@cohortapp/agent-sdk 2.17.0 → 2.18.4

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 (526) hide show
  1. package/.claude/settings.json +18 -0
  2. package/.env.example +18 -5
  3. package/README.md +1 -0
  4. package/bin/maestro.mjs +62 -0
  5. package/docs/guides/billing-console-keys.md +60 -0
  6. package/docs/guides/front-door-session.md +54 -9
  7. package/docs/guides/mac-mini.md +20 -25
  8. package/docs/guides/setup-wizard.md +1 -1
  9. package/docs/runbooks/fleet-rollout.md +156 -0
  10. package/docs/runbooks/mac-mini-bootstrap.md +12 -14
  11. package/lib/action-executor.js +19 -3
  12. package/lib/budget-guard.mjs +279 -3
  13. package/lib/channels/base-adapter.mjs +3 -1
  14. package/lib/channels/contract.mjs +2 -1
  15. package/lib/channels/inbox-item.mjs +8 -0
  16. package/lib/claude-bin.mjs +5 -6
  17. package/lib/cli/doctor-checks.mjs +141 -10
  18. package/lib/cli/global-setup-extras.mjs +5 -1
  19. package/lib/cli/inbox.mjs +100 -15
  20. package/lib/cli/seat-auth.mjs +463 -0
  21. package/lib/cli/session.mjs +80 -12
  22. package/lib/collective/capture.mjs +8 -6
  23. package/lib/collective/global-config.mjs +63 -1
  24. package/lib/collective/presence.mjs +142 -5
  25. package/lib/comms/send-gate.mjs +559 -1
  26. package/lib/diagnostics/alerts.mjs +49 -0
  27. package/lib/diagnostics/cadence-output-freshness.mjs +288 -0
  28. package/lib/engine/agents/definitions.mjs +343 -0
  29. package/lib/engine/agents/persist.mjs +275 -0
  30. package/lib/engine/agents/runtime.mjs +748 -0
  31. package/lib/engine/agents/usage.mjs +95 -0
  32. package/lib/engine/auth-status.mjs +139 -0
  33. package/lib/engine/budget.mjs +194 -0
  34. package/lib/engine/cli.mjs +1204 -0
  35. package/lib/engine/commands/index.mjs +269 -0
  36. package/lib/engine/context/budget.mjs +219 -0
  37. package/lib/engine/context/cache.mjs +125 -0
  38. package/lib/engine/context/child-env.mjs +215 -0
  39. package/lib/engine/context/compaction.mjs +342 -0
  40. package/lib/engine/context/images.mjs +90 -0
  41. package/lib/engine/context/instructions.mjs +327 -0
  42. package/lib/engine/context/lazy-instructions.mjs +169 -0
  43. package/lib/engine/context/manager.mjs +182 -0
  44. package/lib/engine/context/real-path.mjs +91 -0
  45. package/lib/engine/context/secret-values.mjs +163 -0
  46. package/lib/engine/context/settings.mjs +274 -0
  47. package/lib/engine/context/stream-input.mjs +159 -0
  48. package/lib/engine/guard.mjs +152 -0
  49. package/lib/engine/hooks.mjs +713 -0
  50. package/lib/engine/loop.mjs +560 -0
  51. package/lib/engine/mcp/client.mjs +254 -0
  52. package/lib/engine/mcp/config.mjs +301 -0
  53. package/lib/engine/mcp/http.mjs +201 -0
  54. package/lib/engine/mcp/index.mjs +146 -0
  55. package/lib/engine/mcp/jsonrpc.mjs +147 -0
  56. package/lib/engine/mcp/naming.mjs +66 -0
  57. package/lib/engine/mcp/resources.mjs +89 -0
  58. package/lib/engine/mcp/results.mjs +133 -0
  59. package/lib/engine/mcp/stdio.mjs +137 -0
  60. package/lib/engine/mcp/supervisor.mjs +116 -0
  61. package/lib/engine/messages.mjs +104 -0
  62. package/lib/engine/output/json.mjs +164 -0
  63. package/lib/engine/output/stream-json.mjs +266 -0
  64. package/lib/engine/permissions.mjs +845 -0
  65. package/lib/engine/process-identity.mjs +164 -0
  66. package/lib/engine/process-tree.mjs +551 -0
  67. package/lib/engine/prompt.mjs +60 -0
  68. package/lib/engine/session/store.mjs +299 -0
  69. package/lib/engine/session-runtime/args.mjs +97 -0
  70. package/lib/engine/session-runtime/host.mjs +143 -0
  71. package/lib/engine/session-runtime/inbox.mjs +122 -0
  72. package/lib/engine/session-runtime/notifications.mjs +129 -0
  73. package/lib/engine/session-runtime/registry.mjs +328 -0
  74. package/lib/engine/session-runtime/runner.mjs +344 -0
  75. package/lib/engine/session-runtime/socket.mjs +212 -0
  76. package/lib/engine/session-runtime/wakeup.mjs +115 -0
  77. package/lib/engine/skills/index.mjs +321 -0
  78. package/lib/engine/tools/bash-background.mjs +533 -0
  79. package/lib/engine/tools/bash.mjs +216 -0
  80. package/lib/engine/tools/edit.mjs +97 -0
  81. package/lib/engine/tools/glob.mjs +81 -0
  82. package/lib/engine/tools/grep.mjs +224 -0
  83. package/lib/engine/tools/index.mjs +84 -0
  84. package/lib/engine/tools/list-agents.mjs +32 -0
  85. package/lib/engine/tools/ls.mjs +127 -0
  86. package/lib/engine/tools/monitor.mjs +82 -0
  87. package/lib/engine/tools/notebook-edit.mjs +218 -0
  88. package/lib/engine/tools/read.mjs +103 -0
  89. package/lib/engine/tools/schedule-wakeup.mjs +45 -0
  90. package/lib/engine/tools/schema.mjs +144 -0
  91. package/lib/engine/tools/send-message.mjs +77 -0
  92. package/lib/engine/tools/session.mjs +70 -0
  93. package/lib/engine/tools/todo.mjs +144 -0
  94. package/lib/engine/tools/toolsearch.mjs +217 -0
  95. package/lib/engine/tools/walk.mjs +193 -0
  96. package/lib/engine/tools/web-switch.mjs +31 -0
  97. package/lib/engine/tools/webfetch-html.mjs +387 -0
  98. package/lib/engine/tools/webfetch-net.mjs +340 -0
  99. package/lib/engine/tools/webfetch.mjs +198 -0
  100. package/lib/engine/tools/websearch.mjs +91 -0
  101. package/lib/engine/tools/workflow.mjs +95 -0
  102. package/lib/engine/tools/write.mjs +76 -0
