agent-hitch 0.2.5 → 0.2.7

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 (395) hide show
  1. package/README.md +14 -1
  2. package/dist/scripts/canary-buildkit-secrets.js +166 -0
  3. package/dist/scripts/canary-buildkit-secrets.js.map +1 -0
  4. package/dist/scripts/canary-collision-domains.js +77 -0
  5. package/dist/scripts/canary-collision-domains.js.map +1 -0
  6. package/dist/scripts/canary-docker-image-observation.js +181 -0
  7. package/dist/scripts/canary-docker-image-observation.js.map +1 -0
  8. package/dist/scripts/canary-gpu-hardware.js +95 -0
  9. package/dist/scripts/canary-gpu-hardware.js.map +1 -0
  10. package/dist/scripts/canary-harbor-load.js +365 -0
  11. package/dist/scripts/canary-harbor-load.js.map +1 -0
  12. package/dist/scripts/canary-packaged-harness.js +75 -0
  13. package/dist/scripts/canary-packaged-harness.js.map +1 -0
  14. package/dist/scripts/canary-resource-load.js +188 -0
  15. package/dist/scripts/canary-resource-load.js.map +1 -0
  16. package/dist/scripts/check-architecture.js +8 -4
  17. package/dist/scripts/check-architecture.js.map +1 -1
  18. package/dist/scripts/docker-reaper-crash-child.js +21 -0
  19. package/dist/scripts/docker-reaper-crash-child.js.map +1 -0
  20. package/dist/src/adapters/catalog.js +1 -0
  21. package/dist/src/adapters/catalog.js.map +1 -1
  22. package/dist/src/adapters/discovery.js +1 -1
  23. package/dist/src/adapters/discovery.js.map +1 -1
  24. package/dist/src/adapters/providers/claude.js +6 -0
  25. package/dist/src/adapters/providers/claude.js.map +1 -1
  26. package/dist/src/adapters/providers/codex.js +6 -0
  27. package/dist/src/adapters/providers/codex.js.map +1 -1
  28. package/dist/src/adapters/providers/deepseek.js +6 -0
  29. package/dist/src/adapters/providers/deepseek.js.map +1 -1
  30. package/dist/src/adapters/providers/opencode.js +6 -0
  31. package/dist/src/adapters/providers/opencode.js.map +1 -1
  32. package/dist/src/adapters/providers/pi.js +6 -0
  33. package/dist/src/adapters/providers/pi.js.map +1 -1
  34. package/dist/src/artifacts/index.js +1 -0
  35. package/dist/src/artifacts/index.js.map +1 -1
  36. package/dist/src/backends/harbor/backend.js +55 -104
  37. package/dist/src/backends/harbor/backend.js.map +1 -1
  38. package/dist/src/backends/harbor/environment-config.js +92 -0
  39. package/dist/src/backends/harbor/environment-config.js.map +1 -0
  40. package/dist/src/backends/harbor/index.js +5 -1
  41. package/dist/src/backends/harbor/index.js.map +1 -1
  42. package/dist/src/backends/harbor/model-proxy-config.js +23 -0
  43. package/dist/src/backends/harbor/model-proxy-config.js.map +1 -0
  44. package/dist/src/backends/harbor/process.js +177 -0
  45. package/dist/src/backends/harbor/process.js.map +1 -0
  46. package/dist/src/backends/harbor/runtime-toolchain.js +30 -0
  47. package/dist/src/backends/harbor/runtime-toolchain.js.map +1 -0
  48. package/dist/src/backends/harbor/supervisor.js +74 -0
  49. package/dist/src/backends/harbor/supervisor.js.map +1 -0
  50. package/dist/src/backends/harbor/task-resources.js +206 -0
  51. package/dist/src/backends/harbor/task-resources.js.map +1 -0
  52. package/dist/src/backends/harbor/tools.js +6 -29
  53. package/dist/src/backends/harbor/tools.js.map +1 -1
  54. package/dist/src/backends/harbor/verifier-config.js +11 -0
  55. package/dist/src/backends/harbor/verifier-config.js.map +1 -0
  56. package/dist/src/backends/index.js +4 -1
  57. package/dist/src/backends/index.js.map +1 -1
  58. package/dist/src/cli/arguments.js +14 -0
  59. package/dist/src/cli/arguments.js.map +1 -1
  60. package/dist/src/cli/commands/daemon.js +135 -8
  61. package/dist/src/cli/commands/daemon.js.map +1 -1
  62. package/dist/src/cli/commands/eval.js +214 -4
  63. package/dist/src/cli/commands/eval.js.map +1 -1
  64. package/dist/src/cli/commands/images.js +40 -0
  65. package/dist/src/cli/commands/images.js.map +1 -0
  66. package/dist/src/cli/commands/run.js +6 -3
  67. package/dist/src/cli/commands/run.js.map +1 -1
  68. package/dist/src/cli/commands/runs.js +25 -3
  69. package/dist/src/cli/commands/runs.js.map +1 -1
  70. package/dist/src/cli/commands/worker.js +79 -0
  71. package/dist/src/cli/commands/worker.js.map +1 -0
  72. package/dist/src/cli/main.js +4 -0
  73. package/dist/src/cli/main.js.map +1 -1
  74. package/dist/src/cli/output.js +72 -3
  75. package/dist/src/cli/output.js.map +1 -1
  76. package/dist/src/control-plane/build-admission.js +69 -0
  77. package/dist/src/control-plane/build-admission.js.map +1 -0
  78. package/dist/src/control-plane/build-records.js +6 -0
  79. package/dist/src/control-plane/build-records.js.map +1 -0
  80. package/dist/src/control-plane/collisions.js +40 -0
  81. package/dist/src/control-plane/collisions.js.map +1 -0
  82. package/dist/src/control-plane/eval-control-state.js +22 -0
  83. package/dist/src/control-plane/eval-control-state.js.map +1 -0
  84. package/dist/src/control-plane/eval-control-work.js +39 -0
  85. package/dist/src/control-plane/eval-control-work.js.map +1 -0
  86. package/dist/src/control-plane/eval-image-resolution.js +52 -0
  87. package/dist/src/control-plane/eval-image-resolution.js.map +1 -0
  88. package/dist/src/control-plane/eval-image-services.js +30 -0
  89. package/dist/src/control-plane/eval-image-services.js.map +1 -0
  90. package/dist/src/control-plane/eval-records.js +252 -0
  91. package/dist/src/control-plane/eval-records.js.map +1 -0
  92. package/dist/src/control-plane/eval-recovery.js +181 -0
  93. package/dist/src/control-plane/eval-recovery.js.map +1 -0
  94. package/dist/src/control-plane/eval-scheduler.js +447 -0
  95. package/dist/src/control-plane/eval-scheduler.js.map +1 -0
  96. package/dist/src/control-plane/index.js +20 -0
  97. package/dist/src/control-plane/index.js.map +1 -0
  98. package/dist/src/control-plane/local-worker.js +49 -0
  99. package/dist/src/control-plane/local-worker.js.map +1 -0
  100. package/dist/src/control-plane/model-capture-planning.js +12 -0
  101. package/dist/src/control-plane/model-capture-planning.js.map +1 -0
  102. package/dist/src/control-plane/remote-result-transport.js +172 -0
  103. package/dist/src/control-plane/remote-result-transport.js.map +1 -0
