orchestration.ts +85 -24
  280. package/lib/task/commit-identity.ts +31 -0
  281. package/lib/task/commit-operation.ts +2 -16
  282. package/lib/task/control-authority.ts +513 -6
  283. package/lib/task/control-recovery.ts +174 -0
  284. package/lib/task/create-service.ts +131 -6
  285. package/lib/task/create.ts +83 -1
  286. package/lib/task/decision-details.ts +3 -71
  287. package/lib/task/events.ts +292 -75
  288. package/lib/task/finalization-state.ts +11 -0
  289. package/lib/task/human-override.ts +3 -8
  290. package/lib/task/implementation-inputs.ts +3 -3
  291. package/lib/task/index.ts +16 -3
  292. package/lib/task/intermediate-cleanup.ts +455 -0
  293. package/lib/task/invalidation.ts +18 -62
  294. package/lib/task/ledger-intents.ts +7 -24
  295. package/lib/task/ledger.ts +1 -3
  296. package/lib/task/lifecycle.ts +17 -1
  297. package/lib/task/local-artifact-finalization.ts +223 -388
  298. package/lib/task/manual-validation-completion.ts +40 -0
  299. package/lib/task/manual-validation-evidence.ts +209 -0
  300. package/lib/task/manual-validation-receipt.ts +152 -0
  301. package/lib/task/manual-validation-shared.ts +50 -0
  302. package/lib/task/manual-validation-transaction.ts +304 -0
  303. package/lib/task/markdown.ts +70 -0
  304. package/lib/task/orchestration.ts +191 -68
  305. package/lib/task/qualification-audit.ts +588 -0
  306. package/lib/task/qualification-confirmation.ts +117 -0
  307. package/lib/task/qualification-intents.ts +150 -0
  308. package/lib/task/review-artifacts.ts +21 -41
  309. package/lib/task/review-command.ts +23 -0
  310. package/lib/task/review-finalization.ts +154 -153
  311. package/lib/task/review-fingerprint.ts +2 -2
  312. package/lib/task/sections.ts +33 -52
  313. package/lib/task/verification-config.ts +12 -0
  314. package/lib/task/verification-engine.ts +165 -370
  315. package/lib/task/verification-types.ts +17 -0
  316. package/lib/task/verification.ts +1 -1
  317. package/lib/task/workflow-command.ts +18 -0
  318. package/lib/task/workflow-warning-intents.ts +0 -13
  319. package/lib/task/workflow-warnings.ts +21 -2
  320. package/lib/update.ts +6 -0
  321. package/package.json +7 -2
  322. package/templates/.agents/hooks/lifecycle-delegation.js +6 -6
  323. package/templates/.agents/rules/README.en.md +2 -0
  324. package/templates/.agents/rules/README.zh-CN.md +2 -0
  325. package/templates/.agents/rules/decision-qualification.en.md +22 -0
  326. package/templates/.agents/rules/decision-qualification.zh-CN.md +22 -0
  327. package/templates/.agents/rules/evidence-reporting.en.md +28 -0
  328. package/templates/.agents/rules/evidence-reporting.zh-CN.md +28 -0
  329. package/templates/.agents/rules/label-milestone-setup.en.md +7 -7
  330. package/templates/.agents/rules/label-milestone-setup.zh-CN.md +7 -7
  331. package/templates/.agents/rules/local-artifact-repair.en.md +5 -1
  332. package/templates/.agents/rules/local-artifact-repair.zh-CN.md +5 -1
  333. package/templates/.agents/rules/security-alerts.en.md +8 -8
  334. package/templates/.agents/rules/security-alerts.zh-CN.md +8 -8
  335. package/templates/.agents/rules/testing-discipline.en.md +2 -0
  336. package/templates/.agents/rules/testing-discipline.zh-CN.md +2 -0
  337. package/templates/.agents/scripts/package.json +3 -0
  338. package/templates/.agents/skills/analyze-task/SKILL.en.md +15 -2
  339. package/templates/.agents/skills/analyze-task/SKILL.zh-CN.md +19 -2
  340. package/templates/.agents/skills/analyze-task/config/verify.en.json +1 -12
  341. package/templates/.agents/skills/analyze-task/config/verify.zh-CN.json +1 -12
  342. package/templates/.agents/skills/close-codescan/SKILL.en.md +2 -2
  343. package/templates/.agents/skills/close-codescan/SKILL.zh-CN.md +2 -2
  344. package/templates/.agents/skills/close-dependabot/SKILL.en.md +2 -2
  345. package/templates/.agents/skills/close-dependabot/SKILL.zh-CN.md +2 -2
  346. package/templates/.agents/skills/code-task/SKILL.en.md +14 -1
  347. package/templates/.agents/skills/code-task/SKILL.zh-CN.md +14 -1
  348. package/templates/.agents/skills/code-task/config/verify.en.json +5 -14
  349. package/templates/.agents/skills/code-task/config/verify.zh-CN.json +5 -14
  350. package/templates/.agents/skills/code-task/reference/code-rules.en.md +5 -4
  351. package/templates/.agents/skills/code-task/reference/code-rules.zh-CN.md +5 -4
  352. package/templates/.agents/skills/code-task/reference/dual-mode.en.md +1 -1
  353. package/templates/.agents/skills/code-task/reference/dual-mode.zh-CN.md +1 -1
  354. package/templates/.agents/skills/code-task/reference/fix-mode.en.md +3 -1
  355. package/templates/.agents/skills/code-task/reference/fix-mode.zh-CN.md +3 -1
  356. package/templates/.agents/skills/code-task/reference/report-template.en.md +42 -15