  103. package/lib/engine/tui/line-editor.mjs +137 -0
  104. package/lib/engine/tui/render.mjs +86 -0
  105. package/lib/engine/tui/tui.mjs +274 -0
  106. package/lib/engine/wire/anthropic-messages.mjs +263 -0
  107. package/lib/engine/wire/effort.mjs +36 -0
  108. package/lib/engine/wire/errors.mjs +496 -0
  109. package/lib/engine/wire/http.mjs +441 -0
  110. package/lib/engine/wire/index.mjs +76 -0
  111. package/lib/engine/wire/openai-chat.mjs +332 -0
  112. package/lib/engine/wire/prompt-cache.mjs +79 -0
  113. package/lib/engine/wire/search.mjs +140 -0
  114. package/lib/engine/wire/sse.mjs +114 -0
  115. package/lib/engine/wire/stall.mjs +349 -0
  116. package/lib/engine/wire/token-provider.mjs +175 -0
  117. package/lib/engine/wire/usage.mjs +192 -0
  118. package/lib/engine/workflow/host.mjs +524 -0
  119. package/lib/engine/workflow/journal.mjs +188 -0
  120. package/lib/engine/workflow/json-schema.mjs +171 -0
  121. package/lib/engine/workflow/meta.mjs +329 -0
  122. package/lib/engine/workflow/notifications.mjs +52 -0
  123. package/lib/engine/workflow/runtime.mjs +447 -0
  124. package/lib/engine/workflow/sandbox.mjs +534 -0
  125. package/lib/engine/workflow/worker.mjs +141 -0
  126. package/lib/engine/workflow/worktree.mjs +74 -0
  127. package/lib/execution/disposition.mjs +1 -1
  128. package/lib/execution/intake.mjs +10 -0
  129. package/lib/execution/surface-policy.mjs +15 -0
  130. package/lib/learning/curator.mjs +8 -6
  131. package/lib/learning/reflect.mjs +8 -6
  132. package/lib/model-router/catalog/cohort.yaml +137 -0
  133. package/lib/model-router/catalog.mjs +118 -1
  134. package/lib/model-router/failover.mjs +67 -16
  135. package/lib/model-router/llm-task.mjs +39 -3
  136. package/lib/model-router/resolve.mjs +89 -3
  137. package/lib/model-router/spawn.mjs +46 -47
  138. package/lib/model-router/taxonomy.mjs +126 -4
  139. package/lib/org/cost-sync.mjs +141 -11
  140. package/lib/org/inbound/broadcast.mjs +289 -0
  141. package/lib/org/inbound/collective.mjs +375 -0
  142. package/lib/org/inbound/directedness.mjs +96 -8
  143. package/lib/org/inbound/facts.mjs +78 -2
  144. package/lib/org/inbound/project.mjs +22 -0
  145. package/lib/org/inbound/surfaces.mjs +14 -0
  146. package/lib/org/llm-token.mjs +879 -0
  147. package/lib/org/mesh.mjs +61 -0
  148. package/lib/org/messaging.mjs +3 -1
  149. package/lib/org/protocol.checksum +1 -1
  150. package/lib/org/protocol.mjs +15 -0
  151. package/lib/org/quota.mjs +520 -0
  152. package/lib/org/tool-surface.mjs +104 -16
  153. package/lib/org/ui-parity.mjs +16 -1
  154. package/lib/org/work-ledger.mjs +37 -6
  155. package/lib/rate-guard.mjs +114 -1
  156. package/lib/resource-governor.mjs +41 -6
  157. package/lib/runtime/adapter.mjs +823 -0
  158. package/lib/runtime/child-env.mjs +191 -0
  159. package/lib/runtime/legacy-shell-guard.mjs +97 -0
  160. package/lib/runtime/seat-engine.mjs +162 -0
  161. package/lib/session/ask-ledger.mjs +271 -0
  162. package/lib/session/current-work.mjs +676 -0
  163. package/lib/session/feed-core.mjs +40 -3
  164. package/lib/session/launch-args.mjs +56 -4
  165. package/lib/session/status-summary.mjs +26 -9
  166. package/lib/session/upgrade-notice.mjs +42 -0
  167. package/lib/setup/claude-probe.mjs +117 -13
  168. package/lib/setup/enrich.mjs +13 -10
  169. package/lib/setup/sections/model.mjs +39 -13
  170. package/lib/telemetry/collect.mjs +208 -9
  171. package/lib/upgrade/ignored-drift.mjs +105 -0
  172. package/lib/voice/post-call-brief.mjs +30 -17
  173. package/package.json +13 -3
  174. package/plugins/maestro-skills/skills/board-work.md +5 -0
  175. package/plugins/maestro-skills/skills/inbound-triage.md +56 -15
  176. package/plugins/maestro-skills/skills/main-session.md +18 -7
  177. package/scripts/ci/check-tarball-fidelity.mjs +126 -2
  178. package/scripts/ci/run-tests.mjs +47 -19
  179. package/scripts/cohort-llm/api-key-helper.mjs +92 -0
  180. package/scripts/collective/hook-runner.mjs +29 -2
  181. package/scripts/continuous-monitor.sh +13 -0
  182. package/scripts/cost/track-claude-usage.mjs +15 -0
  183. package/scripts/daemon/agent-daemon.mjs +408 -20
  184. package/scripts/daemon/assurance.mjs +48 -12
  185. package/scripts/daemon/cadence-consumer.mjs +218 -68
  186. package/scripts/daemon/cadence-handlers.mjs +73 -4
  187. package/scripts/daemon/classifier.mjs +75 -26
  188. package/scripts/daemon/context-compiler.mjs +51 -37
  189. package/scripts/daemon/deliver.mjs +30 -1
  190. package/scripts/daemon/dispatcher.mjs +595 -149
  191. package/scripts/daemon/health.mjs +14 -1
  192. package/scripts/daemon/maestro-daemon.mjs +11 -0
  193. package/scripts/daemon/prompt-builder.mjs +24 -0
  194. package/scripts/daemon/responder.mjs +246 -79
  195. package/scripts/daemon/sdk-version.mjs +98 -16
  196. package/scripts/eval/probe-gateway.mjs +635 -0
  197. package/scripts/eval/replay/extract.mjs +270 -0
  198. package/scripts/eval/replay/grade.mjs +260 -0
  199. package/scripts/eval/replay/lib/config.mjs +50 -0
  200. package/scripts/eval/replay/lib/effects.mjs +65 -0
  201. package/scripts/eval/replay/lib/fixture.mjs +188 -0
  202. package/scripts/eval/replay/lib/judge.mjs +72 -0
  203. package/scripts/eval/replay/lib/redact.mjs +136 -0