  104. package/dist/src/control-plane/remote-work-coordinator.js +332 -0
  105. package/dist/src/control-plane/remote-work-coordinator.js.map +1 -0
  106. package/dist/src/control-plane/remote-work-inputs.js +266 -0
  107. package/dist/src/control-plane/remote-work-inputs.js.map +1 -0
  108. package/dist/src/control-plane/remote-work-recovery.js +181 -0
  109. package/dist/src/control-plane/remote-work-recovery.js.map +1 -0
  110. package/dist/src/control-plane/remote-worker-artifacts.js +135 -0
  111. package/dist/src/control-plane/remote-worker-artifacts.js.map +1 -0
  112. package/dist/src/control-plane/remote-worker-client.js +202 -0
  113. package/dist/src/control-plane/remote-worker-client.js.map +1 -0
  114. package/dist/src/control-plane/remote-worker-configuration.js +34 -0
  115. package/dist/src/control-plane/remote-worker-configuration.js.map +1 -0
  116. package/dist/src/control-plane/remote-worker-credentials.js +58 -0
  117. package/dist/src/control-plane/remote-worker-credentials.js.map +1 -0
  118. package/dist/src/control-plane/remote-worker-protocol.js +510 -0
  119. package/dist/src/control-plane/remote-worker-protocol.js.map +1 -0
  120. package/dist/src/control-plane/remote-worker-runner.js +214 -0
  121. package/dist/src/control-plane/remote-worker-runner.js.map +1 -0
  122. package/dist/src/control-plane/remote-workers.js +377 -0
  123. package/dist/src/control-plane/remote-workers.js.map +1 -0
  124. package/dist/src/control-plane/rerun-scheduler.js +411 -0
  125. package/dist/src/control-plane/rerun-scheduler.js.map +1 -0
  126. package/dist/src/control-plane/rerun-submission.js +56 -0
  127. package/dist/src/control-plane/rerun-submission.js.map +1 -0
  128. package/dist/src/control-plane/resources.js +182 -0
  129. package/dist/src/control-plane/resources.js.map +1 -0
  130. package/dist/src/control-plane/scheduler-eval-entry.js +32 -0
  131. package/dist/src/control-plane/scheduler-eval-entry.js.map +1 -0
  132. package/dist/src/control-plane/synthetic-result.js +21 -0
  133. package/dist/src/control-plane/synthetic-result.js.map +1 -0
  134. package/dist/src/control-plane/work-admission.js +21 -0
  135. package/dist/src/control-plane/work-admission.js.map +1 -0
  136. package/dist/src/control-plane/work-dispatcher.js +229 -0
  137. package/dist/src/control-plane/work-dispatcher.js.map +1 -0
  138. package/dist/src/controller-runtime/hash.js +12 -1
  139. package/dist/src/controller-runtime/hash.js.map +1 -1
  140. package/dist/src/controller-runtime/index.js +1 -1
  141. package/dist/src/controller-runtime/index.js.map +1 -1
  142. package/dist/src/controller-runtime/store.js +15 -6
  143. package/dist/src/controller-runtime/store.js.map +1 -1
  144. package/dist/src/daemon/client.js.map +1 -1
  145. package/dist/src/daemon/event-stream.js +46 -0
  146. package/dist/src/daemon/event-stream.js.map +1 -0
  147. package/dist/src/daemon/health.js +66 -0
  148. package/dist/src/daemon/health.js.map +1 -0
  149. package/dist/src/daemon/launcher.js +28 -1
  150. package/dist/src/daemon/launcher.js.map +1 -1
  151. package/dist/src/daemon/logging.js +23 -0
  152. package/dist/src/daemon/logging.js.map +1 -0
  153. package/dist/src/daemon/scheduler.js +26 -5
  154. package/dist/src/daemon/scheduler.js.map +1 -1
  155. package/dist/src/daemon/server.js +255 -98
  156. package/dist/src/daemon/server.js.map +1 -1
  157. package/dist/src/daemon/telemetry.js +163 -0
  158. package/dist/src/daemon/telemetry.js.map +1 -0
  159. package/dist/src/daemon/worker-routes.js +219 -0
  160. package/dist/src/daemon/worker-routes.js.map +1 -0
  161. package/dist/src/domain/bundles.js +2 -0
  162. package/dist/src/domain/bundles.js.map +1 -0
  163. package/dist/src/domain/control-plane.js +2 -0
  164. package/dist/src/domain/control-plane.js.map +1 -0
  165. package/dist/src/domain/execution-evidence.js +2 -0
  166. package/dist/src/domain/execution-evidence.js.map +1 -0
  167. package/dist/src/domain/execution-plan.js +2 -0
  168. package/dist/src/domain/execution-plan.js.map +1 -0
  169. package/dist/src/domain/images.js +2 -0
  170. package/dist/src/domain/images.js.map +1 -0
  171. package/dist/src/domain/index.js +9 -0
  172. package/dist/src/domain/index.js.map +1 -1
  173. package/dist/src/domain/interactions.js +2 -0
  174. package/dist/src/domain/interactions.js.map +1 -0
  175. package/dist/src/domain/providers.js +2 -0
  176. package/dist/src/domain/providers.js.map +1 -0
  177. package/dist/src/domain/resources.js +2 -0
  178. package/dist/src/domain/resources.js.map +1 -0
  179. package/dist/src/domain/workers.js +2 -0
  180. package/dist/src/domain/workers.js.map +1 -0
  181. package/dist/src/evals/collect-only-rerun.js +182 -0
  182. package/dist/src/evals/collect-only-rerun.js.map +1 -0
  183. package/dist/src/evals/directory.js +30 -0
  184. package/dist/src/evals/directory.js.map +1 -0
  185. package/dist/src/evals/docker-engine-stats.js +89 -0
  186. package/dist/src/evals/docker-engine-stats.js.map +1 -0
  187. package/dist/src/evals/docker-ownership.js +51 -0
  188. package/dist/src/evals/docker-ownership.js.map +1 -0
  189. package/dist/src/evals/docker-reaper.js +144 -0
  190. package/dist/src/evals/docker-reaper.js.map +1 -0
  191. package/dist/src/evals/docker-resource-observer.js +207 -0
  192. package/dist/src/evals/docker-resource-observer.js.map +1 -0
  193. package/dist/src/evals/environment-image-planning.js +164 -0
  194. package/dist/src/evals/environment-image-planning.js.map +1 -0
  195. package/dist/src/evals/eval-artifact-planning.js +56 -0
  196. package/dist/src/evals/eval-artifact-planning.js.map +1 -0
  197. package/dist/src/evals/eval-finalization.js +10 -0
  198. package/dist/src/evals/eval-finalization.js.map +1 -0
  199. package/dist/src/evals/eval-lifecycle-events.js +16 -0
  200. package/dist/src/evals/eval-lifecycle-events.js.map +1 -0
  201. package/dist/src/evals/events.js +53 -5
  202. package/dist/src/evals/events.js.map +1 -1
  203. package/dist/src/evals/execution-evidence.js +116 -0
  204. package/dist/src/evals/execution-evidence.js.map +1 -0
  205. package/dist/src/evals/execution-leases.js +342 -0
  206. package/dist/src/evals/execution-leases.js.map +1 -0