  357. package/templates/.agents/skills/code-task/reference/report-template.zh-CN.md +39 -12
  358. package/templates/.agents/skills/complete-manual-validation/SKILL.en.md +37 -21
  359. package/templates/.agents/skills/complete-manual-validation/SKILL.zh-CN.md +37 -21
  360. package/templates/.agents/skills/complete-manual-validation/config/verify.en.json +1 -1
  361. package/templates/.agents/skills/complete-manual-validation/config/verify.zh-CN.json +1 -1
  362. package/templates/.agents/skills/complete-manual-validation/reference/report-template.en.md +4 -0
  363. package/templates/.agents/skills/complete-manual-validation/reference/report-template.zh-CN.md +4 -0
  364. package/templates/.agents/skills/complete-manual-validation/reference/summary-update.en.md +16 -7
  365. package/templates/.agents/skills/complete-manual-validation/reference/summary-update.zh-CN.md +16 -7
  366. package/templates/.agents/skills/complete-task/SKILL.en.md +10 -23
  367. package/templates/.agents/skills/complete-task/SKILL.zh-CN.md +10 -23
  368. package/templates/.agents/skills/create-task/SKILL.en.md +8 -0
  369. package/templates/.agents/skills/create-task/SKILL.zh-CN.md +7 -0
  370. package/templates/.agents/skills/import-codescan/SKILL.en.md +1 -1
  371. package/templates/.agents/skills/import-codescan/SKILL.zh-CN.md +1 -1
  372. package/templates/.agents/skills/import-dependabot/SKILL.en.md +1 -1
  373. package/templates/.agents/skills/import-dependabot/SKILL.zh-CN.md +1 -1
  374. package/templates/.agents/skills/init-labels/SKILL.en.md +8 -8
  375. package/templates/.agents/skills/init-labels/SKILL.zh-CN.md +8 -8
  376. package/templates/.agents/skills/init-milestones/SKILL.en.md +9 -11
  377. package/templates/.agents/skills/init-milestones/SKILL.zh-CN.md +9 -11
  378. package/templates/.agents/skills/plan-task/SKILL.en.md +14 -1
  379. package/templates/.agents/skills/plan-task/SKILL.zh-CN.md +16 -1
  380. package/templates/.agents/skills/plan-task/config/verify.en.json +4 -14
  381. package/templates/.agents/skills/plan-task/config/verify.zh-CN.json +4 -14
  382. package/templates/.agents/skills/release/SKILL.en.md +7 -3
  383. package/templates/.agents/skills/release/SKILL.zh-CN.md +7 -3
  384. package/templates/.agents/skills/review-analysis/SKILL.en.md +14 -1
  385. package/templates/.agents/skills/review-analysis/SKILL.zh-CN.md +14 -1
  386. package/templates/.agents/skills/review-analysis/config/verify.en.json +1 -18
  387. package/templates/.agents/skills/review-analysis/config/verify.zh-CN.json +1 -18
  388. package/templates/.agents/skills/review-analysis/reference/report-template.en.md +50 -23
  389. package/templates/.agents/skills/review-analysis/reference/report-template.zh-CN.md +45 -18
  390. package/templates/.agents/skills/review-code/SKILL.en.md +14 -1
  391. package/templates/.agents/skills/review-code/SKILL.zh-CN.md +14 -1
  392. package/templates/.agents/skills/review-code/config/verify.en.json +1 -23
  393. package/templates/.agents/skills/review-code/config/verify.zh-CN.json +1 -23
  394. package/templates/.agents/skills/review-code/reference/report-template.en.md +49 -22
  395. package/templates/.agents/skills/review-code/reference/report-template.zh-CN.md +45 -18
  396. package/templates/.agents/skills/review-plan/SKILL.en.md +14 -1
  397. package/templates/.agents/skills/review-plan/SKILL.zh-CN.md +14 -1
  398. package/templates/.agents/skills/review-plan/config/verify.en.json +1 -18
  399. package/templates/.agents/skills/review-plan/config/verify.zh-CN.json +1 -18
  400. package/templates/.agents/skills/review-plan/reference/report-template.en.md +49 -22
  401. package/templates/.agents/skills/review-plan/reference/report-template.zh-CN.md +45 -18
  402. package/templates/.agents/skills/review-pr/SKILL.en.md +5 -1
  403. package/templates/.agents/skills/review-pr/SKILL.zh-CN.md +5 -1
  404. package/templates/.agents/skills/review-pr/reference/report-template.en.md +4 -3
  405. package/templates/.agents/skills/review-pr/reference/report-template.zh-CN.md +4 -3
  406. package/templates/.agents/skills/run-manual-validation/SKILL.en.md +25 -4
  407. package/templates/.agents/skills/run-manual-validation/SKILL.zh-CN.md +25 -4
  408. package/templates/.agents/skills/run-manual-validation/reference/discovery-and-execution.en.md +4 -1
  409. package/templates/.agents/skills/run-manual-validation/reference/discovery-and-execution.zh-CN.md +4 -1
  410. package/templates/.agents/skills/run-manual-validation/reference/report-template.en.md +5 -1
  411. package/templates/.agents/skills/run-manual-validation/reference/report-template.zh-CN.md +5 -1
  412. package/templates/.agents/skills/run-task/SKILL.en.md +1 -1
  413. package/templates/.agents/skills/run-task/SKILL.zh-CN.md +1 -1
  414. package/templates/.agents/skills/test/SKILL.en.md +29 -7