  204. package/scripts/eval/replay/lib/sandbox.mjs +170 -0
  205. package/scripts/eval/replay/lib/schema-check.mjs +63 -0
  206. package/scripts/eval/replay/lib/transcript.mjs +76 -0
  207. package/scripts/eval/replay/mcp-replay-stub.mjs +101 -0
  208. package/scripts/eval/replay/report.mjs +185 -0
  209. package/scripts/eval/replay/run.mjs +404 -0
  210. package/scripts/fleet/rollout.mjs +1094 -0
  211. package/scripts/hooks/pre-send-audit.sh +36 -245
  212. package/scripts/hooks/pre-write-yaml-validate.mjs +275 -0
  213. package/scripts/hooks/validate-state-yaml.sh +190 -0
  214. package/scripts/huddle/huddle-llm.mjs +361 -0
  215. package/scripts/huddle/huddle-server.mjs +46 -121
  216. package/scripts/local-triggers/autoupdate.sh +448 -78
  217. package/scripts/local-triggers/run-trigger.sh +13 -0
  218. package/scripts/maintenance/pin-integrity.mjs +364 -0
  219. package/scripts/poll-slack-events.sh +41 -9
  220. package/scripts/poller/slack-socket-mode.mjs +28 -3
  221. package/scripts/session/supervisor.mjs +80 -13
  222. package/scripts/spawn-session.sh +13 -0
  223. package/bin/maestro.test.mjs +0 -1574
  224. package/lib/action-executor.test.mjs +0 -871
  225. package/lib/archetype.test.mjs +0 -132
  226. package/lib/assurance/plan-note.test.mjs +0 -234
  227. package/lib/assurance/room-budget.test.mjs +0 -486
  228. package/lib/assurance/tier.test.mjs +0 -174
  229. package/lib/autonomy.test.mjs +0 -66
  230. package/lib/backlog.test.mjs +0 -302
  231. package/lib/backup/policy.test.mjs +0 -305
  232. package/lib/budget-escalate.test.mjs +0 -232
  233. package/lib/budget-guard.envelope.test.mjs +0 -476
  234. package/lib/budget-guard.test.mjs +0 -427
  235. package/lib/cadence-bus-requeue.test.mjs +0 -83
  236. package/lib/cadence-bus-schedule.test.mjs +0 -194
  237. package/lib/cadence-bus.test.mjs +0 -720
  238. package/lib/cadences.test.mjs +0 -230
  239. package/lib/capability/inventory.test.mjs +0 -232
  240. package/lib/capability.test.mjs +0 -78
  241. package/lib/channels/base-adapter.test.mjs +0 -590
  242. package/lib/channels/channels.test.mjs +0 -371
  243. package/lib/channels/contract.test.mjs +0 -162
  244. package/lib/channels/inbox-item.test.mjs +0 -368
  245. package/lib/channels/orgmail/adapter.test.mjs +0 -448
  246. package/lib/channels/pairing.test.mjs +0 -270
  247. package/lib/channels/repeat-suppressor.test.mjs +0 -134
  248. package/lib/channels/slack-adapter.test.mjs +0 -212
  249. package/lib/channels/telegram-adapter.test.mjs +0 -306
  250. package/lib/channels/voice/adapter.test.mjs +0 -278
  251. package/lib/channels/whatsapp/adapter-baileys.test.mjs +0 -359
  252. package/lib/channels/whatsapp/baileys-typing.test.mjs +0 -154
  253. package/lib/charter.test.mjs +0 -89
  254. package/lib/claude-bin.test.mjs +0 -131
  255. package/lib/cli/board.test.mjs +0 -227
  256. package/lib/cli/design.test.mjs +0 -270
  257. package/lib/cli/doctor-checks.test.mjs +0 -336
  258. package/lib/cli/global-setup-extras.test.mjs +0 -462
  259. package/lib/cli/inbox.test.mjs +0 -230
  260. package/lib/cli/session-ack.test.mjs +0 -63
  261. package/lib/cli/session.test.mjs +0 -613
  262. package/lib/collective/capture.test.mjs +0 -121
  263. package/lib/collective/cards.test.mjs +0 -114
  264. package/lib/collective/config.test.mjs +0 -123
  265. package/lib/collective/global-config.test.mjs +0 -220
  266. package/lib/collective/global-skills.test.mjs +0 -126
  267. package/lib/collective/presence.test.mjs +0 -95
  268. package/lib/collective/recall.test.mjs +0 -116
  269. package/lib/collective/vendor-skills.test.mjs +0 -306
  270. package/lib/comms/send-gate.test.mjs +0 -770
  271. package/lib/comms.test.mjs +0 -41
  272. package/lib/context/budget.test.mjs +0 -252
  273. package/lib/context/history-scope.test.mjs +0 -79
  274. package/lib/cost/ledger-row.test.mjs +0 -183
  275. package/lib/design/design-md.test.mjs +0 -318
  276. package/lib/design/fixtures/DESIGN.golden.md +0 -238
  277. package/lib/design/fixtures/PRODUCT.golden.md +0 -67
  278. package/lib/design/fixtures/foundation.json +0 -133
  279. package/lib/design/refresh-gate.test.mjs +0 -144
  280. package/lib/design/write.test.mjs +0 -241
  281. package/lib/diagnostics/alerts.test.mjs +0 -318
  282. package/lib/diagnostics/backup-freshness.test.mjs +0 -185
  283. package/lib/diagnostics/counters.test.mjs +0 -206
  284. package/lib/diagnostics/events.test.mjs +0 -290
  285. package/lib/diagnostics/otel.test.mjs +0 -196
  286. package/lib/diagnostics/trace.test.mjs +0 -251
  287. package/lib/env-compat.test.mjs +0 -104
  288. package/lib/execution/disposition.test.mjs +0 -553
  289. package/lib/execution/drive.test.mjs +0 -270
  290. package/lib/execution/effects.test.mjs +0 -344
  291. package/lib/execution/intake.test.mjs +0 -389
  292. package/lib/execution/journal.test.mjs +0 -261
  293. package/lib/execution/match.test.mjs +0 -235
  294. package/lib/execution/pipeline.test.mjs +0 -392
  295. package/lib/execution/route.test.mjs +0 -186
  296. package/lib/execution/surface-policy.test.mjs +0 -162
  297. package/lib/fs-atomic.test.mjs +0 -72
  298. package/lib/fs-ownership.test.mjs +0 -158
  299. package/lib/goals/admission.test.mjs +0 -164
  300. package/lib/goals/classify.test.mjs +0 -167