  207. package/dist/src/evals/execution-plan-artifacts.js +72 -0
  208. package/dist/src/evals/execution-plan-artifacts.js.map +1 -0
  209. package/dist/src/evals/execution-plan-images.js +69 -0
  210. package/dist/src/evals/execution-plan-images.js.map +1 -0
  211. package/dist/src/evals/execution-plan-resources.js +174 -0
  212. package/dist/src/evals/execution-plan-resources.js.map +1 -0
  213. package/dist/src/evals/execution-plan.js +415 -0
  214. package/dist/src/evals/execution-plan.js.map +1 -0
  215. package/dist/src/evals/harbor-artifact-builder-image.js +99 -0
  216. package/dist/src/evals/harbor-artifact-builder-image.js.map +1 -0
  217. package/dist/src/evals/harbor-artifact-builder.js +354 -0
  218. package/dist/src/evals/harbor-artifact-builder.js.map +1 -0
  219. package/dist/src/evals/harbor-bridge-error.js +39 -3
  220. package/dist/src/evals/harbor-bridge-error.js.map +1 -1
  221. package/dist/src/evals/harbor-phase-timings.js +60 -0
  222. package/dist/src/evals/harbor-phase-timings.js.map +1 -0
  223. package/dist/src/evals/index.js +26 -3
  224. package/dist/src/evals/index.js.map +1 -1
  225. package/dist/src/evals/infrastructure-retry.js +218 -0
  226. package/dist/src/evals/infrastructure-retry.js.map +1 -0
  227. package/dist/src/evals/interaction-capture-import.js +49 -0
  228. package/dist/src/evals/interaction-capture-import.js.map +1 -0
  229. package/dist/src/evals/local-docker-provider.js +372 -0
  230. package/dist/src/evals/local-docker-provider.js.map +1 -0
  231. package/dist/src/evals/local-eval-planning.js +80 -0
  232. package/dist/src/evals/local-eval-planning.js.map +1 -0
  233. package/dist/src/evals/model-capture-plan.js +22 -0
  234. package/dist/src/evals/model-capture-plan.js.map +1 -0
  235. package/dist/src/evals/model-capture-runtime.js +114 -0
  236. package/dist/src/evals/model-capture-runtime.js.map +1 -0
  237. package/dist/src/evals/model-proxy-runtime-state.js +53 -0
  238. package/dist/src/evals/model-proxy-runtime-state.js.map +1 -0
  239. package/dist/src/evals/planned-execution-support.js +54 -0
  240. package/dist/src/evals/planned-execution-support.js.map +1 -0
  241. package/dist/src/evals/planned-execution.js +461 -0
  242. package/dist/src/evals/planned-execution.js.map +1 -0
  243. package/dist/src/evals/planned-remote-execution.js +40 -0
  244. package/dist/src/evals/planned-remote-execution.js.map +1 -0
  245. package/dist/src/evals/planned-retry-lifecycle.js +186 -0
  246. package/dist/src/evals/planned-retry-lifecycle.js.map +1 -0
  247. package/dist/src/evals/prepared-harness.js +76 -0
  248. package/dist/src/evals/prepared-harness.js.map +1 -0
  249. package/dist/src/evals/progress.js +4 -4
  250. package/dist/src/evals/progress.js.map +1 -1
  251. package/dist/src/evals/records.js +20 -3
  252. package/dist/src/evals/records.js.map +1 -1
  253. package/dist/src/evals/recovery.js +192 -0
  254. package/dist/src/evals/recovery.js.map +1 -0
  255. package/dist/src/evals/remote-infrastructure-retry.js +99 -0
  256. package/dist/src/evals/remote-infrastructure-retry.js.map +1 -0
  257. package/dist/src/evals/request.js +14 -1
  258. package/dist/src/evals/request.js.map +1 -1
  259. package/dist/src/evals/rerun-inputs.js +89 -0
  260. package/dist/src/evals/rerun-inputs.js.map +1 -0
  261. package/dist/src/evals/rerun-slots.js +12 -3
  262. package/dist/src/evals/rerun-slots.js.map +1 -1
  263. package/dist/src/evals/rerun-types.js +49 -0
  264. package/dist/src/evals/rerun-types.js.map +1 -0
  265. package/dist/src/evals/rerun.js +113 -75
  266. package/dist/src/evals/rerun.js.map +1 -1
  267. package/dist/src/evals/result-helpers.js +65 -0
  268. package/dist/src/evals/result-helpers.js.map +1 -0
  269. package/dist/src/evals/resume-state.js +36 -0
  270. package/dist/src/evals/resume-state.js.map +1 -0
  271. package/dist/src/evals/service-types.js +2 -0
  272. package/dist/src/evals/service-types.js.map +1 -0
  273. package/dist/src/evals/service.js +327 -237
  274. package/dist/src/evals/service.js.map +1 -1
  275. package/dist/src/evals/task-resources.js +209 -0
  276. package/dist/src/evals/task-resources.js.map +1 -0
  277. package/dist/src/evals/trial-environment-evidence.js +87 -0
  278. package/dist/src/evals/trial-environment-evidence.js.map +1 -0
  279. package/dist/src/evals/trial-execution-evidence.js +13 -0
  280. package/dist/src/evals/trial-execution-evidence.js.map +1 -0
  281. package/dist/src/evals/trial-import-identity.js +31 -0
  282. package/dist/src/evals/trial-import-identity.js.map +1 -0
  283. package/dist/src/evals/trial-import.js +166 -168
  284. package/dist/src/evals/trial-import.js.map +1 -1
  285. package/dist/src/evals/trial-publication-recovery.js +104 -0
  286. package/dist/src/evals/trial-publication-recovery.js.map +1 -0
  287. package/dist/src/evals/trial-publication.js +25 -0
  288. package/dist/src/evals/trial-publication.js.map +1 -0
  289. package/dist/src/evals/verifier-diagnostics.js +235 -0
  290. package/dist/src/evals/verifier-diagnostics.js.map +1 -0
  291. package/dist/src/evals/work-item-artifacts.js +23 -0
  292. package/dist/src/evals/work-item-artifacts.js.map +1 -0
  293. package/dist/src/foundation/config.js +17 -0
  294. package/dist/src/foundation/config.js.map +1 -1
  295. package/dist/src/foundation/credential-redaction.js +137 -0
  296. package/dist/src/foundation/credential-redaction.js.map +1 -0
  297. package/dist/src/foundation/environment-image-references.js +59 -0
  298. package/dist/src/foundation/environment-image-references.js.map +1 -0
  299. package/dist/src/foundation/executable.js +45 -5
  300. package/dist/src/foundation/executable.js.map +1 -1
  301. package/dist/src/foundation/index.js +5 -2
  302. package/dist/src/foundation/index.js.map +1 -1
  303. package/dist/src/foundation/process-identity.js +69 -0
  304. package/dist/src/foundation/process-identity.js.map +1 -0
  305. package/dist/src/images/context.js +47 -0
  306. package/dist/src/images/context.js.map +1 -0
  307. package/dist/src/images/docker-buildkit.js +137 -0
  308. package/dist/src/images/docker-buildkit.js.map +1 -0
  309. package/dist/src/images/dockerfile.js +67 -0
  310. package/dist/src/images/dockerfile.js.map +1 -0
