@gajae-code/coding-agent 0.14.2 → 0.15.0

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 (345) hide show
  1. package/CHANGELOG.md +120 -1
  2. package/dist/types/async/job-manager.d.ts +4 -1
  3. package/dist/types/autoresearch/dashboard.d.ts +29 -4
  4. package/dist/types/autoresearch/data-context.d.ts +25 -0
  5. package/dist/types/autoresearch/git.d.ts +65 -7
  6. package/dist/types/autoresearch/harness.d.ts +58 -0
  7. package/dist/types/autoresearch/prompts.d.ts +51 -0
  8. package/dist/types/autoresearch/runs.d.ts +173 -0
  9. package/dist/types/autoresearch/types.d.ts +9 -139
  10. package/dist/types/capability/types.d.ts +20 -0
  11. package/dist/types/cli/accounts-cli.d.ts +1 -0
  12. package/dist/types/commands/{rlm.d.ts → autoresearch.d.ts} +2 -4
  13. package/dist/types/config/autorouting-contract.d.ts +146 -0
  14. package/dist/types/config/autorouting-generator.d.ts +40 -0
  15. package/dist/types/config/autorouting-tier-map.d.ts +16844 -0
  16. package/dist/types/config/autorouting.d.ts +35 -0
  17. package/dist/types/config/model-discovery-manager.d.ts +8 -1
  18. package/dist/types/config/model-registry.d.ts +30 -1
  19. package/dist/types/config/provider-selection-policy.d.ts +23 -0
  20. package/dist/types/config/settings-schema.d.ts +61 -3
  21. package/dist/types/config/settings.d.ts +3 -0
  22. package/dist/types/coordinator/contract.d.ts +1 -1
  23. package/dist/types/coordinator-mcp/durability.d.ts +29 -0
  24. package/dist/types/coordinator-mcp/event-webhook.d.ts +50 -0
  25. package/dist/types/coordinator-mcp/question-state.d.ts +1 -1
  26. package/dist/types/coordinator-mcp/server.d.ts +14 -0
  27. package/dist/types/crash/index-store.d.ts +3 -0
  28. package/dist/types/crash/record-loader.d.ts +2 -0
  29. package/dist/types/crash/upstream/relay.d.ts +5 -3
  30. package/dist/types/cursor.d.ts +12 -0
  31. package/dist/types/defaults/gjc-defaults.d.ts +35 -1
  32. package/dist/types/defaults/gjc-skills.generated.d.ts +1 -1
  33. package/dist/types/discovery/helpers.d.ts +1 -1
  34. package/dist/types/edit/index.d.ts +2 -1
  35. package/dist/types/eval/py/executor.d.ts +2 -0
  36. package/dist/types/extensibility/extensions/loader.d.ts +32 -0
  37. package/dist/types/extensibility/gjc-plugins/types.d.ts +1 -1
  38. package/dist/types/extensibility/skills.d.ts +2 -0
  39. package/dist/types/gjc-runtime/autoresearch-runtime.d.ts +194 -0
  40. package/dist/types/gjc-runtime/gc-runtime.d.ts +1 -1
  41. package/dist/types/gjc-runtime/psmux-detect.d.ts +1 -1
  42. package/dist/types/gjc-runtime/python-transcript.d.ts +19 -0
  43. package/dist/types/gjc-runtime/ralplan-runtime.d.ts +2 -8
  44. package/dist/types/gjc-runtime/session-layout.d.ts +6 -3
  45. package/dist/types/gjc-runtime/state-renderer.d.ts +0 -22
  46. package/dist/types/gjc-runtime/state-schema.d.ts +7 -7
  47. package/dist/types/gjc-runtime/tmux-common.d.ts +2 -2
  48. package/dist/types/gjc-runtime/ultragoal-change-set.d.ts +38 -1
  49. package/dist/types/gjc-runtime/ultragoal-runtime.d.ts +2 -24
  50. package/dist/types/gjc-runtime/ultragoal-validation-policy.d.ts +73 -0
  51. package/dist/types/gjc-runtime/workflow-recovery-projection.d.ts +100 -0
  52. package/dist/types/hooks/skill-keywords.d.ts +1 -1
  53. package/dist/types/main.d.ts +0 -10
  54. package/dist/types/modes/acp/acp-agent.d.ts +13 -6
  55. package/dist/types/modes/components/model-selector.d.ts +14 -0
  56. package/dist/types/modes/components/skill-hud/render.d.ts +6 -1
  57. package/dist/types/modes/components/smart-routing-panel.d.ts +82 -0
  58. package/dist/types/modes/controllers/selector-controller.d.ts +14 -0
  59. package/dist/types/modes/interactive-mode.d.ts +3 -0
  60. package/dist/types/modes/types.d.ts +9 -0
  61. package/dist/types/rlm/artifacts.d.ts +4 -4