  301. package/lib/goals/collaborate.test.mjs +0 -336
  302. package/lib/goals/gaps.test.mjs +0 -284
  303. package/lib/goals/loop.test.mjs +0 -845
  304. package/lib/hooks/bus.test.mjs +0 -387
  305. package/lib/identity/persona.test.mjs +0 -142
  306. package/lib/kpi-sensors.test.mjs +0 -278
  307. package/lib/kpi.test.mjs +0 -244
  308. package/lib/learning/config.test.mjs +0 -75
  309. package/lib/learning/counters.test.mjs +0 -69
  310. package/lib/learning/curator-consolidate.test.mjs +0 -238
  311. package/lib/learning/curator.test.mjs +0 -106
  312. package/lib/learning/reflect.test.mjs +0 -0
  313. package/lib/learning/session-index.test.mjs +0 -125
  314. package/lib/learning/skill-writer.test.mjs +0 -210
  315. package/lib/mandate/audit.test.mjs +0 -195
  316. package/lib/mandate/contract.test.mjs +0 -185
  317. package/lib/mandate/derive.test.mjs +0 -274
  318. package/lib/mandate/model.test.mjs +0 -164
  319. package/lib/mandate/refresh.test.mjs +0 -389
  320. package/lib/mcp/server.test.mjs +0 -426
  321. package/lib/model-router/auth-profiles.test.mjs +0 -580
  322. package/lib/model-router/catalog.test.mjs +0 -385
  323. package/lib/model-router/economics.test.mjs +0 -438
  324. package/lib/model-router/failover.test.mjs +0 -439
  325. package/lib/model-router/health.test.mjs +0 -338
  326. package/lib/model-router/integration-coverage.test.mjs +0 -831
  327. package/lib/model-router/integration.test.mjs +0 -564
  328. package/lib/model-router/ledger.test.mjs +0 -415
  329. package/lib/model-router/llm-task.test.mjs +0 -392
  330. package/lib/model-router/org-credentials.test.mjs +0 -265
  331. package/lib/model-router/pricing-refresh.test.mjs +0 -286
  332. package/lib/model-router/reconcile.test.mjs +0 -316
  333. package/lib/model-router/repair.test.mjs +0 -180
  334. package/lib/model-router/spawn.test.mjs +0 -446
  335. package/lib/model-router/taxonomy.test.mjs +0 -410
  336. package/lib/model-router.test.mjs +0 -1207
  337. package/lib/org/activity.test.mjs +0 -134
  338. package/lib/org/approvals.test.mjs +0 -216
  339. package/lib/org/awareness.test.mjs +0 -159
  340. package/lib/org/board-mine-cache.test.mjs +0 -53
  341. package/lib/org/board.test.mjs +0 -187
  342. package/lib/org/bootstrap-context.test.mjs +0 -153
  343. package/lib/org/client.test.mjs +0 -1206
  344. package/lib/org/cohort-client.test.mjs +0 -126
  345. package/lib/org/cost-sync.test.mjs +0 -153
  346. package/lib/org/doctor.test.mjs +0 -346
  347. package/lib/org/engagement-ledger.test.mjs +0 -112
  348. package/lib/org/engagement.test.mjs +0 -739
  349. package/lib/org/handoff.test.mjs +0 -269
  350. package/lib/org/inbound/directedness.test.mjs +0 -668
  351. package/lib/org/inbound/facts.test.mjs +0 -471
  352. package/lib/org/inbound/hydrate.test.mjs +0 -908
  353. package/lib/org/inbound/index.test.mjs +0 -429
  354. package/lib/org/inbound/project.test.mjs +0 -287
  355. package/lib/org/integration-tools.test.mjs +0 -160
  356. package/lib/org/keys.test.mjs +0 -92
  357. package/lib/org/knowledge.test.mjs +0 -326
  358. package/lib/org/leases.test.mjs +0 -235
  359. package/lib/org/mesh-directives.test.mjs +0 -110
  360. package/lib/org/mesh-integration.test.mjs +0 -127
  361. package/lib/org/mesh.test.mjs +0 -400
  362. package/lib/org/messaging.test.mjs +0 -471
  363. package/lib/org/param-contract.test.mjs +0 -477
  364. package/lib/org/policy.test.mjs +0 -237
  365. package/lib/org/protocol.checksum.test.mjs +0 -90
  366. package/lib/org/protocol.test.mjs +0 -323
  367. package/lib/org/push.test.mjs +0 -792
  368. package/lib/org/registry.test.mjs +0 -100
  369. package/lib/org/resource-tools.test.mjs +0 -361
  370. package/lib/org/tool-access.test.mjs +0 -144
  371. package/lib/org/tool-surface-integration.test.mjs +0 -120
  372. package/lib/org/tool-surface.test.mjs +0 -1268
  373. package/lib/org/typing.test.mjs +0 -291
  374. package/lib/org/ui-parity.test.mjs +0 -560
  375. package/lib/org/verify.test.mjs +0 -194
  376. package/lib/org/work-ledger.test.mjs +0 -273
  377. package/lib/plan/adoption-e2e.test.mjs +0 -366
  378. package/lib/plan/budget-enforcement.test.mjs +0 -400
  379. package/lib/plan/compile.test.mjs +0 -382
  380. package/lib/plan/emit.test.mjs +0 -269
  381. package/lib/plan/explain.test.mjs +0 -188
  382. package/lib/prompts/parallelism.test.mjs +0 -177
  383. package/lib/rag/rag.test.mjs +0 -505
  384. package/lib/rate-guard.test.mjs +0 -272
  385. package/lib/reactive-gate.test.mjs +0 -57
  386. package/lib/render.test.mjs +0 -68
  387. package/lib/resource-governor.test.mjs +0 -488
  388. package/lib/scheduling/dynamic-jobs.test.mjs +0 -344
  389. package/lib/scheduling/jitter.test.mjs +0 -140
  390. package/lib/secrets/broker.test.mjs +0 -280
  391. package/lib/secrets/providers.test.mjs +0 -274
  392. package/lib/security/audit-engine.test.mjs +0 -424
  393. package/lib/security/coerce-args.test.mjs +0 -281
  394. package/lib/security/dangerous-tools.test.mjs +0 -68
  395. package/lib/security/external-content.test.mjs +0 -84
  396. package/lib/security/redact.test.mjs +0 -441
  397. package/lib/security/secret-equal.test.mjs +0 -55
  398. package/lib/session/config.test.mjs +0 -92
  399. package/lib/session/feed-core.test.mjs +0 -198