  415. package/templates/.agents/skills/test/SKILL.zh-CN.md +29 -7
  416. package/templates/.agents/skills/update-agent-infra/scripts/sync-templates.js +14 -0
  417. package/templates/.agents/templates/task.en.md +6 -0
  418. package/templates/.agents/templates/task.zh-CN.md +6 -0
  419. package/templates/.claude/commands/run-manual-validation.en.md +1 -1
  420. package/templates/.claude/commands/run-manual-validation.zh-CN.md +1 -1
  421. package/dist/lib/platform/adapters.js +0 -56
  422. package/lib/platform/adapters.ts +0 -197
  423. package/templates/.agents/scripts/security-alerts.github.sh +0 -151
  424. package/templates/.agents/scripts/security-alerts.sh +0 -31
  425. package/templates/.agents/skills/init-labels/scripts/init-labels.github.sh +0 -146
  426. package/templates/.agents/skills/init-labels/scripts/init-labels.sh +0 -32
  427. package/templates/.agents/skills/init-milestones/scripts/init-milestones.github.sh +0 -294
  428. package/templates/.agents/skills/init-milestones/scripts/init-milestones.sh +0 -6
@@ -0,0 +1,365 @@
1
+ import { createHash } from 'node:crypto';
2
+ import type { SpawnSyncReturns } from 'node:child_process';
3
+ import { runProbe } from '../shell.ts';
4
+ import { getAdapter } from './index.ts';
5
+
6
+ export type SandboxAuthorityEvidenceV1 = Readonly<{
7
+ version: 1;
8
+ provider: string;
9
+ lockDomain: string;
10
+ routeKind: 'default' | 'context' | 'endpoint' | 'wsl2';
11
+ routeSelector: Readonly<Record<string, string>>;
12
+ normalizedEndpoint: string;
13
+ endpointFingerprint: string;
14
+ daemonIdentity: Readonly<{
15
+ kind: 'docker-server-id';
16
+ fingerprint: string;
17
+ }>;
18
+ apiVersion: Readonly<{ major: number; minor: number }>;
19
+ authorityFingerprint: string;
20
+ }>;
21
+
22
+ export type AuthorityProbe = (cmd: string, args: string[], options?: { timeout?: number }) => SpawnSyncReturns<string | Buffer>;
23
+
24
+ export type AuthorityCaptureOptions = Readonly<{
25
+ probe?: AuthorityProbe;
26
+ timeoutMs?: number;
27
+ env?: NodeJS.ProcessEnv;
28
+ lockDomain?: string;
29
+ route?: SandboxAuthorityEvidenceV1;
30
+ }>;
31
+
32
+ export type AuthorityVerification = Readonly<{
33
+ state: 'verified' | 'unknown' | 'conflict';
34
+ reason?: string;
35
+ }>;
36
+
37
+ function digest(value: string): string {
38
+ return createHash('sha256').update(value).digest('hex');
39
+ }
40
+
41
+ function canonical(value: unknown): string {
42
+ if (Array.isArray(value)) return `[${value.map(canonical).join(',')}]`;
43
+ if (!value || typeof value !== 'object') return JSON.stringify(value) ?? 'null';
44
+ return `{${Object.entries(value as Record<string, unknown>)
45
+ .sort(([left], [right]) => left.localeCompare(right))
46
+ .map(([key, entry]) => `${JSON.stringify(key)}:${canonical(entry)}`)
47
+ .join(',')}}`;
48
+ }
49
+
50
+ function text(value: string | Buffer | null | undefined): string {
51
+ return typeof value === 'string' ? value : value?.toString('utf8') ?? '';
52
+ }
53
+
54
+ function safeEndpoint(value: string): string {
55
+ if (!value) return 'docker-default';
56
+ try {
57
+ const parsed = new URL(value);
58
+ const hasUnreplayableParts = Boolean(parsed.username || parsed.password || parsed.search || parsed.hash);
59
+ parsed.username = '';
60
+ parsed.password = '';
61
+ parsed.search = '';
62
+ parsed.hash = '';
63
+ const normalized = parsed.toString().replace(/\/$/u, '');
64
+ return hasUnreplayableParts
65
+ ? `${normalized}#<redacted>`
66
+ : normalized;
67
+ } catch {
68
+ const normalized = value
69
+ .replace(/\/\/[^/@]+@/u, '//<redacted>@')
70
+ .replace(/[?&](?:token|password|secret|key)=[^&]*/giu, '');
71
+ return /[@?#]/u.test(value) ? `${normalized.replace(/[?#].*$/u, '')}#<redacted>` : normalized;
72
+ }
73
+ }
74
+
75
+ function lockDomainFor(engine: string, endpoint: string): string {
76
+ return digest(`${process.platform}\0${process.arch}\0${engine}\0${endpoint}`);
77
+ }
78
+
79
+ type SandboxAuthorityRoute = Readonly<{
80
+ kind: SandboxAuthorityEvidenceV1['routeKind'];
81
+ selector: Readonly<Record<string, string>>;
82
+ endpoint: string;
83
+ command: (args: string[]) => { cmd: string; args: string[] };
84
+ }>;
85
+
86
+ function contextRoute(context: string): SandboxAuthorityRoute {
87
+ return {
88
+ kind: 'context',
89
+ selector: { context },
90
+ endpoint: `docker-context://${context}`,
91
+ command: (args) => ({ cmd: 'docker', args: ['--context', context, ...args] })
92
+ };
93
+ }
94
+
95
+ function endpointRoute(endpoint: string, rawEndpoint: string, unreplayable?: string): SandboxAuthorityRoute {
96
+ return {
97
+ kind: 'endpoint',
98
+ selector: {
99
+ source: 'DOCKER_HOST',
100
+ endpoint,
101
+ ...(unreplayable ? { unreplayable } : {})
102
+ },
103
+ endpoint,
104
+ command: (args) => ({ cmd: 'docker', args: ['--host', rawEndpoint, ...args] })
105
+ };
106
+ }
107
+
108