  62. package/dist/types/runtime/optional-runtime-services.d.ts +2 -1
  63. package/dist/types/runtime/workspace-tree-service.d.ts +5 -1
  64. package/dist/types/runtime-mcp/config.d.ts +6 -0
  65. package/dist/types/runtime-mcp/manager.d.ts +10 -0
  66. package/dist/types/sdk/acp/adapter.d.ts +0 -4
  67. package/dist/types/sdk/broker/broker.d.ts +1 -1
  68. package/dist/types/sdk/broker/lifecycle.d.ts +57 -2
  69. package/dist/types/sdk/broker/session-index.d.ts +15 -0
  70. package/dist/types/sdk/bus/chat-daemon-control.d.ts +6 -0
  71. package/dist/types/sdk/bus/chat-daemon-runtime.d.ts +4 -0
  72. package/dist/types/sdk/bus/index.d.ts +8 -0
  73. package/dist/types/sdk/bus/kind-aware-reconciliation.d.ts +2 -0
  74. package/dist/types/sdk/bus/telegram-daemon-contract.d.ts +12 -2
  75. package/dist/types/sdk/client/client.d.ts +39 -4
  76. package/dist/types/sdk/host/host.d.ts +1 -0
  77. package/dist/types/sdk/host/internal-autorouting-state.d.ts +3 -0
  78. package/dist/types/sdk/host/model-pin.d.ts +29 -0
  79. package/dist/types/sdk/host/session-runtime.d.ts +2 -1
  80. package/dist/types/sdk/lifecycle-session.d.ts +13 -0
  81. package/dist/types/sdk/router/session-router.d.ts +27 -1
  82. package/dist/types/sdk/session.d.ts +11 -1
  83. package/dist/types/session/agent-session.d.ts +36 -6
  84. package/dist/types/session/artifacts.d.ts +16 -1
  85. package/dist/types/session/internal/managed-session-storage.d.ts +7 -0
  86. package/dist/types/session/session-manager.d.ts +84 -2
  87. package/dist/types/session/session-staging-paths.d.ts +3 -0
  88. package/dist/types/skill-state/canonical-skills.d.ts +1 -1
  89. package/dist/types/skill-state/workflow-hud.d.ts +12 -21
  90. package/dist/types/skill-state/workflow-mutation-guard.d.ts +1 -0
  91. package/dist/types/slash-commands/helpers/autorouting-status.d.ts +8 -0
  92. package/dist/types/system-prompt.d.ts +23 -0
  93. package/dist/types/task/executor.d.ts +35 -4
  94. package/dist/types/task/index.d.ts +17 -3
  95. package/dist/types/task/receipt.d.ts +3 -1
  96. package/dist/types/task/types.d.ts +144 -0
  97. package/dist/types/tools/atomic-file-write.d.ts +72 -0
  98. package/dist/types/tools/descriptors.d.ts +1 -0
  99. package/dist/types/tools/index.d.ts +36 -1
  100. package/dist/types/tools/move-session.d.ts +40 -0
  101. package/dist/types/tools/python.d.ts +12 -0
  102. package/dist/types/tools/resource-gc.d.ts +0 -3
  103. package/dist/types/tools/search.d.ts +8 -1
  104. package/dist/types/tools/skill.d.ts +9 -4
  105. package/dist/types/tools/subagent.d.ts +33 -3
  106. package/dist/types/utils/xml-escape.d.ts +2 -0
  107. package/dist/types/workflow/workflow-intent-diff.d.ts +2 -2
  108. package/dist/types/workflow/workflow-intent-report-consumer.d.ts +1 -1
  109. package/package.json +11 -12
  110. package/scripts/build-sdk-package-smoke.ts +7 -5
  111. package/scripts/check-autorouting-tier-map.ts +183 -0
  112. package/scripts/generate-sdk-operation-inventory.ts +12 -0
  113. package/scripts/generate-tool-catalog.ts +40 -5
  114. package/src/async/job-manager.ts +5 -2
  115. package/src/autoresearch/dashboard.ts +134 -314
  116. package/src/autoresearch/data-context.ts +35 -0
  117. package/src/autoresearch/git.ts +186 -26
  118. package/src/autoresearch/{helpers.ts → harness.ts} +65 -40
  119. package/src/autoresearch/{prompt-setup.md → prompts/prompt-setup.md} +7 -7
  120. package/src/autoresearch/prompts/prompt.md +103 -0
  121. package/src/autoresearch/prompts.ts +126 -0
  122. package/src/autoresearch/runs.ts +708 -0
  123. package/src/autoresearch/types.ts +9 -150
  124. package/src/capability/index.ts +4 -4
  125. package/src/capability/types.ts +20 -0
  126. package/src/cli/accounts-cli.ts +21 -5
  127. package/src/cli/config-cli.ts +25 -1