  400. package/lib/session/first-run.test.mjs +0 -121
  401. package/lib/session/frontdoor.test.mjs +0 -205
  402. package/lib/session/handoffs.test.mjs +0 -183
  403. package/lib/session/identity.test.mjs +0 -180
  404. package/lib/session/inbox-claims.test.mjs +0 -286
  405. package/lib/session/launch-args.test.mjs +0 -157
  406. package/lib/session/liveness.test.mjs +0 -100
  407. package/lib/session/status-summary.test.mjs +0 -118
  408. package/lib/session-permissions.test.mjs +0 -120
  409. package/lib/setup/claude-probe.test.mjs +0 -187
  410. package/lib/setup/completeness.test.mjs +0 -110
  411. package/lib/setup/context-pack.test.mjs +0 -89
  412. package/lib/setup/enrich.test.mjs +0 -115
  413. package/lib/setup/enroll-from-cohort.test.mjs +0 -300
  414. package/lib/setup/integration.test.mjs +0 -162
  415. package/lib/setup/io.test.mjs +0 -77
  416. package/lib/setup/runner.test.mjs +0 -132
  417. package/lib/setup/sections/identity.test.mjs +0 -234
  418. package/lib/setup/sections/inventory.test.mjs +0 -198
  419. package/lib/setup/sections/learning.test.mjs +0 -81
  420. package/lib/setup/sections/mandate.test.mjs +0 -388
  421. package/lib/setup/sections/messaging.test.mjs +0 -127
  422. package/lib/setup/sections/model.test.mjs +0 -240
  423. package/lib/setup/sections/org.test.mjs +0 -346
  424. package/lib/setup/sections/orgmail.test.mjs +0 -118
  425. package/lib/setup/sections/recovery.test.mjs +0 -98
  426. package/lib/setup/sections/subagents.test.mjs +0 -429
  427. package/lib/setup/sections/verify.test.mjs +0 -175
  428. package/lib/setup/sot.test.mjs +0 -81
  429. package/lib/setup/state.test.mjs +0 -115
  430. package/lib/singleton.test.mjs +0 -151
  431. package/lib/subagents/cli.test.mjs +0 -389
  432. package/lib/subagents/client.test.mjs +0 -309
  433. package/lib/subagents/gap.test.mjs +0 -234
  434. package/lib/subagents/lock.test.mjs +0 -248
  435. package/lib/subagents/manifest.test.mjs +0 -175
  436. package/lib/subagents/refs.test.mjs +0 -204
  437. package/lib/subagents/resolve.test.mjs +0 -422
  438. package/lib/subagents/schema.test.mjs +0 -328
  439. package/lib/telemetry/alerts.test.mjs +0 -109
  440. package/lib/telemetry/collect.test.mjs +0 -1274
  441. package/lib/tool-definitions-integration.test.mjs +0 -83
  442. package/lib/tool-definitions.test.mjs +0 -437
  443. package/lib/upgrade/global-refresh.test.mjs +0 -65
  444. package/lib/upgrade/launchd-reconcile.test.mjs +0 -272
  445. package/lib/upgrade/post-steps.test.mjs +0 -200
  446. package/lib/upgrade/verify.test.mjs +0 -164
  447. package/lib/util/fetch-timeout.test.mjs +0 -202
  448. package/lib/util/reconnect.test.mjs +0 -369
  449. package/lib/util/unhandled.test.mjs +0 -216
  450. package/lib/voice/outbound.test.mjs +0 -69
  451. package/lib/voice/session-rotation.test.mjs +0 -114
  452. package/lib/voice/stt.test.mjs +0 -226
  453. package/lib/voice/voice.test.mjs +0 -990
  454. package/scripts/cadence/enqueue-cadence-tick.test.mjs +0 -187
  455. package/scripts/ci/check-docs-accuracy.test.mjs +0 -409
  456. package/scripts/ci/check-durable-write-seam.test.mjs +0 -90
  457. package/scripts/ci/check-no-build-artifacts.test.mjs +0 -71
  458. package/scripts/ci/check-no-residual-identity.test.mjs +0 -202
  459. package/scripts/ci/check-skill-packs.test.mjs +0 -495
  460. package/scripts/ci/check-subagent-frontmatter.test.mjs +0 -124
  461. package/scripts/ci/check.test.mjs +0 -194
  462. package/scripts/ci/conformance-org-api.test.mjs +0 -425
  463. package/scripts/cloud-relay/voice/relay-identity.test.mjs +0 -96
  464. package/scripts/collective/hook-runner.test.mjs +0 -173
  465. package/scripts/cost/fleet-digest.test.mjs +0 -207
  466. package/scripts/cost/track-claude-usage-pricing.test.mjs +0 -183
  467. package/scripts/cost/track-claude-usage.test.mjs +0 -148
  468. package/scripts/daemon/agent-daemon-board-mine.test.mjs +0 -96
  469. package/scripts/daemon/agent-daemon-design.test.mjs +0 -238
  470. package/scripts/daemon/agent-daemon-frontdoor.test.mjs +0 -60
  471. package/scripts/daemon/agent-daemon.test.mjs +0 -995
  472. package/scripts/daemon/assurance-e2e.test.mjs +0 -613
  473. package/scripts/daemon/assurance.test.mjs +0 -1791
  474. package/scripts/daemon/board-mirror.test.mjs +0 -165
  475. package/scripts/daemon/cadence-consumer-frontdoor.test.mjs +0 -393
  476. package/scripts/daemon/cadence-consumer-governance.test.mjs +0 -276
  477. package/scripts/daemon/cadence-consumer.test.mjs +0 -776
  478. package/scripts/daemon/cadence-handlers.test.mjs +0 -837
  479. package/scripts/daemon/classifier-identity.test.mjs +0 -137
  480. package/scripts/daemon/classifier.test.mjs +0 -266
  481. package/scripts/daemon/classify-kind.test.mjs +0 -40
  482. package/scripts/daemon/context-compiler.test.mjs +0 -406
  483. package/scripts/daemon/deliver.test.mjs +0 -564
  484. package/scripts/daemon/dispatcher-cooldown.test.mjs +0 -122
  485. package/scripts/daemon/dispatcher-governance.test.mjs +0 -1013
  486. package/scripts/daemon/dispatcher-resume.test.mjs +0 -166
  487. package/scripts/daemon/dispatcher-session-continuity.test.mjs +0 -365
  488. package/scripts/daemon/execution-ladder.test.mjs +0 -470
  489. package/scripts/daemon/goal-steward-cadence.test.mjs +0 -312