+ function tlsEnvironmentReason(env: NodeJS.ProcessEnv): string | undefined {
109
+ return env.DOCKER_TLS_VERIFY || env.DOCKER_CERT_PATH || env.DOCKER_TLS
110
+ ? 'tls-environment'
111
+ : undefined;
112
+ }
113
+
114
+ function routeFor(engine: string, env: NodeJS.ProcessEnv): SandboxAuthorityRoute {
115
+ const context = getAdapter(engine).dockerContext;
116
+ if (context) {
117
+ return contextRoute(context);
118
+ }
119
+ if (env.DOCKER_CONTEXT) {
120
+ return contextRoute(env.DOCKER_CONTEXT);
121
+ }
122
+ if (env.DOCKER_HOST) {
123
+ return endpointRoute(safeEndpoint(env.DOCKER_HOST), env.DOCKER_HOST, tlsEnvironmentReason(env));
124
+ }
125
+ return contextRoute('default');
126
+ }
127
+
128
+ function validRouteName(value: string): boolean {
129
+ return /^[A-Za-z0-9][A-Za-z0-9_.@/-]*$/u.test(value);
130
+ }
131
+
132
+ function resolveWslDistro(probe: AuthorityProbe, timeoutMs?: number, explicit?: string): string {
133
+ if (explicit) {
134
+ if (!validRouteName(explicit)) throw new Error('SANDBOX_AUTHORITY_ROUTE_CAPTURE_FAILED');
135
+ return explicit;
136
+ }
137
+ const result = probe(
138
+ 'wsl.exe',
139
+ ['--list', '--verbose'],
140
+ timeoutMs === undefined ? {} : { timeout: timeoutMs }
141
+ );
142
+ if (result.status !== 0) throw new Error('SANDBOX_AUTHORITY_ROUTE_CAPTURE_FAILED');
143
+ const output = text(result.stdout).replace(/\0/gu, '').replace(/\r/gu, '');
144
+ const defaultLine = output.split('\n').find((line) => /^\s*\*\s+/u.test(line));
145
+ const distro = defaultLine?.replace(/^\s*\*\s+/u, '').replace(/\s+(?:Running|Stopped)\s+\d+\s*$/iu, '').trim();
146
+ if (!distro || !validRouteName(distro)) throw new Error('SANDBOX_AUTHORITY_ROUTE_CAPTURE_FAILED');
147
+ return distro;
148
+ }
149
+
150
+ function wslRoute(
151
+ distro: string,
152
+ context: string | undefined,
153
+ env: NodeJS.ProcessEnv
154
+ ): SandboxAuthorityRoute {
155
+ const endpoint = env.DOCKER_HOST
156
+ ? safeEndpoint(env.DOCKER_HOST)
157
+ : `wsl://${distro}/${context ?? 'default'}`;
158
+ const unreplayable = tlsEnvironmentReason(env)
159
+ ?? (endpoint.includes('<redacted>') ? 'ambient-endpoint-secret' : undefined);
160
+ return {
161
+ kind: 'wsl2',
162
+ selector: {
163
+ distro,
164
+ ...(context ? { context } : {}),
165
+ ...(unreplayable ? { unreplayable } : {}),
166
+ ...(env.DOCKER_HOST ? { endpoint } : {})
167
+ },
168
+ endpoint,
169
+ command: (args) => ({
170
+ cmd: 'wsl.exe',
171
+ args: [
172
+ '--distribution', distro,
173
+ '--exec', 'docker',
174
+ ...(context ? ['--context', context] : []),
175
+ ...(env.DOCKER_HOST ? ['--host', env.DOCKER_HOST] : []),
176
+ ...args
177
+ ]
178
+ })
179
+ };
180
+ }
181
+
182
+ function currentWslRoute(
183
+ env: NodeJS.ProcessEnv,
184
+ probe: AuthorityProbe,
185
+ timeoutMs?: number
186
+ ): SandboxAuthorityRoute {
187
+ const distro = resolveWslDistro(probe, timeoutMs, env.WSL_DISTRO_NAME);
188
+ let context = env.DOCKER_CONTEXT;
189
+ if (!context && !env.DOCKER_HOST) {
190
+ const result = probe(
191
+ 'wsl.exe',
192
+ ['--distribution', distro, '--exec', 'docker', 'context', 'show'],
193
+ timeoutMs === undefined ? {} : { timeout: timeoutMs }
194
+ );
195
+ context = text(result.stdout).trim();
196
+ if (result.status !== 0 || !validRouteName(context)) {
197
+ throw new Error('SANDBOX_AUTHORITY_ROUTE_CAPTURE_FAILED');
198
+ }
199
+ }
200
+ return wslRoute(distro, context, env);
201
+ }
202
+
203
+ function routeFromEvidence(evidence: SandboxAuthorityEvidenceV1): SandboxAuthorityRoute {
204
+ return {
205
+ kind: evidence.routeKind,
206
+ selector: evidence.routeSelector,
207
+ endpoint: evidence.normalizedEndpoint,
208
+ command: (args) => commandForSandboxAuthority(evidence, 'docker', args)
209
+ };
210
+ }
211
+
212
+ function currentContextRoute(
213
+ engine: string,
214
+ env: NodeJS.ProcessEnv,
215
+ probe: AuthorityProbe,
216
+ timeoutMs?: number
217
+ ): SandboxAuthorityRoute {
218
+ const configured = routeFor(engine, env);
219
+ if (configured.kind !== 'context' || configured.selector.context !== 'default') return configured;
220
+ if (env.DOCKER_CONTEXT || env.DOCKER_HOST || getAdapter(engine).dockerContext) return configured;
221
+ const result = probe('docker', ['context', 'show'], timeoutMs === undefined ? {} : { timeout: timeoutMs });
222
+ const context = text(result.stdout).trim();
223
+ if (result.status !== 0 || !/^[A-Za-z0-9][A-Za-z0-9_.@/-]*$/u.test(context)) {
224
+ throw new Error('SANDBOX_AUTHORITY_ROUTE_CAPTURE_FAILED');
225
+ }
226
+ return contextRoute(context);
227
+ }
228
+
229
+ /** Build a command using only the route persisted with the sandbox authority. */
230
+ export function commandForSandboxAuthority(
231
+ evidence: SandboxAuthorityEvidenceV1,
232
+ cmd: string,
233
+ args: string[] = []
234
+ ): { cmd: string; args: string[] } {
235
+ if (cmd !== 'docker') {