  128. package/src/cli/fast-help.ts +1 -2
  129. package/src/cli/list-models.ts +5 -8
  130. package/src/cli/setup-cli.ts +8 -0
  131. package/src/cli.ts +2 -21
  132. package/src/commands/autoresearch.ts +33 -0
  133. package/src/commands/coordinator.ts +15 -0
  134. package/src/commands/launch.ts +2 -4
  135. package/src/commands/quick-lane.ts +1 -1
  136. package/src/commands/sdk.ts +24 -0
  137. package/src/commands/state.ts +1 -1
  138. package/src/config/autorouting-contract.ts +474 -0
  139. package/src/config/autorouting-generator.ts +238 -0
  140. package/src/config/autorouting-tier-map.ts +6108 -0
  141. package/src/config/autorouting.ts +120 -0
  142. package/src/config/model-discovery-manager.ts +42 -11
  143. package/src/config/model-registry.ts +348 -68
  144. package/src/config/provider-selection-policy.ts +60 -9
  145. package/src/config/settings-schema.ts +195 -6
  146. package/src/config/settings.ts +62 -1
  147. package/src/config.ts +2 -3
  148. package/src/coordinator/contract.ts +0 -1
  149. package/src/coordinator-mcp/codex-handoff.ts +68 -42
  150. package/src/coordinator-mcp/durability.ts +200 -0
  151. package/src/coordinator-mcp/event-webhook.ts +278 -0
  152. package/src/coordinator-mcp/question-state.ts +109 -46
  153. package/src/coordinator-mcp/server.ts +932 -504
  154. package/src/crash/index-store.ts +126 -22
  155. package/src/crash/record-loader.ts +7 -0
  156. package/src/crash/upstream/relay.ts +82 -34
  157. package/src/cursor.ts +84 -1
  158. package/src/debug/index.ts +26 -0
  159. package/src/defaults/gjc/skills/autoresearch/SKILL.md +116 -0
  160. package/src/defaults/gjc/skills/autoresearch/auto-critic.md +46 -0
  161. package/src/defaults/gjc/skills/autoresearch/auto-iterate.md +42 -0
  162. package/src/defaults/gjc/skills/deep-interview/SKILL.md +17 -37
  163. package/src/defaults/gjc/skills/ralplan/SKILL.md +21 -18
  164. package/src/defaults/gjc/skills/ultragoal/SKILL.md +6 -17
  165. package/src/defaults/gjc/skills/ultragoal/ai-slop-cleaner.md +2 -2
  166. package/src/defaults/gjc-defaults.ts +89 -2
  167. package/src/defaults/gjc-skills.generated.ts +24 -17
  168. package/src/discovery/builtin.ts +18 -5
  169. package/src/discovery/helpers.ts +6 -6
  170. package/src/edit/index.ts +5 -2
  171. package/src/eval/py/executor.ts +8 -0
  172. package/src/extensibility/extensions/loader.ts +135 -13
  173. package/src/extensibility/extensions/wrapper.ts +1 -2
  174. package/src/extensibility/gjc-plugins/schema.ts +1 -1
  175. package/src/extensibility/runtime-skill-discovery.ts +2 -2
  176. package/src/extensibility/skill-management.ts +2 -3
  177. package/src/extensibility/skills.ts +16 -8
  178. package/src/gjc-runtime/autoresearch-runtime.ts +1509 -0
  179. package/src/gjc-runtime/gc-render.ts +0 -1
  180. package/src/gjc-runtime/gc-runtime.ts +1 -11
  181. package/src/gjc-runtime/launch-tmux.ts +3 -3
  182. package/src/gjc-runtime/psmux-detect.ts +2 -2
  183. package/src/gjc-runtime/python-transcript.ts +58 -0
  184. package/src/gjc-runtime/ralplan-runtime.ts +6 -18
  185. package/src/gjc-runtime/session-layout.ts +18 -7
  186. package/src/gjc-runtime/session-state-sidecar.ts +58 -9
  187. package/src/gjc-runtime/state-renderer.ts +0 -38
  188. package/src/gjc-runtime/state-runtime.ts +43 -22
  189. package/src/gjc-runtime/state-schema.ts +1 -1
  190. package/src/gjc-runtime/tmux-common.ts +7 -7
  191. package/src/gjc-runtime/ultragoal-change-set.ts +232 -11
  192. package/src/gjc-runtime/ultragoal-runtime.ts +309 -114
  193. package/src/gjc-runtime/ultragoal-validation-policy.ts +247 -0
  194. package/src/gjc-runtime/workflow-command-ref.ts +15 -11
  195. package/src/gjc-runtime/workflow-manifest.generated.json +413 -532
  196. package/src/gjc-runtime/workflow-manifest.ts +43 -83