  311. package/dist/src/images/gc.js +261 -0
  312. package/dist/src/images/gc.js.map +1 -0
  313. package/dist/src/images/index.js +10 -0
  314. package/dist/src/images/index.js.map +1 -0
  315. package/dist/src/images/manifest.js +109 -0
  316. package/dist/src/images/manifest.js.map +1 -0
  317. package/dist/src/images/ownership.js +5 -0
  318. package/dist/src/images/ownership.js.map +1 -0
  319. package/dist/src/images/records.js +32 -0
  320. package/dist/src/images/records.js.map +1 -0
  321. package/dist/src/images/registry.js +154 -0
  322. package/dist/src/images/registry.js.map +1 -0
  323. package/dist/src/images/service.js +224 -0
  324. package/dist/src/images/service.js.map +1 -0
  325. package/dist/src/model-access/capture.js +255 -0
  326. package/dist/src/model-access/capture.js.map +1 -0
  327. package/dist/src/model-access/index.js +5 -0
  328. package/dist/src/model-access/index.js.map +1 -0
  329. package/dist/src/model-access/policy.js +49 -0
  330. package/dist/src/model-access/policy.js.map +1 -0
  331. package/dist/src/model-access/proxy.js +336 -0
  332. package/dist/src/model-access/proxy.js.map +1 -0
  333. package/dist/src/model-access/records.js +167 -0
  334. package/dist/src/model-access/records.js.map +1 -0
  335. package/dist/src/runs/bundle.js +450 -0
  336. package/dist/src/runs/bundle.js.map +1 -0
  337. package/dist/src/runs/events.js +12 -1
  338. package/dist/src/runs/events.js.map +1 -1
  339. package/dist/src/runs/executor.js +24 -17
  340. package/dist/src/runs/executor.js.map +1 -1
  341. package/dist/src/runs/index.js +3 -1
  342. package/dist/src/runs/index.js.map +1 -1
  343. package/dist/src/runs/manifest.js +5 -5
  344. package/dist/src/runs/manifest.js.map +1 -1
  345. package/dist/src/runs/request.js +6 -1
  346. package/dist/src/runs/request.js.map +1 -1
  347. package/dist/src/runs/training-candidate.js +161 -0
  348. package/dist/src/runs/training-candidate.js.map +1 -0
  349. package/dist/src/trajectories/provider-capture.js +15 -24
  350. package/dist/src/trajectories/provider-capture.js.map +1 -1
  351. package/dist/src/trajectories/providers/deepseek.js +3 -3
  352. package/dist/src/trajectories/providers/deepseek.js.map +1 -1
  353. package/dist/src/workers/index.js +3 -0
  354. package/dist/src/workers/index.js.map +1 -0
  355. package/dist/src/workers/remote-harbor-work-spec.js +97 -0
  356. package/dist/src/workers/remote-harbor-work-spec.js.map +1 -0
  357. package/dist/src/workers/remote-harbor-worker.js +220 -0
  358. package/dist/src/workers/remote-harbor-worker.js.map +1 -0
  359. package/docs/schemas/environment-build-index.schema.json +13 -0
  360. package/docs/schemas/environment-build-record.schema.json +25 -0
  361. package/docs/schemas/environment-image-manifest.schema.json +62 -0
  362. package/docs/schemas/eval-control.schema.json +36 -0
  363. package/docs/schemas/eval-execution-plan.schema.json +216 -0
  364. package/docs/schemas/eval-request.schema.json +2 -0
  365. package/docs/schemas/eval-rerun-result.schema.json +53 -0
  366. package/docs/schemas/eval-rerun-state.schema.json +61 -0
  367. package/docs/schemas/eval-rerun-submission.schema.json +58 -0
  368. package/docs/schemas/eval-result.schema.json +34 -1
  369. package/docs/schemas/eval-submission-input.schema.json +14 -0
  370. package/docs/schemas/eval-submission.schema.json +68 -0
  371. package/docs/schemas/eval-trial-publication.schema.json +39 -0
  372. package/docs/schemas/execution-evidence.schema.json +96 -0
  373. package/docs/schemas/execution-lease.schema.json +50 -0
  374. package/docs/schemas/execution-provider-status.schema.json +65 -0
  375. package/docs/schemas/execution-worker.schema.json +53 -0
  376. package/docs/schemas/harbor-process-exit.schema.json +14 -0
  377. package/docs/schemas/interaction-capture-ref.schema.json +45 -0
  378. package/docs/schemas/local-provider-execution.schema.json +33 -0
  379. package/docs/schemas/model-interaction.schema.json +41 -0
  380. package/docs/schemas/remote-result-envelope.schema.json +32 -0
  381. package/docs/schemas/remote-tree-envelope.schema.json +25 -0
  382. package/docs/schemas/remote-work-offer.schema.json +58 -0
  383. package/docs/schemas/remote-work-receipt.schema.json +50 -0
  384. package/docs/schemas/remote-worker-artifact.schema.json +17 -0
  385. package/docs/schemas/remote-worker-event.schema.json +18 -0
  386. package/docs/schemas/remote-worker-heartbeat.schema.json +37 -0
  387. package/docs/schemas/remote-worker-registration.schema.json +52 -0
  388. package/docs/schemas/result-bundle-index.schema.json +133 -0
  389. package/docs/schemas/run-environment-images.schema.json +39 -0
  390. package/docs/schemas/training-data-candidate.schema.json +46 -0
  391. package/integrations/harbor/hitch_harbor_agent.py +300 -547
  392. package/integrations/harbor/hitch_harbor_environment.py +331 -0
  393. package/integrations/harbor/hitch_harbor_task_resources.py +275 -0
  394. package/integrations/harbor/hitch_harbor_verifier.py +337 -0
  395. package/package.json +16 -2
@@ -0,0 +1,218 @@
1
+ import path from "node:path";
2
+ import { setTimeout as delay } from "node:timers/promises";
3
+ import { runHarborBackend } from "../backends/index.js";
4
+ import { HitchError } from "../foundation/index.js";
5
+ import { importEvalTrialRun, importEvalTrialRuns, TrialBundlePendingError, validateEvalTrialReferences } from "./trial-import.js";
6
+ import { replaceInvalidEvalProgressTrial, writeEvalProgress } from "./progress.js";
7
+ const INFRASTRUCTURE_REASONS = new Set([
8
+ "infrastructure_failure",
9
+ "verifier_infrastructure_failure",
10
+ "verifier_result_missing",
11
+ ]);
12
+ // Verifier failures are retried inside the original live Harbor trial by the
13
+ // custom verifier wrapper. Starting another Harbor trial would execute the
14
+ // candidate agent again, so only non-verifier trial infrastructure failures
15
+ // are eligible for this outer retry path.
16
+ const RETRYABLE_TRIAL_INFRASTRUCTURE_REASONS = new Set([
17
+ "infrastructure_failure",
18
+ ]);
19
+ export async function runInfrastructureRetries(options) {
20
+ let progress = options.progress;
21
+ let retryCandidates = retryableInfrastructureTrials(options.initialRefs);
22