  490. package/scripts/daemon/inbox-deferral-session.test.mjs +0 -49
  491. package/scripts/daemon/inbox-deferral.test.mjs +0 -336
  492. package/scripts/daemon/inbox-wake.test.mjs +0 -199
  493. package/scripts/daemon/integration.test.mjs +0 -149
  494. package/scripts/daemon/lib/self-echo.test.mjs +0 -153
  495. package/scripts/daemon/lib/session-router.test.mjs +0 -554
  496. package/scripts/daemon/prompt-builder-preamble.test.mjs +0 -210
  497. package/scripts/daemon/prompt-builder.test.mjs +0 -556
  498. package/scripts/daemon/responder-cost.test.mjs +0 -68
  499. package/scripts/daemon/responder-history.test.mjs +0 -221
  500. package/scripts/daemon/sdk-version.test.mjs +0 -31
  501. package/scripts/daemon/session-lock.test.mjs +0 -252
  502. package/scripts/daemon/session-outcomes.test.mjs +0 -533
  503. package/scripts/daemon/typing-registry.test.mjs +0 -102
  504. package/scripts/hooks/pre-send-audit.test.mjs +0 -354
  505. package/scripts/huddle/huddle-prompt.test.mjs +0 -176
  506. package/scripts/local-triggers/autoupdate.test.mjs +0 -518
  507. package/scripts/local-triggers/generate-plists.test.mjs +0 -456
  508. package/scripts/media-generation/brand-clause.test.mjs +0 -135
  509. package/scripts/org/send-orgmail.first-contact.test.mjs +0 -102
  510. package/scripts/poller/inbox-privilege-injection.test.mjs +0 -167
  511. package/scripts/poller/inbox-scan-poller.test.mjs +0 -295
  512. package/scripts/poller/lib/cloud-relay-dedup.test.mjs +0 -133
  513. package/scripts/poller/slack-socket-mode.test.mjs +0 -805
  514. package/scripts/poller-launchd/install.test.mjs +0 -243
  515. package/scripts/restore-from-backup.test.mjs +0 -181
  516. package/scripts/session/feed.test.mjs +0 -196
  517. package/scripts/session/supervisor-sh.test.mjs +0 -218
  518. package/scripts/session/supervisor.test.mjs +0 -482
  519. package/scripts/setup/configure-macos.test.mjs +0 -306
  520. package/scripts/setup/gen-subagent-manifest.test.mjs +0 -124
  521. package/scripts/setup/generate-agent-package-json.test.mjs +0 -143
  522. package/scripts/setup/generate-capability.test.mjs +0 -134
  523. package/scripts/setup/init-agent.test.mjs +0 -370
  524. package/scripts/setup/init-skill-marketplace.test.mjs +0 -193
  525. package/scripts/vendor/sync-skill-packs.test.mjs +0 -103
  526. package/scripts/watchdog/memory-watchdog.test.mjs +0 -64
@@ -1,392 +0,0 @@
1
- /**
2
- * lib/model-router/llm-task.test.mjs — the governed cheap one-shot path.
3
- *
4
- * Drives llmTask with an injected fetchImpl (no network) over a hermetic catalog,
5
- * asserting the component-brief contract:
6
- * - routes to the CHEAPEST direct-harness model meeting minGrade;
7
- * - writes ONE ledger row (real writeLedgerRow into a tmp dir);
8
- * - tool-less by construction (no tools[] in the request body);
9
- * - non-Anthropic payloads pass through the injected PII mask (§7.4);
10
- * - HTTP/transport faults classify onto the taxonomy and fail-open (no throw);
11
- * - the kill switch forces an Anthropic direct row;
12
- * - missing credential / open breaker / grade floor are honored.
13
- *
14
- * House style: node:test, node:assert/strict, hermetic.
15
- */
16
-
17
- import { test } from "node:test";
18
- import assert from "node:assert/strict";
19
- import { mkdtempSync, rmSync, existsSync, readFileSync, mkdirSync } from "node:fs";
20
- import { tmpdir } from "node:os";
21
- import { join } from "node:path";
22
-
23
- import { llmTask, pickDirectModel, _internals } from "./llm-task.mjs";
24
- import { loadCatalog } from "./catalog.mjs";
25
- import { sumLedgerToday } from "./ledger.mjs";
26
-
27
- // ---------------------------------------------------------------------------
28
- // A hermetic catalog: a cheap grade-B direct row, a cheaper grade-C row, and an
29
- // anthropic direct row. We inject raw YAML via catalogText so no FS is touched.
30
- // ---------------------------------------------------------------------------
31
-
32
- const CHEAP_YAML = `
33
- provider: cheapco
34
- auth_env: CHEAPCO_API_KEY
35
- endpoints:
36
- openai: https://api.cheapco.test
37
- models:
38
- - id: cheap-b
39
- status: available
40
- context_tokens: 200000
41
- cost: { input: 0.20, output: 0.40 }
42
- cost_provenance: { source: official, fetched: "2026-06-10", volatile: false }
43
- compat: { cache_control: none, max_tokens_field: max_tokens }
44
- tool_reliability: B
45
- harness: { session: false, direct: true, batch: false }
46
- - id: cheap-c
47
- status: available
48
- context_tokens: 100000
49
- cost: { input: 0.05, output: 0.10 }
50
- cost_provenance: { source: official, fetched: "2026-06-10", volatile: false }
51
- compat: { cache_control: none, max_tokens_field: max_tokens }
52
- tool_reliability: C
53
- harness: { session: false, direct: true, batch: false }
54
- `;
55
-
56
- const ANTH_YAML = `
57
- provider: anthropic
58
- auth_env: ANTHROPIC_API_KEY
59
- endpoints:
60
- anthropic: https://api.anthropic.com
61
- models:
62
- - id: claude-haiku-4-5
63
- status: available
64
- context_tokens: 200000
65
- cost: { input: 1.0, output: 5.0, cache_read: 0.1, cache_write: 1.25 }
66
- cost_provenance: { source: official, fetched: "2026-06-10", volatile: false }
67
- compat: { cache_control: anthropic }