  197. package/src/gjc-runtime/workflow-recovery-projection.ts +616 -0
  198. package/src/hooks/skill-keywords.ts +3 -3
  199. package/src/hooks/skill-state.ts +2 -2
  200. package/src/internal-urls/docs-index.generated.ts +26 -24
  201. package/src/lsp/index.ts +25 -5
  202. package/src/main.ts +2 -27
  203. package/src/modes/acp/acp-agent.ts +200 -75
  204. package/src/modes/components/model-selector.ts +216 -13
  205. package/src/modes/components/oauth-selector.ts +5 -5
  206. package/src/modes/components/skill-hud/render.ts +173 -35
  207. package/src/modes/components/smart-routing-panel.ts +531 -0
  208. package/src/modes/components/tool-status-header.ts +6 -3
  209. package/src/modes/controllers/command-controller.ts +17 -7
  210. package/src/modes/controllers/input-controller.ts +44 -9
  211. package/src/modes/controllers/runtime-mcp-command-controller.ts +4 -0
  212. package/src/modes/controllers/selector-controller.ts +298 -65
  213. package/src/modes/interactive-mode.ts +8 -1
  214. package/src/modes/shared/agent-wire/workflow-gate-types.ts +1 -1
  215. package/src/modes/types.ts +9 -1
  216. package/src/prompts/agents/planner.md +1 -1
  217. package/src/prompts/system/system-prompt.md +1 -1
  218. package/src/prompts/system/volatile-project-context.md +2 -1
  219. package/src/prompts/tools/move-session.md +15 -0
  220. package/src/prompts/tools/python.md +12 -0
  221. package/src/prompts/tools/skill-discovery.md +1 -1
  222. package/src/prompts/tools/skill.md +2 -2
  223. package/src/prompts/tools/task-summary.md +1 -0
  224. package/src/prompts/tools/task.md +6 -0
  225. package/src/rlm/artifacts.ts +12 -10
  226. package/src/runtime/optional-runtime-services.ts +3 -2
  227. package/src/runtime/workspace-tree-service.ts +11 -3
  228. package/src/runtime-mcp/config.ts +8 -1
  229. package/src/runtime-mcp/manager.ts +33 -0
  230. package/src/sdk/acp/adapter.ts +48 -1
  231. package/src/sdk/broker/broker.ts +30 -2
  232. package/src/sdk/broker/discovery.ts +217 -2
  233. package/src/sdk/broker/lifecycle.ts +385 -59
  234. package/src/sdk/broker/session-index.ts +243 -48
  235. package/src/sdk/broker/transport.ts +1 -0
  236. package/src/sdk/bus/chat-daemon-control.ts +8 -2
  237. package/src/sdk/bus/chat-daemon-runtime.ts +5 -0
  238. package/src/sdk/bus/index.ts +208 -11
  239. package/src/sdk/bus/kind-aware-reconciliation.ts +16 -0
  240. package/src/sdk/bus/reconciliation-store.ts +27 -5
  241. package/src/sdk/bus/telegram-daemon-contract.ts +12 -2
  242. package/src/sdk/client/client.ts +204 -22
  243. package/src/sdk/host/host.ts +16 -0
  244. package/src/sdk/host/internal-autorouting-state.ts +13 -0
  245. package/src/sdk/host/model-pin.ts +101 -0
  246. package/src/sdk/host/reverse-leases.ts +2 -1
  247. package/src/sdk/host/session-runtime.test.ts +224 -1
  248. package/src/sdk/host/session-runtime.ts +125 -37
  249. package/src/sdk/lifecycle/service.ts +3 -1
  250. package/src/sdk/lifecycle-session.ts +42 -1
  251. package/src/sdk/protocol/operation-inventory.generated.json +66 -0
  252. package/src/sdk/router/session-router.ts +375 -36
  253. package/src/sdk/session.ts +547 -47
  254. package/src/session/account-inventory.ts +1 -1
  255. package/src/session/agent-session.ts +765 -189
  256. package/src/session/artifacts.ts +234 -4
  257. package/src/session/internal/managed-session-storage.ts +91 -4
  258. package/src/session/provider-safety-stop-hint.ts +3 -2
  259. package/src/session/session-manager.ts +775 -16
  260. package/src/session/session-staging-paths.ts +8 -0
  261. package/src/setup/hermes/templates/operator-instructions.v1.md +0 -1
  262. package/src/setup/host-plugin-setup.ts +2 -2
  263. package/src/skill-state/active-state.ts +2 -2
  264. package/src/skill-state/canonical-skills.ts +1 -1