+ const runs = [];
23
+ for (let retry = 1; retry <= options.request.infrastructure_retries && retryCandidates.length > 0; retry += 1) {
24
+ if (options.signal?.aborted)
25
+ break;
26
+ const retryTriggers = [...retryCandidates];
27
+ const taskNames = [...new Set(retryCandidates.map((ref) => ref.task_id))].sort();
28
+ const backoffMs = options.request.infrastructure_retry_backoff_ms * retry;
29
+ options.sink.emit({
30
+ type: "eval.infrastructure-retry.scheduled",
31
+ execution_kind: "physical-infrastructure-retry",
32
+ candidate_executes: true,
33
+ attempt: options.logicalAttempt,
34
+ retry,
35
+ tasks: taskNames,
36
+ backoff_ms: backoffMs,
37
+ });
38
+ if (backoffMs > 0) {
39
+ if (options.signal)
40
+ await delay(backoffMs, undefined, { signal: options.signal });
41
+ else
42
+ await delay(backoffMs);
43
+ }
44
+ if (options.signal?.aborted)
45
+ break;
46
+ const backendDirectory = path.join(options.backendBaseDirectory || path.join(options.evalDirectory, "infrastructure-retries"), `retry-${String(retry).padStart(4, "0")}`, options.request.attempts === 1 ? "harbor" : attemptDirectoryName(options.logicalAttempt));
47
+ const harborJobDirectory = path.join(backendDirectory, "job");
48
+ const retryRefs = [];
49
+ const selectedTasks = new Set(taskNames);
50
+ let lifecycle;
51
+ const retryEnvironmentImages = () => lifecycle?.environmentImages ?? options.environmentImages;
52
+ const executionEvidence = async () => lifecycle?.captureExecutionEvidence
53
+ ? { executionEvidence: await lifecycle.captureExecutionEvidence() }
54
+ : {};
55
+ const publish = async (ref) => {
56
+ if (ref.attempt !== options.logicalAttempt || !selectedTasks.has(ref.task_id)) {
57
+ throw new HitchError(`Harbor infrastructure retry returned an unselected trial: ${ref.task_id}#${ref.attempt}`, {
58
+ code: "eval_infrastructure_retry_trial_mismatch",
59
+ exitCode: 12,
60
+ });
61
+ }
62
+ const existing = retryRefs.find((current) => current.trial_id === ref.trial_id);
63
+ if (existing) {
64
+ if (JSON.stringify(existing) !== JSON.stringify(ref))
65
+ throw new Error(`infrastructure retry trial identity changed: ${ref.trial_id}`);
66
+ return;
67
+ }
68
+ retryRefs.push(ref);
69
+ await validateEvalTrialReferences(options.root, options.evalId, [ref], {
70
+ benchmarkId: options.request.benchmark_id,
71
+ benchmarkRevision: options.request.benchmark_revision,
72
+ });
73
+ if (ref.observation_status === "valid") {
74
+ progress = replaceInvalidEvalProgressTrial(progress, ref);
75
+ await writeEvalProgress(options.evalDirectory, progress);
76
+ }
77
+ options.sink.emit({
78
+ type: ref.observation_status === "valid"
79
+ ? "eval.infrastructure-retry.repaired"
80
+ : "eval.infrastructure-retry.failed",
81
+ attempt: options.logicalAttempt,
82
+ retry,
83
+ task_id: ref.task_id,
84
+ trial_id: ref.trial_id,
85
+ run_id: ref.run_id,
86
+ observation_status: ref.observation_status,
87
+ ...(ref.invalid_reason ? { invalid_reason: ref.invalid_reason } : {}),
88
+ generation: progress.generation,
89
+ });
90
+ };
91
+ let run;
92
+ try {
93
+ lifecycle = await options.beginRetry?.({ retry, logicalAttempt: options.logicalAttempt, taskNames, triggers: retryTriggers, backendDirectory });
94
+ run = await runHarborBackend({
95
+ evalId: options.evalId,
96
+ evalDirectory: options.evalDirectory,
97
+ backendDirectory,
98
+ logicalAttempt: options.logicalAttempt,
99
+ taskNames,
100
+ request: { ...options.request, attempts: 1 },
101
+ root: options.root,
102
+ resolvedRevision: options.resolvedRevision,
103
+ runtimeDirectory: options.controllerRuntime.directory,
104
+ runtimeId: options.controllerRuntime.runtime_id,
105
+ preparedArtifact: options.preparedArtifact,
106
+ ...(options.executionResources ? { executionResources: options.executionResources } : {}),
107
+ ...(options.resolvedImages ? { resolvedImages: options.resolvedImages } : {}),
108
+ ...(options.interactionCaptureExporter ? { modelProxy: options.interactionCaptureExporter.route } : {}),
109
+ ...lifecycle?.backend,
110
+ env: options.env,
111
+ ...(options.harborExecutable !== undefined ? { harborExecutable: options.harborExecutable } : {}),
112
+ ...(options.signal ? { signal: options.signal } : {}),
113
+ ...(options.trialBundleGraceMs === undefined ? {} : { trialBundleGraceMs: options.trialBundleGraceMs }),
114
+ emit: (event) => options.sink.emit({ ...event, execution_kind: "physical-infrastructure-retry", infrastructure_retry: retry, logical_attempt: options.logicalAttempt, ...(lifecycle ? { work_id: lifecycle.workId, lease_id: lifecycle.leaseId } : {}) }),
115
+ onTrialSettled: async (trial, context) => {
116
+ try {
117
+ const environmentImages = retryEnvironmentImages();
118
+ const ref = await importEvalTrialRun({
119
+ root: options.root,
120
+ evalId: options.evalId,
121
+ evalDirectory: options.evalDirectory,
122
+ harborJobDirectory,
123
+ expectedAttempt: options.logicalAttempt,
124
+ request: options.request,
125
+ resolvedRevision: options.resolvedRevision,
126
+ benchmarkId: options.request.benchmark_id,
127
+ benchmarkRevision: options.request.benchmark_revision,
128
+ publicationMode: "replace-invalid",
129
+ runtimeId: options.controllerRuntime.runtime_id,
130
+ env: options.env,
131
+ ...(options.modelCapturePlan ? { modelCapturePlan: options.modelCapturePlan } : {}),
132
+ ...(options.interactionCaptureExporter ? {
133
+ interactionCaptureExporter: options.interactionCaptureExporter,
134
+ } : {}),
135
+ ...await executionEvidence(),
136
+ ...(environmentImages ? { environmentImages } : {}),
137
+ requireCompleteMarker: true,
138
+ allowMissingBundleDiagnostic: context.bundleWaitExpired,
139
+ }, trial, retryRefs.length, retryRefs);
140
+ await publish(ref);
141
+ return true;
142
+ }
143
+ catch (error) {
144
+ if (error instanceof TrialBundlePendingError)
145
+ return false;
146
+ throw error;
147
+ }
148
+ },
149
+ });
150
+ const environmentImages = retryEnvironmentImages();
151
+ const terminalRefs = await importEvalTrialRuns({
152
+ root: options.root,
153
+ evalId: options.evalId,
154
+ evalDirectory: options.evalDirectory,
155
+ harborJobDirectory,
156
+ expectedAttempt: options.logicalAttempt,
157
+ request: options.request,
158
+ resolvedRevision: options.resolvedRevision,