68
- tool_reliability: A
69
- harness: { session: true, direct: true, batch: true }
70
- `;
71
-
72
- function hermeticCatalog() {
73
- const res = loadCatalog("/tmp/nope", {
74
- catalogText: { cheapco: CHEAP_YAML, anthropic: ANTH_YAML },
75
- });
76
- assert.ok(res.byRef.size >= 3, "catalog parsed three rows");
77
- return res;
78
- }
79
-
80
- /** A config that permits cheapco for public data (otherwise deny-default). */
81
- function publicConfig() {
82
- return {
83
- schema_version: 2,
84
- backends: {
85
- anthropic: { allowed_data_classes: ["public", "internal", "sensitive"] },
86
- cheapco: { allowed_data_classes: ["public"] },
87
- },
88
- defaults: { data_class: "public" },
89
- };
90
- }
91
-
92
- /** A fake OpenAI-compatible response. */
93
- function openaiResponse(text, usage = {}) {
94
- return {
95
- ok: true,
96
- status: 200,
97
- async json() {
98
- return {
99
- choices: [{ message: { role: "assistant", content: text } }],
100
- usage: { prompt_tokens: usage.in ?? 100, completion_tokens: usage.out ?? 20 },
101
- };
102
- },
103
- };
104
- }
105
-
106
- /** A fake anthropic-messages response. */
107
- function anthropicResponse(text, usage = {}) {
108
- return {
109
- ok: true,
110
- status: 200,
111
- async json() {
112
- return {
113
- content: [{ type: "text", text }],
114
- usage: { input_tokens: usage.in ?? 100, output_tokens: usage.out ?? 20 },
115
- };
116
- },
117
- };
118
- }
119
-
120
- const baseDeps = (over = {}) => ({
121
- catalog: hermeticCatalog(),
122
- config: publicConfig(),
123
- env: { CHEAPCO_API_KEY: "cheap-key", ANTHROPIC_API_KEY: "anth-key" },
124
- now: () => 1_700_000_000_000,
125
- isOpen: () => ({ open: false, until: 0, reason: null, strikes: 0 }),
126
- ...over,
127
- });
128
-
129
- // ---------------------------------------------------------------------------
130
- // pickDirectModel
131
- // ---------------------------------------------------------------------------
132
-
133
- test("pickDirectModel returns the cheapest qualifying direct row for minGrade C", () => {
134
- const catalog = hermeticCatalog();
135
- const r = pickDirectModel(catalog, { data_class: "public" }, { ...baseDeps({ catalog }), minGrade: "C" });
136
- assert.equal(r.row.ref, "cheapco/cheap-c", "cheapest (C) wins at minGrade C");
137
- });
138
-
139
- test("pickDirectModel skips below-grade rows for minGrade B", () => {
140
- const catalog = hermeticCatalog();
141
- const r = pickDirectModel(catalog, { data_class: "public" }, { ...baseDeps({ catalog }), minGrade: "B" });
142
- assert.equal(r.row.ref, "cheapco/cheap-b", "cheap-c is grade C, excluded; cheap-b wins");
143
- });
144
-
145
- test("pickDirectModel denies cheapco for non-public data_class (deny-default)", () => {
146
- const catalog = hermeticCatalog();
147
- const r = pickDirectModel(
148
- catalog,
149
- { data_class: "sensitive" },
150
- { ...baseDeps({ catalog }), minGrade: "C" }
151
- );
152
- // cheapco rows denied → anthropic haiku is the only allowed direct row.
153
- assert.equal(r.row.ref, "anthropic/claude-haiku-4-5");
154
- });
155
-
156
- test("pickDirectModel skips a row whose credential is absent", () => {
157
- const catalog = hermeticCatalog();
158
- const r = pickDirectModel(
159
- catalog,
160
- { data_class: "public" },
161
- { ...baseDeps({ catalog, env: { ANTHROPIC_API_KEY: "anth-key" } }), minGrade: "C" }
162
- );
163
- // No CHEAPCO_API_KEY → cheapco skipped → anthropic haiku.
164
- assert.equal(r.row.ref, "anthropic/claude-haiku-4-5");
165
- });
166
-
167
- test("pickDirectModel skips a row with an open breaker", () => {
168
- const catalog = hermeticCatalog();
169
- const r = pickDirectModel(catalog, { data_class: "public" }, {
170
- ...baseDeps({ catalog }),
171
- minGrade: "C",
172
- isOpen: (key) => ({ open: key.startsWith("cheapco"), until: 9e15, reason: "rate_limit", strikes: 1 }),
173
- });
174
- assert.equal(r.row.ref, "anthropic/claude-haiku-4-5", "cheapco breaker open → anthropic");
175
- });
176
-
177
- // ---------------------------------------------------------------------------
178
- // llmTask happy path
179
- // ---------------------------------------------------------------------------
180
-
181
- test("routes to the cheapest direct model, returns text + usage + estCostUSD, writes a ledger row", async () => {
182
- const rows = [];
183
- const out = await llmTask("classify this", {
184
- ...baseDeps(),
185
- minGrade: "C",
186
- request: { task_class: "classify.inbox", data_class: "public", agent: "alex" },
187
- fetchImpl: async (url, init) => {
188
- // tool-less invariant: no tools[] ever sent.
189
- const body = JSON.parse(init.body);
190
- assert.equal(body.tools, undefined, "no tools[] in the request");
191
- assert.match(url, /cheapco/, "cheapest cheapco model selected");
192
- return openaiResponse("spam", { in: 80, out: 4 });
193
- },
194
- writeLedger: (catalog, entry) => rows.push(entry),
195
- });
196
-
197
- assert.equal(out.ok, true);
198
- assert.equal(out.ref, "cheapco/cheap-c");
199
- assert.equal(out.text, "spam");
200
- assert.equal(out.usage.inputTokens, 80);
201
- assert.equal(out.usage.outputTokens, 4);
202
- assert.ok(out.estCostUSD != null);
203
- // Exactly one ledger row, joined to the task + harness + exit reason.