  265. package/src/skill-state/initial-phase.ts +1 -1
  266. package/src/skill-state/workflow-hud.ts +41 -41
  267. package/src/skill-state/workflow-mutation-guard.ts +26 -7
  268. package/src/slash-commands/builtin-registry.ts +77 -0
  269. package/src/slash-commands/helpers/autorouting-status.ts +55 -0
  270. package/src/system-prompt.ts +102 -5
  271. package/src/task/discovery.ts +2 -3
  272. package/src/task/executor.ts +617 -43
  273. package/src/task/index.ts +262 -15
  274. package/src/task/receipt.ts +30 -0
  275. package/src/task/render.ts +11 -1
  276. package/src/task/types.ts +316 -0
  277. package/src/tools/atomic-file-write.ts +594 -0
  278. package/src/tools/bisect.ts +3 -2
  279. package/src/tools/descriptors.test.ts +3 -0
  280. package/src/tools/descriptors.ts +24 -0
  281. package/src/tools/image-gen.ts +5 -1
  282. package/src/tools/index.ts +62 -5
  283. package/src/tools/move-session.ts +107 -0
  284. package/src/tools/python.ts +205 -0
  285. package/src/tools/read.ts +88 -15
  286. package/src/tools/renderers.ts +2 -0
  287. package/src/tools/resource-gc.ts +8 -177
  288. package/src/tools/search.ts +30 -6
  289. package/src/tools/skill-discovery.ts +1 -0
  290. package/src/tools/skill.ts +37 -7
  291. package/src/tools/subagent-render.ts +119 -13
  292. package/src/tools/subagent.ts +91 -74
  293. package/src/tools/tool-catalog.generated.ts +64 -2
  294. package/src/tools/write.ts +16 -12
  295. package/src/utils/xml-escape.ts +14 -0
  296. package/src/web/search/providers/codex.ts +2 -2
  297. package/src/workflow/workflow-intent-diff.ts +13 -12
  298. package/src/workflow/workflow-intent-report-consumer.ts +6 -6
  299. package/dist/types/autoresearch/helpers.d.ts +0 -24
  300. package/dist/types/autoresearch/index.d.ts +0 -2
  301. package/dist/types/autoresearch/state.d.ts +0 -17
  302. package/dist/types/autoresearch/storage.d.ts +0 -142
  303. package/dist/types/autoresearch/tools/init-experiment.d.ts +0 -31
  304. package/dist/types/autoresearch/tools/log-experiment.d.ts +0 -23
  305. package/dist/types/autoresearch/tools/run-experiment.d.ts +0 -8
  306. package/dist/types/autoresearch/tools/update-notes.d.ts +0 -12
  307. package/dist/types/commands/team.d.ts +0 -24
  308. package/dist/types/gjc-runtime/team-gc.d.ts +0 -7
  309. package/dist/types/gjc-runtime/team-launch.d.ts +0 -61
  310. package/dist/types/gjc-runtime/team-notify.d.ts +0 -45
  311. package/dist/types/gjc-runtime/team-runtime.d.ts +0 -336
  312. package/dist/types/gjc-runtime/team-store.d.ts +0 -168
  313. package/dist/types/gjc-runtime/team-worker-heartbeat.d.ts +0 -43
  314. package/dist/types/gjc-runtime/team-worker-memory-guard.d.ts +0 -92
  315. package/dist/types/gjc-runtime/team-workers.d.ts +0 -83
  316. package/dist/types/rlm/index.d.ts +0 -47
  317. package/dist/types/rlm/preset.d.ts +0 -23
  318. package/dist/types/rlm/python-tool.d.ts +0 -16
  319. package/src/autoresearch/command-resume.md +0 -14
  320. package/src/autoresearch/index.ts +0 -536
  321. package/src/autoresearch/prompt.md +0 -103
  322. package/src/autoresearch/resume-message.md +0 -10
  323. package/src/autoresearch/state.ts +0 -273
  324. package/src/autoresearch/storage.ts +0 -699
  325. package/src/autoresearch/tools/init-experiment.ts +0 -272
  326. package/src/autoresearch/tools/log-experiment.ts +0 -524
  327. package/src/autoresearch/tools/run-experiment.ts +0 -475
  328. package/src/autoresearch/tools/update-notes.ts +0 -109
  329. package/src/commands/rlm.ts +0 -19
  330. package/src/commands/team.ts +0 -238
  331. package/src/defaults/gjc/skills/deep-interview/auto-research-greenfield.md +0 -42
  332. package/src/defaults/gjc/skills/team/SKILL.md +0 -467
  333. package/src/gjc-runtime/team-gc.ts +0 -49
  334. package/src/gjc-runtime/team-launch.ts +0 -404