159
+ benchmarkId: options.request.benchmark_id,
160
+ benchmarkRevision: options.request.benchmark_revision,
161
+ publicationMode: "replace-invalid",
162
+ runtimeId: options.controllerRuntime.runtime_id,
163
+ env: options.env,
164
+ ...(options.modelCapturePlan ? { modelCapturePlan: options.modelCapturePlan } : {}),
165
+ ...(options.interactionCaptureExporter ? {
166
+ interactionCaptureExporter: options.interactionCaptureExporter,
167
+ } : {}),
168
+ ...await executionEvidence(),
169
+ ...(environmentImages ? { environmentImages } : {}),
170
+ rawResult: run.rawResult,
171
+ }, retryRefs);
172
+ for (const ref of terminalRefs)
173
+ await publish(ref);
174
+ if (run.rawResult !== null)
175
+ assertBackendTrialSet(run.rawResult, retryRefs);
176
+ runs.push({ attempt: options.logicalAttempt, retry, tasks: taskNames, triggers: retryTriggers, refs: retryRefs, run, ...(lifecycle ? { leaseId: lifecycle.leaseId, workId: lifecycle.workId } : {}) });
177
+ }
178
+ finally {
179
+ await lifecycle?.close();
180
+ }
181
+ options.sink.emit({
182
+ type: "eval.infrastructure-retry.completed",
183
+ execution_kind: "physical-infrastructure-retry",
184
+ candidate_executes: true,
185
+ attempt: options.logicalAttempt,
186
+ retry,
187
+ tasks: taskNames,
188
+ repaired_tasks: retryRefs.filter((ref) => ref.observation_status === "valid").map((ref) => ref.task_id).sort(),
189
+ remaining_tasks: retryableInfrastructureTrials(retryRefs).map((ref) => ref.task_id).sort(),
190
+ });
191
+ if (run.backend.process_exit_code !== 0 || run.rawResult === null || options.stopAfterResult?.(run.rawResult))
192
+ break;
193
+ retryCandidates = retryableInfrastructureTrials(retryRefs);
194
+ }
195
+ return { progress, runs };
196
+ }
197
+ export function retryableInfrastructureTrials(trials) {
198
+ return trials.filter((trial) => trial.observation_status === "invalid"
199
+ && trial.invalid_reason !== undefined
200
+ && RETRYABLE_TRIAL_INFRASTRUCTURE_REASONS.has(trial.invalid_reason));
201
+ }
202
+ export function infrastructureFailureTrials(trials) {
203
+ return trials.filter((trial) => trial.observation_status === "invalid"
204
+ && trial.invalid_reason !== undefined
205
+ && INFRASTRUCTURE_REASONS.has(trial.invalid_reason));
206
+ }
207
+ function assertBackendTrialSet(rawResult, refs) {
208
+ const trials = Array.isArray(rawResult.trial_results) ? rawResult.trial_results : [];
209
+ const backendIds = new Set(trials.map((trial) => typeof trial.trial_name === "string" ? trial.trial_name : null).filter((value) => value !== null));
210
+ const refIds = new Set(refs.map((ref) => ref.trial_id));
211
+ if (backendIds.size !== trials.length || backendIds.size !== refIds.size || [...backendIds].some((id) => !refIds.has(id))) {
212
+ throw new Error("Harbor infrastructure retry trial set does not match published eval progress");
213
+ }
214
+ }
215
+ function attemptDirectoryName(attempt) {
216
+ return `attempt-${String(attempt).padStart(4, "0")}`;
217
+ }
218
+ //# sourceMappingURL=infrastructure-retry.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"infrastructure-retry.js","sourceRoot":"","sources":["../../../src/evals/infrastructure-retry.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,UAAU,IAAI,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAE3D,OAAO,EAAE,gBAAgB,EAAE,MAAM,sBAAsB,CAAC;AAIxD,OAAO,EAAE,UAAU,EAAE,MAAM,wBAAwB,CAAC;AAEpD,OAAO,EAAE,kBAAkB,EAAE,mBAAmB,EAAE,uBAAuB,EAAE,2BAA2B,EAAE,MAAM,mBAAmB,CAAC;AAClI,OAAO,EAAE,+BAA+B,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAC;AAInF,MAAM,sBAAsB,GAAG,IAAI,GAAG,CAAC;IACrC,wBAAwB;IACxB,iCAAiC;IACjC,yBAAyB;CAC1B,CAAC,CAAC;AAEH,6EAA6E;AAC7E,2EAA2E;AAC3E,4EAA4E;AAC5E,0CAA0C;AAC1C,MAAM,sCAAsC,GAAG,IAAI,GAAG,CAAC;IACrD,wBAAwB;CACzB,CAAC,CAAC;AA6DH,MAAM,CAAC,KAAK,UAAU,wBAAwB,CAC5C,OAAwC;IAExC,IAAI,QAAQ,GAAG,OAAO,CAAC,QAAQ,CAAC;IAChC,IAAI,eAAe,GAAG,6BAA6B,CAAC,OAAO,CAAC,WAAW,CAAC,CAAC;IACzE,MAAM,IAAI,GAA6B,EAAE,CAAC;IAC1C,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,IAAI,OAAO,CAAC,OAAO,CAAC,sBAAsB,IAAI,eAAe,CAAC,MAAM,GAAG,CAAC,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;QAC9G,IAAI,OAAO,CAAC,MAAM,EAAE,OAAO;YAAE,MAAM;QACnC,MAAM,aAAa,GAAG,CAAC,GAAG,eAAe,CAAC,CAAC;QAC3C,MAAM,SAAS,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QACjF,MAAM,SAAS,GAAG,OAAO,CAAC,OAAO,CAAC,+BAA+B,GAAG,KAAK,CAAC;QAC1E,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC;YAChB,IAAI,EAAE,qCAAqC;YAC3C,cAAc,EAAE,+BAA+B;YAC/C,kBAAkB,EAAE,IAAI;YACxB,OAAO,EAAE,OAAO,CAAC,cAAc;YAC/B,KAAK;YACL,KAAK,EAAE,SAAS;YAChB,UAAU,EAAE,SAAS;SACtB,CAAC,CAAC;QACH,IAAI,SAAS,GAAG,CAAC,EAAE,CAAC;YAClB,IAAI,OAAO,CAAC,MAAM;gBAAE,MAAM,KAAK,CAAC,SAAS,EAAE,SAAS,EAAE,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC;;gBAC7E,MAAM,KAAK,CAAC,SAAS,CAAC,CAAC;QAC9B,CAAC;QACD,IAAI,OAAO,CAAC,MAAM,EAAE,OAAO;YAAE,MAAM;QAEnC,MAAM,gBAAgB,GAAG,IAAI,CAAC,IAAI,CAChC,OAAO,CAAC,oBAAoB,IAAI,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,aAAa,EAAE,wBAAwB,CAAC,EAC1F,SAAS,MAAM,CAAC,KAAK,CAAC,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,EACzC,OAAO,CAAC,OAAO,CAAC,QAAQ,KAAK,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,oBAAoB,CAAC,OAAO,CAAC,cAAc,CAAC,CACzF,CAAC;QACF,MAAM,kBAAkB,GAAG,IAAI,CAAC,IAAI,CAAC,gBAAgB,EAAE,KAAK,CAAC,CAAC;QAC9D,MAAM,SAAS,GAAqB,EAAE,CAAC;QACvC,MAAM,aAAa,GAAG,IAAI,GAAG,CAAC,SAAS,CAAC,CAAC;QACzC,IAAI,SAAmD,CAAC;QACxD,MAAM,sBAAsB,GAAG,GAAyC,EAAE,CAAC,SAAS,EAAE,iBAAiB,IAAI,OAAO,CAAC,iBAAiB,CAAC;QACrI,MAAM,iBAAiB,GAAG,KAAK,IAA0D,EAAE,CAAC,SAAS,EAAE,wBAAwB;YAC7H,CAAC,CAAC,EAAE,iBAAiB,EAAE,MAAM,SAAS,CAAC,wBAAwB,EAAE,EAAE;YACnE,CAAC,CAAC,EAAE,CAAC;QACP,MAAM,OAAO,GAAG,KAAK,EAAE,GAAmB,EAAiB,EAAE;YAC3D,IAAI,GAAG,CAAC,OAAO,KAAK,OAAO,CAAC,cAAc,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;gBAC9E,MAAM,IAAI,UAAU,CAAC,6DAA6D,GAAG,CAAC,OAAO,IAAI,GAAG,CAAC,OAAO,EAAE,EAAE;oBAC9G,IAAI,EAAE,0CAA0C;oBAChD,QAAQ,EAAE,EAAE;iBACb,CAAC,CAAC;YACL,CAAC;YACD,MAAM,QAAQ,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,QAAQ,KAAK,GAAG,CAAC,QAAQ,CAAC,CAAC;YAChF,IAAI,QAAQ,EAAE,CAAC;gBACb,IAAI,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC;oBAAE,MAAM,IAAI,KAAK,CAAC,gDAAgD,GAAG,CAAC,QAAQ,EAAE,CAAC,CAAC;gBACtI,OAAO;YACT,CAAC;YACD,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACpB,MAAM,2BAA2B,