236
+ throw new Error('SANDBOX_AUTHORITY_ROUTE_UNSUPPORTED');
237
+ }
238
+ if (evidence.routeKind === 'context') {
239
+ const context = evidence.routeSelector.context;
240
+ if (!context) throw new Error('SANDBOX_AUTHORITY_ROUTE_INVALID');
241
+ return { cmd, args: ['--context', context, ...args] };
242
+ }
243
+ if (evidence.routeKind === 'endpoint') {
244
+ const endpoint = evidence.routeSelector.endpoint;
245
+ if (!endpoint || endpoint.includes('<redacted>') || evidence.routeSelector.unreplayable) {
246
+ throw new Error('SANDBOX_AUTHORITY_ROUTE_UNREPLAYABLE');
247
+ }
248
+ return { cmd, args: ['--host', endpoint, ...args] };
249
+ }
250
+ if (evidence.routeKind === 'wsl2') {
251
+ const distro = evidence.routeSelector.distro;
252
+ const context = evidence.routeSelector.context;
253
+ const endpoint = evidence.routeSelector.endpoint;
254
+ if (!distro || evidence.routeSelector.unreplayable || endpoint?.includes('<redacted>')) {
255
+ throw new Error('SANDBOX_AUTHORITY_ROUTE_UNREPLAYABLE');
256
+ }
257
+ return {
258
+ cmd: 'wsl.exe',
259
+ args: [
260
+ '--distribution', distro,
261
+ '--exec', cmd,
262
+ ...(context ? ['--context', context] : []),
263
+ ...(endpoint && !endpoint.startsWith('wsl://') ? ['--host', endpoint] : []),
264
+ ...args
265
+ ]
266
+ };
267
+ }
268
+ return { cmd, args: ['--context', 'default', ...args] };
269
+ }
270
+
271
+ export function authorityVersionArgs(): string[] {
272
+ return ['version', '--format', '{{json .Server}}'];
273
+ }
274
+
275
+ export function captureSandboxAuthority(
276
+ engine: string,
277
+ options: AuthorityCaptureOptions = {}
278
+ ): SandboxAuthorityEvidenceV1 {
279
+ const env = options.env ?? process.env;
280
+ const probe = options.probe ?? runProbe;
281
+ const route = options.route
282
+ ? routeFromEvidence(options.route)
283
+ : engine === 'wsl2'
284
+ ? currentWslRoute(env, probe, options.timeoutMs)
285
+ : currentContextRoute(engine, env, probe, options.timeoutMs);
286
+ function probeJson(args: string[]): unknown {
287
+ const command = options.route
288
+ ? commandForSandboxAuthority(options.route, 'docker', args)
289
+ : route.command(args);
290
+ try {
291
+ const response = probe(
292
+ command.cmd,
293
+ command.args,
294
+ options.timeoutMs === undefined ? {} : { timeout: options.timeoutMs }
295
+ );
296
+ if (response.status !== 0) throw new Error('SANDBOX_AUTHORITY_CAPTURE_FAILED');
297
+ return JSON.parse(text(response.stdout).trim());
298
+ } catch {
299
+ throw new Error('SANDBOX_AUTHORITY_CAPTURE_FAILED');
300
+ }
301
+ }
302
+ const server = probeJson(authorityVersionArgs());
303
+ if (!server || typeof server !== 'object' || Array.isArray(server)) {
304
+ throw new Error('SANDBOX_AUTHORITY_CAPTURE_FAILED');
305
+ }
306
+ const record = server as Record<string, unknown>;
307
+ const api = typeof record.ApiVersion === 'string' ? /^(\d+)\.(\d+)$/u.exec(record.ApiVersion) : null;
308
+ if (!api) throw new Error('SANDBOX_AUTHORITY_CAPTURE_FAILED');
309
+ // Docker version exposes API metadata; the daemon ID belongs to Docker info.
310
+ const serverId = probeJson(['info', '--format', '{{json .ID}}']);
311
+ if (typeof serverId !== 'string' || !serverId.trim()) throw new Error('SANDBOX_AUTHORITY_CAPTURE_FAILED');
312
+ const normalizedEndpoint = route.endpoint;
313
+ const evidenceWithoutFingerprint = {
314
+ version: 1 as const,
315
+ provider: engine,
316
+ lockDomain: options.lockDomain ?? options.route?.lockDomain ?? lockDomainFor(engine, normalizedEndpoint),
317
+ routeKind: route.kind,
318
+ routeSelector: route.selector,
319
+ normalizedEndpoint,
320
+ endpointFingerprint: digest(normalizedEndpoint),
321
+ daemonIdentity: { kind: 'docker-server-id' as const, fingerprint: digest(serverId) },
322
+ apiVersion: { major: Number(api[1]), minor: Number(api[2]) }
323
+ };
324
+ return {
325
+ ...evidenceWithoutFingerprint,
326
+ authorityFingerprint: digest(canonical(evidenceWithoutFingerprint))
327
+ };
328
+ }
329
+
330
+ export function verifySandboxAuthority(
331
+ expected: SandboxAuthorityEvidenceV1 | null | undefined,
332
+ actual: SandboxAuthorityEvidenceV1 | null
333
+ ): AuthorityVerification {
334
+ if (!expected) return { state: 'unknown', reason: 'SANDBOX_AUTHORITY_EVIDENCE_MISSING' };
335
+ if (!actual) return { state: 'unknown', reason: 'SANDBOX_AUTHORITY_UNAVAILABLE' };
336
+ if (expected.authorityFingerprint !== actual.authorityFingerprint
337
+ || expected.lockDomain !== actual.lockDomain
338
+ || expected.routeKind !== actual.routeKind
339
+ || expected.endpointFingerprint !== actual.endpointFingerprint
340
+ || expected.daemonIdentity.fingerprint !== actual.daemonIdentity.fingerprint) {
341
+ return { state: 'conflict', reason: 'SANDBOX_AUTHORITY_DRIFT' };
342
+ }
343
+ return { state: 'verified' };
344
+ }
345
+
346