  335. package/src/gjc-runtime/team-notify.ts +0 -326
  336. package/src/gjc-runtime/team-runtime.ts +0 -5923
  337. package/src/gjc-runtime/team-store.ts +0 -878
  338. package/src/gjc-runtime/team-worker-heartbeat.ts +0 -149
  339. package/src/gjc-runtime/team-worker-memory-guard.ts +0 -363
  340. package/src/gjc-runtime/team-workers.ts +0 -1123
  341. package/src/prompts/system/rlm-report-command.md +0 -1
  342. package/src/prompts/system/rlm-research.md +0 -23
  343. package/src/rlm/index.ts +0 -365
  344. package/src/rlm/preset.ts +0 -76
  345. package/src/rlm/python-tool.ts +0 -68
@@ -1,272 +0,0 @@
1
- import * as path from "node:path";
2
-
3
- import { Text } from "@gajae-code/tui";
4
- import * as z from "zod/v4";
5
- import type { ToolDefinition } from "../../extensibility/extensions";
6
- import type { Theme } from "../../modes/theme/theme";
7
- import { replaceTabs, truncateToWidth } from "../../tools/render-utils";
8
- import * as git from "../../utils/git";
9
- import { parseWorkDirDirtyPaths } from "../git";
10
- import { dedupeStrings, normalizePathSpec } from "../helpers";
11
- import { buildExperimentState } from "../state";
12
- import { openAutoresearchStorage, type SessionRow } from "../storage";
13
- import type { AutoresearchToolFactoryOptions, ExperimentState } from "../types";
14
-
15
- export const HARNESS_FILENAME = "autoresearch.sh";
16
- export const DEFAULT_HARNESS_COMMAND = `bash ${HARNESS_FILENAME}`;
17
- const HARNESS_COMMIT_TITLE = "autoresearch: harness setup";
18
-
19
- const initExperimentSchema = z.object({
20
- name: z.string().describe("experiment name"),
21
- goal: z.string().describe("session goal").optional(),
22
- primary_metric: z.string().describe("primary metric name"),
23
- metric_unit: z.string().describe("metric unit (e.g. ms, µs, mb)").optional(),
24
- direction: z
25
- .enum(["lower", "higher"] as const)
26
- .describe("better direction (default lower)")
27
- .optional(),
28
- secondary_metrics: z.array(z.string()).describe("secondary metric names").optional(),
29
- scope_paths: z.array(z.string()).describe("expected-to-modify paths").optional(),
30
- off_limits: z.array(z.string()).describe("off-limits paths").optional(),
31
- constraints: z.array(z.string()).describe("free-form constraints").optional(),
32
- max_iterations: z.number().describe("soft iteration cap per segment").optional(),
33
- new_segment: z.boolean().describe("bump to a new segment in existing session").optional(),
34
- });
35
-
36
- interface InitExperimentDetails {
37
- state: ExperimentState;
38
- createdSession: boolean;
39
- bumpedSegment: boolean;
40
- abandonedRuns: number;
41
- harnessCommitted: boolean;
42
- baselineCommit: string | null;
43
- }
44
-
45
- export function createInitExperimentTool(
46
- options: AutoresearchToolFactoryOptions,
47
- ): ToolDefinition<typeof initExperimentSchema, InitExperimentDetails> {
48
- return {
49
- name: "init_experiment",
50
- label: "Init Experiment",
51
- description:
52
- "Initialize or reconfigure the autoresearch session. On first call (Phase 1 → Phase 2 transition), requires `./autoresearch.sh` to exist and pending harness changes are auto-committed on an autoresearch branch. Pass `new_segment: true` to start a fresh baseline within an existing session.",
53
- parameters: initExperimentSchema,
54
- defaultInactive: true,
55
- async execute(_toolCallId, params, _signal, _onUpdate, ctx) {
56
- const storage = await openAutoresearchStorage(ctx.cwd);
57
- const runtime = options.getRuntime(ctx);
58
-
59
- const direction = params.direction ?? "lower";
60
- const metricUnit = params.metric_unit ?? "";
61
- const scopePaths = dedupeStrings((params.scope_paths ?? []).map(normalizePathSpec));
62
- const offLimits = dedupeStrings((params.off_limits ?? []).map(normalizePathSpec));