CAAC,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,EAAE;gBACrE,WAAW,EAAE,OAAO,CAAC,OAAO,CAAC,YAAY;gBACzC,iBAAiB,EAAE,OAAO,CAAC,OAAO,CAAC,kBAAkB;aACtD,CAAC,CAAC;YACH,IAAI,GAAG,CAAC,kBAAkB,KAAK,OAAO,EAAE,CAAC;gBACvC,QAAQ,GAAG,+BAA+B,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;gBAC1D,MAAM,iBAAiB,CAAC,OAAO,CAAC,aAAa,EAAE,QAAQ,CAAC,CAAC;YAC3D,CAAC;YACD,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC;gBAChB,IAAI,EAAE,GAAG,CAAC,kBAAkB,KAAK,OAAO;oBACtC,CAAC,CAAC,oCAAoC;oBACtC,CAAC,CAAC,kCAAkC;gBACtC,OAAO,EAAE,OAAO,CAAC,cAAc;gBAC/B,KAAK;gBACL,OAAO,EAAE,GAAG,CAAC,OAAO;gBACpB,QAAQ,EAAE,GAAG,CAAC,QAAQ;gBACtB,MAAM,EAAE,GAAG,CAAC,MAAM;gBAClB,kBAAkB,EAAE,GAAG,CAAC,kBAAkB;gBAC1C,GAAG,CAAC,GAAG,CAAC,cAAc,CAAC,CAAC,CAAC,EAAE,cAAc,EAAE,GAAG,CAAC,cAAc,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACrE,UAAU,EAAE,QAAQ,CAAC,UAAU;aAChC,CAAC,CAAC;QACL,CAAC,CAAC;QACF,IAAI,GAAwB,CAAC;QAC7B,IAAI,CAAC;YACH,SAAS,GAAG,MAAM,OAAO,CAAC,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,cAAc,EAAE,OAAO,CAAC,cAAc,EAAE,SAAS,EAAE,QAAQ,EAAE,aAAa,EAAE,gBAAgB,EAAE,CAAC,CAAC;YAChJ,GAAG,GAAG,MAAM,gBAAgB,CAAC;gBAC3B,MAAM,EAAE,OAAO,CAAC,MAAM;gBACtB,aAAa,EAAE,OAAO,CAAC,aAAa;gBACpC,gBAAgB;gBAChB,cAAc,EAAE,OAAO,CAAC,cAAc;gBACtC,SAAS;gBACT,OAAO,EAAE,EAAE,GAAG,OAAO,CAAC,OAAO,EAAE,QAAQ,EAAE,CAAC,EAAE;gBAC5C,IAAI,EAAE,OAAO,CAAC,IAAI;gBAClB,gBAAgB,EAAE,OAAO,CAAC,gBAAgB;gBAC1C,gBAAgB,EAAE,OAAO,CAAC,iBAAiB,CAAC,SAAS;gBACrD,SAAS,EAAE,OAAO,CAAC,iBAAiB,CAAC,UAAU;gBAC/C,gBAAgB,EAAE,OAAO,CAAC,gBAAgB;gBAC1C,GAAG,CAAC,OAAO,CAAC,kBAAkB,CAAC,CAAC,CAAC,EAAE,kBAAkB,EAAE,OAAO,CAAC,kBAAkB,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACzF,GAAG,CAAC,OAAO,CAAC,cAAc,CAAC,CAAC,CAAC,EAAE,cAAc,EAAE,OAAO,CAAC,cAAc,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBAC7E,GAAG,CAAC,OAAO,CAAC,0BAA0B,CAAC,CAAC,CAAC,EAAE,UAAU,EAAE,OAAO,CAAC,0BAA0B,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACvG,GAAG,SAAS,EAAE,OAAO;gBACrB,GAAG,EAAE,OAAO,CAAC,GAAG;gBAChB,GAAG,CAAC,OAAO,CAAC,gBAAgB,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,gBAAgB,EAAE,OAAO,CAAC,gBAAgB,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACjG,GAAG,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACrD,GAAG,CAAC,OAAO,CAAC,kBAAkB,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,kBAAkB,EAAE,OAAO,CAAC,kBAAkB,EAAE,CAAC;gBACvG,IAAI,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,KAAK,EAAE,cAAc,EAAE,+BAA+B,EAAE,oBAAoB,EAAE,KAAK,EAAE,eAAe,EAAE,OAAO,CAAC,cAAc,EAAE,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,SAAS,CAAC,MAAM,EAAE,QAAQ,EAAE,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC;gBACzP,cAAc,EAAE,KAAK,EAAE,KAAK,EAAE,OAAO,EAAoB,EAAE;oBACzD,IAAI,CAAC;wBACH,MAAM,iBAAiB,GAAG,sBAAsB,EAAE,CAAC;wBACnD,MAAM,GAAG,GAAG,MAAM,kBAAkB,CAAC;4BACnC,IAAI,EAAE,OAAO,CAAC,IAAI;4BAClB,MAAM,EAAE,OAAO,CAAC,MAAM;4BACtB,aAAa,EAAE,OAAO,CAAC,aAAa;4BACpC,kBAAkB;4BAClB,eAAe,EAAE,OAAO,CAAC,cAAc;4BACvC,OAAO,EAAE,OAAO,CAAC,OAAO;4BACxB,gBAAgB,EAAE,OAAO,CAAC,gBAAgB;4BAC1C,WAAW,EAAE,OAAO,CAAC,OAAO,CAAC,YAAY;4BACzC,iBAAiB,EAAE,OAAO,CAAC,OAAO,CAAC,kBAAkB;4BACrD,eAAe,EAAE,iBAAiB;4BAClC,SAAS,EAAE,OAAO,CAAC,iBAAiB,CAAC,UAAU;4BAC/C,GAAG,EAAE,OAAO,CAAC,GAAG;4BAChB,GAAG,CAAC,OAAO,CAAC,gBAAgB,CAAC,CAAC,CAAC,EAAE,gBAAgB,EAAE,OAAO,CAAC,gBAAgB,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;4BACnF,GAAG,CAAC,OAAO,CAAC,0BAA0B,CAAC,CAAC,CAAC;gCACvC,0BAA0B,EAAE,OAAO,CAAC,0BAA0B;6BAC/D,CAAC,CAAC,CAAC,EAAE,CAAC;4BACP,GAAG,MAAM,iBAAiB,EAAE;4BAC5B,GAAG,CAAC,iBAAiB,CAAC,CAAC,CAAC,EAAE,iBAAiB,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;4BACnD,qBAAqB,EAAE,IAAI;4BAC3B,4BAA4B,EAAE,OAAO,CAAC,iBAAiB;yBACxD,EAAE,KAAK,EAAE,SAAS,CAAC,MAAM,EAAE,SAAS,CAAC,CAAC;wBACvC,MAAM,OAAO,CAAC,GAAG,CAAC,CAAC;wBACnB,OAAO,IAAI,CAAC;oBACd,CAAC;oBAAC,OAAO,KAAK,EAAE,CAAC;wBACf,IAAI,KAAK,YAAY,uBAAuB;4BAAE,OAAO,KAAK,CAAC;wBAC3D,MAAM,KAAK,CAAC;oBACd,CAAC;gBACH,CAAC;aACF,CAAC,CAAC;YACH,MAAM,iBAAiB,GAAG,sBAAsB,EAAE,CAAC;YACnD,MAAM,YAAY,GAAG,MAAM,mBAAmB,CAAC;gBAC7C,IAAI,EAAE,OAAO,CAAC,IAAI;gBAClB,MAAM,EAAE,OAAO,CAAC,MAAM;gBACtB,aAAa,EAAE,OAAO,CAAC,aAAa;gBACpC,kBAAkB;gBAClB,eAAe,EAAE,OAAO,CAAC,cAAc;gBACvC,OAAO,EAAE,OAAO,CAAC,OAAO;gBACxB,gBAAgB,EAAE,OAAO,CAAC,gBAAgB;gBAC1C,WAAW,EAAE,OAAO,CAAC,OAAO,CAAC,YAAY;gBACzC,iBAAiB,EAAE,OAAO,CAAC,OAAO,CAAC,kBAAkB;gBACrD,eAAe,EAAE,iBAAiB;gBAClC,SAAS,EAAE,OAAO,CAAC,iBAAiB,CAAC,UAAU;gBAC/C,GAAG,EAAE,OAAO,CAAC,GAAG;gBAChB,GAAG,CAAC,OAAO,CAAC,gBAAgB,CAAC,CAAC,CAAC,EAAE,gBAAgB,EAAE,OAAO,CAAC,gBAAgB,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACnF,GAAG,CAAC,OAAO,CAAC,0BAA0B,CAAC,CAAC,CAAC;oBACvC,0BAA0B,EAAE,OAAO,CAAC,0BAA0B;iBAC/D,CAAC,CAAC,CAAC,EAAE,CAAC;gBACP,GAAG,MAAM,iBAAiB,EAAE;gBAC5B,GAAG,CAAC,iBAAiB,CAAC,CAAC,CAAC,EAAE,iBAAiB,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACnD,SAAS,EAAE,GAAG,CAAC,SAAS;aACzB,EAAE,SAAS,CAAC,CAAC;YACd,KAAK,MAAM,GAAG,IAAI,YAAY;gBAAE,MAAM,OAAO,CAAC,GAAG,CAAC,CAAC;YACnD,IAAI,GAAG,CAAC,SAAS,KAAK,IAAI;gBAAE,qBAAqB,CAAC,GAAG,CAAC,SAAS,EAAE,SAAS,CAAC,CAAC;YAC5E,IAAI,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,OAAO,CAAC,cAAc,EAAE,KAAK,EAAE,KAAK,EAAE,SAAS,EAAE,QAAQ,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,EAAE,GAAG,EAAE,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,SAAS,CAAC,OAAO,EAAE,MAAM,EAAE,SAAS,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;QACzM,CAAC;gBAAS,CAAC;YACT,MAAM,SAAS,EAAE,KAAK,EAAE,CAAC;QAC3B,CAAC;QACD,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC;YAChB,IAAI,EAAE,qCAAqC;YAC3C,cAAc,EAAE,+BAA+B;YAC/C,kBAAkB,EAAE,IAAI;YACxB,OAAO,EAAE,OAAO,CAAC,cAAc;YAC/B,KAAK;YACL,KAAK,EAAE,SAAS;YAChB,cAAc,EAAE,SAAS,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,kBAAkB,KAAK,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE;YAC9G,eAAe,EAAE,6BAA6B,CAAC,SAAS,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE;SAC3F,CAAC,CAAC;QACH,IAAI,GAAG,CAAC,OAAO,CAAC,iBAAiB,KAAK,CAAC,IAAI,GAAG,CAAC,SAAS,KAAK,IAAI,IAAI,OAAO,CAAC,eAAe,EAAE,CAAC,GAAG,CAAC,SAAS,CAAC;YAAE,MAAM;QACrH,eAAe,GAAG,6BAA6B,CAAC,SAAS,CAAC,CAAC;IAC7D,CAAC;IACD,OAAO,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC;AAC5B,CAAC;AAED,MAAM,UAAU,6BAA6B,CAAC,MAAiC;IAC7E,OAAO,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,kBAAkB,KAAK,SAAS;WACjE,KAAK,CAAC,cAAc,KAAK,SAAS;WAClC,sCAAsC,CAAC,GAAG,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC,CAAC;AACzE,CAAC;AAED,MAAM,UAAU,2BAA2B,CAAC,MAAiC;IAC3E,OAAO,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,kBAAkB,KAAK,SAAS;WACjE,KAAK,CAAC,cAAc,KAAK,SAAS;WAClC,sBAAsB,CAAC,GAAG,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC,CAAC;AACzD,CAAC;AAED,SAAS,qBAAqB,CAAC,SAAkC,EAAE,IAA+B;IAChG,MAAM,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,SAAS,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,aAA0C,CAAC,CAAC,CAAC,EAAE,CAAC;IAClH,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,KAAK,CAAC,UAAU,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAmB,EAAE,CAAC,KAAK,KAAK,IAAI,CAAC,CAAC,CAAC;IACrK,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC;IACxD,IAAI,UAAU,CAAC,IAAI,KAAK,MAAM,CAAC,MAAM,IAAI,UAAU,CAAC,IAAI,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,GAAG,UAAU,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC;QAC1H,MAAM,IAAI,KAAK,CAAC,8EAA8E,CAAC,CAAC;IAClG,CAAC;AACH,CAAC;AAED,SAAS,oBAAoB,CAAC,OAAe;IAC3C,OAAO,WAAW,MAAM,CAAC,OAAO,CAAC,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC;AACvD,CAAC"}
@@ -0,0 +1,49 @@
1
+ import { HitchError, atomicWriteJSON, ensureDir } from "../foundation/index.js";
2
+ import { loadInteractionCapture } from "../model-access/index.js";
3
+ export async function importTrialInteractionCapture(input, runId, staging) {
4
+ const plan = input.modelCapturePlan;
5
+ await writeTrialCapturePolicy(staging, plan);
6
+ const expectsProxy = plan?.effective_mode === "proxy" || plan?.effective_mode === "hybrid";
7
+ const existing = expectsProxy
8
+ ? await loadInteractionCapture(staging).catch((error) => {
9
+ if (error.code === "ENOENT")
10
+ return null;
11
+ throw error;
12
+ })
13
+ : null;
14
+ if (existing) {
15
+ const ref = existing.ref;
16
+ if (ref.run_id !== runId || ref.mode !== plan?.effective_mode || ref.required !== plan.required
17
+ || ref.topology !== plan.topology || plan.required && ref.completeness !== "complete") {
18
+ throw new HitchError("remote model interaction capture does not match its sealed plan", { code: "model_capture_incomplete", exitCode: 12 });
19
+ }
20
+ return;
21
+ }
22
+ if (!input.interactionCaptureExporter) {
23
+ if (expectsProxy && plan.required) {
24
+ throw new HitchError("required model interaction capture is unavailable", { code: "model_capture_incomplete", exitCode: 12 });
25
+ }
26
+ return;
27
+ }
28
+ const ref = await input.interactionCaptureExporter.finalizeRun(runId, staging);
29
+ if (ref.run_id !== runId || ref.mode !== input.interactionCaptureExporter.route.mode) {
30
+ throw new HitchError("model interaction capture identity does not match the run", { code: "model_capture_incomplete", exitCode: 12 });
31
+ }
32
+ if (input.interactionCaptureExporter.plan.required && ref.completeness !== "complete") {
33
+ throw new HitchError("required model interaction capture is incomplete", { code: "model_capture_incomplete", exitCode: 12 });
34
+ }
35
+ }
36
+ export async function writeTrialCapturePolicy(staging, plan) {
37
+ if (!plan)
38
+ return;
39
+ await ensureDir(`${staging}/interactions`);
40
+ await atomicWriteJSON(`${staging}/interactions/capture.policy.json`, {
41
+ schema_version: "1",
42
+ requested_mode: plan.requested_mode,
43
+ effective_mode: plan.effective_mode,
44
+ required: plan.required,
45
+ ...(plan.topology ? { topology: plan.topology } : {}),
46
+ ...(plan.degraded_reason ? { degraded_reason: plan.degraded_reason } : {}),
47
+ });
48
+ }
49
+ //# sourceMappingURL=interaction-capture-import.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"interaction-capture-import.js","sourceRoot":"","sources":["../../../src/evals/interaction-capture-import.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,UAAU,EAAE,eAAe,EAAE,SAAS,EAAE,MAAM,wBAAwB,CAAC;AAChF,OAAO,EAAE,sBAAsB,EAAE,MAAM,0BAA0B,CAAC;AAGlE,MAAM,CAAC,KAAK,UAAU,6BAA6B,CAAC,KAGnD,EAAE,KAAa,EAAE,OAAe;IAC/B,MAAM,IAAI,GAAG,KAAK,CAAC,gBAAgB,CAAC;IACpC,MAAM,uBAAuB,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;IAC7C,MAAM,YAAY,GAAG,IAAI,EAAE,cAAc,KAAK,OAAO,IAAI,IAAI,EAAE,cAAc,KAAK,QAAQ,CAAC;IAC3F,MAAM,QAAQ,GAAG,YAAY;QAC3B,CAAC,CAAC,MAAM,sBAAsB,CAAC,OAAO,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE;YACtD,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ;gBAAE,OAAO,IAAI,CAAC;YACpE,MAAM,KAAK,CAAC;QACd,CAAC,CAAC;QACF,CAAC,CAAC,IAAI,CAAC;IACT,IAAI,QAAQ,EAAE,CAAC;QACb,MAAM,GAAG,GAAG,QAAQ,CAAC,GAAG,CAAC;QACzB,IAAI,GAAG,CAAC,MAAM,KAAK,KAAK,IAAI,GAAG,CAAC,IAAI,KAAK,IAAI,EAAE,cAAc,IAAI,GAAG,CAAC,QAAQ,KAAK,IAAI,CAAC,QAAQ;eAC1F,GAAG,CAAC,QAAQ,KAAK,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ,IAAI,GAAG,CAAC,YAAY,KAAK,UAAU,EAAE,CAAC;YACxF,MAAM,IAAI,UAAU,CAAC,iEAAiE,EAAE,EAAE,IAAI,EAAE,0BAA0B,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;QAC9I,CAAC;QACD,OAAO;IACT,CAAC;IACD,IAAI,CAAC,KAAK,CAAC,0BAA0B,EAAE,CAAC;QACtC,IAAI,YAAY,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;YAClC,MAAM,IAAI,UAAU,CAAC,mDAAmD,EAAE,EAAE,IAAI,EAAE,0BAA0B,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;QAChI,CAAC;QACD,OAAO;IACT,CAAC;IACD,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,0BAA0B,CAAC,WAAW,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IAC/E,IAAI,GAAG,CAAC,MAAM,KAAK,KAAK,IAAI,GAAG,CAAC,IAAI,KAAK,KAAK,CAAC,0BAA0B,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;QACrF,MAAM,IAAI,UAAU,CAAC,2DAA2D,EAAE,EAAE,IAAI,EAAE,0BAA0B,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;IACxI,CAAC;IACD,IAAI,KAAK,CAAC,0BAA0B,CAAC,IAAI,CAAC,QAAQ,IAAI,GAAG,CAAC,YAAY,KAAK,UAAU,EAAE,CAAC;QACtF,MAAM,IAAI,UAAU,CAAC,kDAAkD,EAAE,EAAE,IAAI,EAAE,0BAA0B,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;IAC/H,CAAC;AACH,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,uBAAuB,CAAC,OAAe,EAAE,IAAyB;IACtF,IAAI,CAAC,IAAI;QAAE,OAAO;IAClB,MAAM,SAAS,CAAC,GAAG,OAAO,eAAe,CAAC,CAAC;IAC3C,MAAM,eAAe,CAAC,GAAG,OAAO,mCAAmC,EAAE;QACnE,cAAc,EAAE,GAAG;QACnB,cAAc,EAAE,IAAI,CAAC,cAAc;QACnC,cAAc,EAAE,IAAI,CAAC,cAAc;QACnC,QAAQ,EAAE,IAAI,CAAC,QAAQ;QACvB,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,IAAI,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACrD,GAAG,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC,CAAC,EAAE,eAAe,EAAE,IAAI,CAAC,eAAe,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;KAC3E,CAAC,CAAC;AACL,CAAC"}