+ export function isSandboxAuthorityEvidence(value: unknown): value is SandboxAuthorityEvidenceV1 {
347
+ if (!value || typeof value !== 'object' || Array.isArray(value)) return false;
348
+ const evidence = value as Record<string, unknown>;
349
+ const selector = evidence.routeSelector as Record<string, unknown> | null;
350
+ const daemon = evidence.daemonIdentity as Record<string, unknown> | null;
351
+ const api = evidence.apiVersion as Record<string, unknown> | null;
352
+ return evidence.version === 1
353
+ && typeof evidence.provider === 'string'
354
+ && /^[a-f0-9]{64}$/u.test(evidence.lockDomain as string)
355
+ && ['default', 'context', 'endpoint', 'wsl2'].includes(evidence.routeKind as string)
356
+ && !!selector && typeof selector === 'object' && !Array.isArray(selector)
357
+ && Object.values(selector).every((entry) => typeof entry === 'string')
358
+ && typeof evidence.normalizedEndpoint === 'string'
359
+ && /^[a-f0-9]{64}$/u.test(evidence.endpointFingerprint as string)
360
+ && !!daemon && daemon.kind === 'docker-server-id'
361
+ && /^[a-f0-9]{64}$/u.test(daemon.fingerprint as string)
362
+ && !!api && Number.isSafeInteger(api.major) && (api.major as number) >= 0
363
+ && Number.isSafeInteger(api.minor) && (api.minor as number) >= 0
364
+ && /^[a-f0-9]{64}$/u.test(evidence.authorityFingerprint as string);
365
+ }
@@ -0,0 +1,132 @@
1
+ import type { SpawnSyncReturns } from 'node:child_process';
2
+ import { commandForEngine, runProbe } from '../shell.ts';
3
+ import { commandForSandboxAuthority, type SandboxAuthorityEvidenceV1 } from './authority.ts';
4
+
5
+ type ProbeResult = SpawnSyncReturns<string | Buffer>;
6
+ type Probe = (cmd: string, args: string[], options?: { timeout?: number }) => ProbeResult;
7
+
8
+ export type ExactContainerDiscovery =
9
+ | Readonly<{ state: 'absent'; id: string; evidence: 'exact-empty' }>
10
+ | Readonly<{
11
+ state: 'found';
12
+ id: string;
13
+ running: boolean;
14
+ labels: Readonly<Record<string, string>>;
15
+ evidence: 'exact-present';
16
+ }>
17
+ | Readonly<{ state: 'conflict'; reason: string }>
18
+ | Readonly<{ state: 'unknown'; reason: string }>;
19
+
20
+ export type ExactContainerDiscoveryOptions = Readonly<{
21
+ probe?: Probe;
22
+ timeoutMs?: number;
23
+ authority?: SandboxAuthorityEvidenceV1;
24
+ }>;
25
+
26
+ const FULL_CONTAINER_ID = /^[a-f0-9]{64}$/u;
27
+
28
+ function outputText(value: string | Buffer | null | undefined): string {
29
+ return typeof value === 'string' ? value : value?.toString('utf8') ?? '';
30
+ }
31
+
32
+ function unknown(reason: string): ExactContainerDiscovery {
33
+ return { state: 'unknown', reason };
34
+ }
35
+
36
+ export function buildExactContainerListArgs(id: string): string[] {
37
+ if (!FULL_CONTAINER_ID.test(id)) throw new Error('SANDBOX_CONTAINER_ID_NOT_FULL');
38
+ return ['container', 'ls', '--all', '--no-trunc', '--filter', `id=${id}`, '--format', '{{.ID}}'];
39
+ }
40
+
41
+ function buildExactContainerInspectArgs(id: string): string[] {
42
+ return ['container', 'inspect', '--format', '{{json .}}', id];
43
+ }
44
+
45
+ function runDockerProbe(
46
+ engine: string,
47
+ args: string[],
48
+ options: ExactContainerDiscoveryOptions
49
+ ): ProbeResult {
50
+ const command = options.authority
51
+ ? commandForSandboxAuthority(options.authority, 'docker', args)
52
+ : commandForEngine(engine, 'docker', args);
53
+ return (options.probe ?? runProbe)(
54
+ command.cmd,
55
+ command.args,
56
+ options.timeoutMs === undefined ? {} : { timeout: options.timeoutMs }
57
+ );
58
+ }
59
+
60
+ function parseExactList(stdout: string, expectedId: string): 'empty' | 'present' | 'conflict' | string {
61
+ const rows = stdout.split(/\r?\n/u).map((row) => row.trim()).filter(Boolean);
62
+ if (rows.length === 0) return 'empty';
63
+ if (rows.length !== 1) return 'exact container query returned multiple rows';
64
+ if (!FULL_CONTAINER_ID.test(rows[0]!)) return 'exact container query returned a malformed ID';
65
+ if (rows[0] !== expectedId) return 'conflict';
66
+ return 'present';
67
+ }
68
+
69
+ function parseExactInspect(stdout: string, expectedId: string): ExactContainerDiscovery {
70
+ let value: unknown;
71
+ try {
72
+ value = JSON.parse(stdout);
73
+ } catch {
74
+ return unknown('exact container inspection returned malformed JSON');
75
+ }
76
+ if (!value || typeof value !== 'object' || Array.isArray(value)) {
77
+ return unknown('exact container inspection returned an incomplete object');
78
+ }
79
+ const inspected = value as {
80
+ Id?: unknown;
81
+ State?: { Running?: unknown };
82
+ Config?: { Labels?: unknown };
83
+ };
84
+ if (inspected.Id !== expectedId || typeof inspected.State?.Running !== 'boolean'
85
+ || !inspected.Config?.Labels || typeof inspected.Config.Labels !== 'object'
86
+ || Array.isArray(inspected.Config.Labels)) {
87
+ return unknown('exact container inspection identity is incomplete');