63
- const constraints = dedupeStrings(params.constraints ?? []);
64
- const secondaryMetrics = dedupeStrings(params.secondary_metrics ?? []);
65
- const goal = params.goal?.trim() || null;
66
- const maxIterations =
67
- params.max_iterations !== undefined && Number.isFinite(params.max_iterations) && params.max_iterations > 0
68
- ? Math.floor(params.max_iterations)
69
- : null;
70
- const branch = (await git.branch.current(ctx.cwd)) ?? null;
71
- const onAutoresearchBranch = branch?.startsWith("autoresearch/") ?? false;
72
-
73
- const existing = storage.getActiveSessionForBranch(branch);
74
- const isNewSegmentInit = existing !== null && params.new_segment === true;
75
- const requiresHarness = !existing || isNewSegmentInit;
76
-
77
- if (requiresHarness) {
78
- const harnessExists = await Bun.file(path.join(ctx.cwd, HARNESS_FILENAME)).exists();
79
- if (!harnessExists) {
80
- return {
81
- content: [
82
- {
83
- type: "text",
84
- text: `Error: ./${HARNESS_FILENAME} does not exist. Phase 1 of autoresearch is harness setup — write \`./${HARNESS_FILENAME}\` so it exits 0 and prints \`METRIC <name>=<value>\`, validate it via \`bash ${HARNESS_FILENAME}\`, then call init_experiment again.`,
85
- },
86
- ],
87
- };
88
- }
89
- }
90
-
91
- let harnessCommitted = false;
92
- let commitWarning: string | null = null;
93
- if (requiresHarness && onAutoresearchBranch) {
94
- const dirty = await detectPendingChanges(ctx.cwd);
95
- if (dirty) {
96
- try {
97
- await git.stage.files(ctx.cwd, []);
98
- const message = buildHarnessCommitMessage(goal, params.name);
99
- await git.commit(ctx.cwd, message);
100
- harnessCommitted = true;
101
- } catch (err) {
102
- commitWarning = `Failed to auto-commit harness changes: ${err instanceof Error ? err.message : String(err)}. Recording baseline at current HEAD; discard may not preserve uncommitted harness files.`;
103
- }
104
- }
105
- }
106
-
107
- const baselineCommit = await tryReadHeadSha(ctx.cwd);
108
-
109
- let session: SessionRow;
110
- let createdSession = false;
111
- let bumpedSegment = false;
112
- let abandonedRuns = 0;
113
-
114
- if (!existing) {
115
- session = storage.openSession({
116
- name: params.name,
117
- goal,
118
- primaryMetric: params.primary_metric,
119
- metricUnit,
120
- direction,
121
- preferredCommand: DEFAULT_HARNESS_COMMAND,
122
- branch,
123
- baselineCommit,
124
- maxIterations,
125
- scopePaths,
126
- offLimits,
127
- constraints,
128
- secondaryMetrics,
129
- });
130
- createdSession = true;
131
- } else {
132
- abandonedRuns = storage.abandonPendingRuns(existing.id);
133
- const updates: Parameters<typeof storage.updateSession>[1] = {
134
- goal,
135
- maxIterations,
136
- scopePaths,
137
- offLimits,
138
- constraints,
139
- secondaryMetrics,
140
- primaryMetric: params.primary_metric,
141
- metricUnit,
142
- direction,
143
- branch,
144
- };
145
- if (isNewSegmentInit) {
146
- updates.baselineCommit = baselineCommit;
147
- }
148
- let updated = storage.updateSession(existing.id, updates);
149
- if (isNewSegmentInit) {
150
- updated = storage.bumpSegment(existing.id);
151
- bumpedSegment = true;
152
- }
153
- session = updated;
154
- }
155
-
156
- const loggedRuns = storage.listLoggedRuns(session.id);
157
- const state = buildExperimentState(session, loggedRuns);
158
- runtime.state = state;
159
- runtime.goal = session.goal;
160
- runtime.autoresearchMode = true;
161
- runtime.autoResumeArmed = true;
162
- runtime.lastAutoResumePendingRunNumber = null;
163
- runtime.lastRunDuration = null;
164
- runtime.lastRunAsi = null;
165
- runtime.lastRunArtifactDir = null;
166
- runtime.lastRunNumber = null;
167
- runtime.lastRunSummary = null;
168
- options.dashboard.updateWidget(ctx, runtime);
169
- options.dashboard.requestRender();