88
+ }
89
+ const rawLabels = inspected.Config.Labels as Record<string, unknown>;
90
+ if (Object.values(rawLabels).some((value) => typeof value !== 'string')) {
91
+ return unknown('exact container inspection labels are malformed');
92
+ }
93
+ return {
94
+ state: 'found',
95
+ id: expectedId,
96
+ running: inspected.State.Running,
97
+ labels: { ...rawLabels as Record<string, string> },
98
+ evidence: 'exact-present'
99
+ };
100
+ }
101
+
102
+ export function discoverExactContainer(
103
+ engine: string,
104
+ id: string,
105
+ options: ExactContainerDiscoveryOptions = {}
106
+ ): ExactContainerDiscovery {
107
+ if (!FULL_CONTAINER_ID.test(id)) return unknown('container identity is not a full 64-character ID');
108
+ let listed: ProbeResult;
109
+ try {
110
+ listed = runDockerProbe(engine, buildExactContainerListArgs(id), options);
111
+ } catch {
112
+ return unknown('exact container query failed before producing a result');
113
+ }
114
+ if (listed.status !== 0) {
115
+ return unknown(`exact container query exited with ${listed.status ?? 'unknown status'}`);
116
+ }
117
+ const listState = parseExactList(outputText(listed.stdout), id);
118
+ if (listState === 'empty') return { state: 'absent', id, evidence: 'exact-empty' };
119
+ if (listState === 'conflict') return { state: 'conflict', reason: 'exact container query returned a different ID' };
120
+ if (listState !== 'present') return unknown(listState);
121
+
122
+ let inspected: ProbeResult;
123
+ try {
124
+ inspected = runDockerProbe(engine, buildExactContainerInspectArgs(id), options);
125
+ } catch {
126
+ return unknown('exact container inspection failed before producing a result');
127
+ }
128
+ if (inspected.status !== 0) {
129
+ return unknown(`exact container inspection exited with ${inspected.status ?? 'unknown status'}`);
130
+ }
131
+ return parseExactInspect(outputText(inspected.stdout).trim(), id);
132
+ }
@@ -1,3 +1,5 @@
1
+ import { PUBLIC_CLI_ROUTE_SELECTORS } from '../internal/cli-route-inventory.ts';
2
+
1
3
  const USAGE = `Usage: ai sandbox <command> [options]
2
4
 
3
5
  Commands:
@@ -18,14 +20,15 @@ Commands:
18
20
  rm <branch | TASK-id> | --unbound | --purge
19
21
  Remove one sandbox; use a full TASK-id for task-bound
20
22
  cleanup and a branch for branch-only cleanup. --unbound
21
- removes completed task-bound and branch-only sandboxes;
22
- active, blocked, and archive tasks are protected.
23
+ removes completed task-bound and branch-only sandboxes
24
+ plus verified auxiliary state; active, blocked, and archive tasks are protected.
23
25
  start [--recreate] <branch | TASK-id | N>
24
26
  Verify or recover an existing sandbox container;
25
27
  optionally replace only the container on failure
26
28
  vm status|start|stop Manage the sandbox VM (macOS) or check the backend (Windows)
27
29
 
28
30
  Run 'ai sandbox <command> --help' for details.`;
31
+ const SANDBOX_OPERATIONS = PUBLIC_CLI_ROUTE_SELECTORS.sandbox;
29
32
 
30
33
  export async function runSandbox(args: string[]): Promise<void> {
31
34
  const [subcommand, ...rest] = args;
@@ -41,6 +44,10 @@ export async function runSandbox(args: string[]): Promise<void> {
41
44
  return;
42
45
  }
43
46
 
47
+ if (!SANDBOX_OPERATIONS.includes(subcommand as typeof SANDBOX_OPERATIONS[number])) {
48
+ throw new Error(`Unknown sandbox command: ${subcommand}`);
49
+ }
50
+
44
51
  switch (subcommand) {
45
52
  case 'create': {
46
53
  const { create } = await import('./commands/create.ts');
@@ -1,3 +1,4 @@
1
+ import { prepareCompletedReentry, publishCompletedReentry, waitForCompletedReentry } from './control/completed-reentry.ts';
1
2
  import fs from 'node:fs';
2
3
  import path from 'node:path';
3
4
  import { spawn } from 'node:child_process';
@@ -59,9 +60,10 @@ import {
59
60
  type BrokerOwner
60
61
  } from './control/lifecycle.ts';
61
62
  import {
62
- appendSandboxControlAudit,
63
63
  readSandboxControlStatus
64
64
  } from './control/state.ts';
65
+ import { validateSandboxControlIdentity } from './control/identity-sentinel.ts';
66
+ import { appendDiagnosticAudit as appendControlDiagnosticAudit } from './control/audit.ts';
65
67
  import { inspectSandboxControlContainer } from './control/container-identity.ts';
66
68
  import {
67
69
  SANDBOX_CONTROL_FUTURE_SKEW_MS,
@@ -223,6 +225,17 @@ export async function startSandboxControlBroker(repoRoot: string, manifestPath:
223
225
  const internalCli = path.resolve(directory, '..', '..', 'bin', `internal-cli${extension}`);
224
226
  const manifest = readSandboxControlManifest(manifestPath);
225
227
  const root = path.dirname(manifestPath);
228
+ const identity = validateSandboxControlIdentity({
229
+ publicStatusDir: manifest.publicStatusDir,
230
+ root,
231
+ mode: manifest.mode,
232
+ taskId: manifest.taskId,
233