170
-
171
- const lines: string[] = [];
172
- if (abandonedRuns > 0) {
173
- lines.push(`Abandoned ${abandonedRuns} pending run${abandonedRuns === 1 ? "" : "s"} before reconfiguring.`);
174
- }
175
- if (harnessCommitted && session.baselineCommit) {
176
- lines.push(`Committed harness setup at ${session.baselineCommit.slice(0, 12)}.`);
177
- }
178
- if (commitWarning) {
179
- lines.push(commitWarning);
180
- }
181
- if (createdSession) {
182
- lines.push(`Started session #${session.id}: ${session.name}`);
183
- } else if (bumpedSegment) {
184
- lines.push(`Bumped segment to ${session.currentSegment} for session #${session.id}: ${session.name}`);
185
- } else {
186
- lines.push(`Updated session #${session.id} (segment ${session.currentSegment}): ${session.name}`);
187
- }
188
- lines.push(
189
- `Metric: ${session.primaryMetric} (${session.metricUnit || "unitless"}, ${session.direction} is better)`,
190
- );
191
- lines.push(`Benchmark entrypoint: ${DEFAULT_HARNESS_COMMAND}`);
192
- if (session.scopePaths.length > 0) {
193
- lines.push(`Files in scope: ${session.scopePaths.join(", ")}`);
194
- }
195
- if (session.offLimits.length > 0) {
196
- lines.push(`Off limits: ${session.offLimits.join(", ")}`);
197
- }
198
- if (session.maxIterations !== null) {
199
- lines.push(`Max iterations per segment: ${session.maxIterations}`);
200
- }
201
- if (session.branch) {
202
- lines.push(`Active branch: ${session.branch}`);
203
- }
204
- if (session.baselineCommit) {
205
- lines.push(`Baseline commit: ${session.baselineCommit.slice(0, 12)}`);
206
- }
207
- if (createdSession) {
208
- lines.push(
209
- "Phase 2: iteration loop is active. Run the baseline experiment with `run_experiment` and log it.",
210
- );
211
- } else if (bumpedSegment) {
212
- lines.push("Run a fresh baseline for the new segment.");
213
- }
214
- if (requiresHarness && !onAutoresearchBranch) {
215
- lines.push(
216
- "Note: not on a dedicated `autoresearch/*` branch — `log_experiment discard` will only revert run-modified files, not reset to baseline.",
217
- );
218
- }
219
-
220
- return {
221
- content: [{ type: "text", text: lines.join("\n") }],
222
- details: {
223
- state,
224
- createdSession,
225
- bumpedSegment,
226
- abandonedRuns,
227
- harnessCommitted,
228
- baselineCommit: session.baselineCommit,
229
- },
230
- };
231
- },
232
- renderCall(args, _options, theme): Text {
233
- return new Text(renderInitCall(args.name, theme), 0, 0);
234
- },
235
- renderResult(result): Text {
236
- const text = replaceTabs(result.content.find(part => part.type === "text")?.text ?? "");
237
- return new Text(text, 0, 0);
238
- },
239
- };
240
- }
241
-
242
- function renderInitCall(name: string, theme: Theme): string {
243
- return `${theme.fg("toolTitle", theme.bold("init_experiment"))} ${theme.fg("accent", truncateToWidth(replaceTabs(name), 100))}`;
244
- }
245
-
246
- async function tryReadHeadSha(cwd: string): Promise<string | null> {
247
- try {
248
- return (await git.head.sha(cwd)) ?? null;
249
- } catch {
250
- return null;
251
- }
252
- }
253
-
254
- async function detectPendingChanges(cwd: string): Promise<boolean> {
255
- try {
256
- const statusText = await git.status(cwd, { porcelainV1: true, untrackedFiles: "all", z: true });
257
- const workDirPrefix = await git.show.prefix(cwd).catch(() => "");
258
- return parseWorkDirDirtyPaths(statusText, workDirPrefix).length > 0;
259
- } catch {
260
- return false;
261
- }
262
- }
263
-
264
- function buildHarnessCommitMessage(goal: string | null, name: string): string {
265
- const lines = [HARNESS_COMMIT_TITLE, "", `Benchmark entrypoint: ${DEFAULT_HARNESS_COMMAND}`];
266
- if (goal) {
267
- lines.push(`Goal: ${goal}`);
268
- } else {
269
- lines.push(`Session: ${name}`);
270
- }
271
- return lines.join("\n");
272
- }