+ generation: manifest.generation,
234
+ controlRootId: manifest.controlRootId
235
+ });
236
+ if (identity.state !== 'valid') {
237
+ throw new Error(`SANDBOX_CONTROL_IDENTITY_${identity.state.replaceAll('-', '_').toUpperCase()}`);
238
+ }
226
239
  if (isSandboxControlRootQuiescing(root)) throw new Error('SANDBOX_CONTROL_QUIESCING');
227
240
  const brokerPath = path.join(root, 'broker.json');
228
241
  let brokerSnapshot: string | null = null;
@@ -295,8 +308,8 @@ export async function startSandboxControlBroker(repoRoot: string, manifestPath:
295
308
  }
296
309
  }
297
310
  if (replacedBrokerRecord) {
298
- appendSandboxControlAudit(manifest, 'broker-observed-crash');
299
- appendSandboxControlAudit(manifest, 'broker-restart');
311
+ appendControlDiagnosticAudit(manifest, 'broker-observed-crash');
312
+ appendControlDiagnosticAudit(manifest, 'broker-restart');
300
313
  }
301
314
  child = spawn(
302
315
  process.execPath,
@@ -337,6 +350,17 @@ async function ensureSandboxControlBroker(params: {
337
350
  });
338
351
  if (!fs.existsSync(control.manifestPath)) return;
339
352
  const validatedManifest = readSandboxControlManifest(control.manifestPath);
353
+ const identity = validateSandboxControlIdentity({
354
+ publicStatusDir: validatedManifest.publicStatusDir,
355
+ root: control.root,
356
+ mode: validatedManifest.mode,
357
+ taskId: validatedManifest.taskId,
358
+ generation: validatedManifest.generation,
359
+ controlRootId: validatedManifest.controlRootId
360
+ });
361
+ if (identity.state !== 'valid') {
362
+ throw new Error(`SANDBOX_CONTROL_IDENTITY_${identity.state.replaceAll('-', '_').toUpperCase()}`);
363
+ }
340
364
  const containerObservation = await inspectSandboxControlContainer(validatedManifest);
341
365
  if (containerObservation.state === 'unknown') {
342
366
  throw new Error(`SANDBOX_CONTROL_CONTAINER_UNKNOWN: ${containerObservation.reason}`);
@@ -653,7 +677,7 @@ function recoveryTaskSources(repoRoot: string, taskId: string): {
653
677
  return { mountPaths, accessiblePaths: [completedSource] };
654
678
  }
655
679
 
656
- function startCompletedSandboxContainer(params: {
680
+ export function startCompletedSandboxContainer(params: {
657
681
  config: SandboxConfig;
658
682
  engine: string;
659
683
  container: string;
@@ -748,7 +772,7 @@ function expectedMounts(params: {
748
772
  ? {}
749
773
  : {
750
774
  runtimeDir: control.runtimeDir,
751
- taskSources: taskSources.mountPaths,
775
+ taskSources: taskSources!.accessiblePaths,
752
776
  taskId: taskId!
753
777
  })
754
778
  }).map((mount) => ({
@@ -757,7 +781,7 @@ function expectedMounts(params: {
757
781
  hostPaths: mount.hostPaths,
758
782
  ...(taskSources !== null && taskId !== null
759
783
  && mount.containerPath === `/workspace/.agents/workspace/active/${taskId}`
760
- ? { sourceAccessiblePaths: taskSources.accessiblePaths }
784
+ ? { sourceAccessiblePaths: taskSources!.accessiblePaths }
761
785
  : {}),
762
786
  expectedRW: !mount.readOnly
763
787
  }));
@@ -1250,7 +1274,17 @@ export async function ensureSandboxReady(params: EnsureSandboxReadyParams): Prom
1250
1274
  throw new Error('Explicit container recreation requested.');
1251
1275
  }
1252
1276
  if (!params.row.running) {
1277
+ let completedReentry: { manifest: SandboxControlManifest; evidence: Awaited<ReturnType<typeof prepareCompletedReentry>> } | null = null;
1253
1278
  if (params.reentry === 'completed' && params.workspace?.mode === 'task-bound') {
1279
+ const control = sandboxControlPaths({
1280
+ base: params.config.controlBase, project: params.config.project,
1281
+ container: params.row.name, identity: params.workspace
1282
+ });
1283
+ const manifest = readSandboxControlManifest(control.manifestPath);
1284
+ if (manifest.taskId !== params.workspace.taskId || manifest.branch !== params.branch) {
1285
+ throw new Error('SANDBOX_COMPLETED_REENTRY_EVIDENCE_INVALID');
1286
+ }
1287
+ completedReentry = { manifest, evidence: await prepareCompletedReentry(manifest) };
1254
1288
  startCompletedSandboxContainer({
1255
1289
  config: params.config,
1256
1290
  engine: params.engine,
@@ -1315,6 +1349,15 @@ export async function ensureSandboxReady(params: EnsureSandboxReadyParams): Prom
1315
1349
  if (final.findings.length > 0) {
1316
1350
  throw new Error(describeFindings(final.findings));
1317
1351
  }
1352
+ if (completedReentry) {
1353
+ const { manifest, evidence } = completedReentry;
1354
+ const current = await prepareCompletedReentry(manifest);
1355
+ if (JSON.stringify(current) !== JSON.stringify(evidence)) {
1356
+ throw new Error('SANDBOX_COMPLETED_REENTRY_EVIDENCE_INVALID');
1357
+ }
1358
+ publishCompletedReentry(manifest, evidence);
1359
+ await waitForCompletedReentry(manifest, evidence);
1360
+ }
1318
1361
  return { container: params.row.name, path: 'recovered', warnings };
1319
1362
  }
1320
1363