204
- assert.equal(rows.length, 1);
205
- assert.equal(rows[0].ref, "cheapco/cheap-c");
206
- assert.equal(rows[0].harness, "direct");
207
- assert.equal(rows[0].task_class, "classify.inbox");
208
- assert.equal(rows[0].exitReason, "ok");
209
- assert.equal(rows[0].agent, "alex");
210
- });
211
-
212
- test("writes a REAL ledger row to state/cost-tracking via writeLedgerRow", async () => {
213
- const root = mkdtempSync(join(tmpdir(), "maestro-llmtask-"));
214
- mkdirSync(join(root, "state", "cost-tracking"), { recursive: true });
215
- try {
216
- const out = await llmTask("hi", {
217
- ...baseDeps(),
218
- minGrade: "C",
219
- request: { task_class: "lookup.gmail", data_class: "public" },
220
- ledgerDeps: { agentRoot: root, now: () => 1_700_000_000_000 },
221
- fetchImpl: async () => openaiResponse("ok", { in: 50, out: 10 }),
222
- });
223
- assert.equal(out.ok, true);
224
- const sum = sumLedgerToday({ agentRoot: root, now: () => 1_700_000_000_000 });
225
- assert.equal(sum.sessions, 1, "one ledger row written to disk");
226
- assert.ok(sum.spentUSD >= 0);
227
- } finally {
228
- rmSync(root, { recursive: true, force: true });
229
- }
230
- });
231
-
232
- test("non-Anthropic payload passes through the injected PII mask (§7.4)", async () => {
233
- let masked = false;
234
- let sentBody = null;
235
- const out = await llmTask("email me at a@b.com", {
236
- ...baseDeps(),
237
- minGrade: "C",
238
- request: { data_class: "public" },
239
- maskInput: (s) => {
240
- masked = true;
241
- return s.replace(/[\w.]+@[\w.]+/g, "***");
242
- },
243
- fetchImpl: async (url, init) => {
244
- sentBody = JSON.parse(init.body);
245
- return openaiResponse("done");
246
- },
247
- writeLedger: () => {},
248
- });
249
- assert.equal(out.ok, true);
250
- assert.equal(masked, true, "mask invoked for the cheapco (non-anthropic) route");
251
- assert.match(sentBody.messages.at(-1).content, /\*\*\*/, "PII masked in the sent prompt");
252
- assert.ok(!/a@b\.com/.test(sentBody.messages.at(-1).content));
253
- });
254
-
255
- test("kill switch forces an Anthropic direct row (mask NOT applied)", async () => {
256
- let masked = false;
257
- let url = null;
258
- const out = await llmTask("anything", {
259
- ...baseDeps({ env: { CHEAPCO_API_KEY: "k", ANTHROPIC_API_KEY: "a", MAESTRO_ROUTER_FORCE_ANTHROPIC: "1" } }),
260
- minGrade: "C",
261
- request: { data_class: "public" },
262
- maskInput: () => {
263
- masked = true;
264
- return "x";
265
- },
266
- fetchImpl: async (u) => {
267
- url = u;
268
- return anthropicResponse("hello");
269
- },
270
- writeLedger: () => {},
271
- });
272
- assert.equal(out.ok, true);
273
- assert.equal(out.provider, "anthropic");
274
- assert.match(url, /anthropic/);
275
- assert.equal(masked, false, "anthropic route is not PII-masked");
276
- });
277
-
278
- // ---------------------------------------------------------------------------
279
- // fault classification (fail-open, never throws)
280
- // ---------------------------------------------------------------------------
281
-
282
- test("HTTP 429 → rate_limit reason, fail-open (no throw), ledger row written", async () => {
283
- const rows = [];
284
- const out = await llmTask("x", {
285
- ...baseDeps(),
286
- minGrade: "C",
287
- request: { data_class: "public" },
288
- fetchImpl: async () => ({
289
- ok: false,
290
- status: 429,
291
- async json() {
292
- return { error: { message: "rate limit exceeded" } };
293
- },
294
- }),
295
- writeLedger: (c, e) => rows.push(e),
296
- });
297
- assert.equal(out.ok, false);
298
- assert.equal(out.reason, "rate_limit");
299
- assert.equal(rows.length, 1);
300
- assert.equal(rows[0].exitReason, "rate_limit");
301
- });
302
-
303
- test("transport throw → classified network, fail-open", async () => {
304
- const out = await llmTask("x", {
305
- ...baseDeps(),
306
- minGrade: "C",
307
- request: { data_class: "public" },
308
- fetchImpl: async () => {
309
- const e = new Error("getaddrinfo ENOTFOUND api.cheapco.test");
310
- e.code = "ENOTFOUND";
311
- throw e;
312
- },
313
- writeLedger: () => {},
314
- });
315
- assert.equal(out.ok, false);
316
- assert.equal(out.reason, "network");
317
- });
318
-
319
- test("2xx with empty content → empty_response soft-fail", async () => {
320
- const out = await llmTask("x", {
321
- ...baseDeps(),
322
- minGrade: "C",
323
- request: { data_class: "public" },
324
- fetchImpl: async () => openaiResponse(" "),
325
- writeLedger: () => {},
326
- });
327
- assert.equal(out.ok, false);
328
- assert.equal(out.reason, "empty_response");
329
- });
330
-
331
- test("no fetchImpl → no_fetch_impl, does not throw", async () => {
332
- const out = await llmTask("x", { ...baseDeps(), request: { data_class: "public" } });
333
- assert.equal(out.ok, false);
334
- assert.equal(out.reason, "no_fetch_impl");
335
- });
336
-
337
- test("no qualifying candidate (no creds anywhere) → no_direct_candidate", async () => {
338
- const out = await llmTask("x", {
339
- ...baseDeps({ env: {} }),
340
- minGrade: "C",
341
- request: { data_class: "public" },
342
- fetchImpl: async () => openaiResponse("y"),
343
- });
344
- assert.equal(out.ok, false);
345
- assert.ok(["missing_credential", "no_direct_candidate"].includes(out.reason));
346
- });
347
-
348
- test("opts.repair promotes a leaked tool call in the response text", async () => {
349
- const out = await llmTask("call the tool", {
350
- ...baseDeps(),
351
- minGrade: "C",
352
- request: { data_class: "public" },
353
- repair: true,
354
- repairOpts: { idFor: (n) => `id_${n}` },
355
- fetchImpl: async () => openaiResponse('```json\n{"name":"do","input":{"k":1}}\n```'),
356
- writeLedger: () => {},
357
- });
358
- assert.equal(out.ok, true);
359
- assert.equal(out.repaired, true);
360
- assert.equal(out.toolCalls[0].name, "do");
361
- assert.deepEqual(out.toolCalls[0].input, { k: 1 });
362
- });
363
-
364
- test("anthropic wire sends x-api-key + anthropic-version, parses content[]", async () => {
365
- let headers = null;
366
- const out = await llmTask("x", {
367
- ...baseDeps(),
368
- minGrade: "A", // forces the anthropic grade-A row
369
- request: { data_class: "public" },
370
- fetchImpl: async (url, init) => {
371
- headers = init.headers;
372
- assert.match(url, /\/v1\/messages$/);
373
- return anthropicResponse("hi", { in: 10, out: 2 });
374
- },
375
- writeLedger: () => {},
376
- });
377
- assert.equal(out.ok, true);
378
- assert.equal(out.provider, "anthropic");
379
- assert.equal(headers["x-api-key"], "anth-key");
380
- assert.equal(headers["anthropic-version"], "2023-06-01");
381
- assert.equal(out.text, "hi");
382
- });
383
-
384
- test("_internals.wireFor + endpointFor pick the right wire per provider", () => {
385
- const catalog = hermeticCatalog();
386
- const cheap = catalog.byRef.get("cheapco/cheap-c");
387
- const anth = catalog.byRef.get("anthropic/claude-haiku-4-5");
388
- assert.equal(_internals.wireFor(cheap), "openai-completions");
389
- assert.equal(_internals.wireFor(anth), "anthropic-messages");
390
- assert.match(_internals.endpointFor(cheap, "openai-completions"), /cheapco.*chat\/completions/);
391
- assert.match(_internals.endpointFor(anth, "anthropic-messages"), /\/v1\/messages$/);
392
- });
@@ -1,265 +0,0 @@
1
- /**
2
- * org-credentials.test.mjs — the router CONSUMES org credentials (single-SoT
3
- * payoff). Verifies resolveOrgCredentialEnv: leases a provider absent from local
4
- * env; local-set provider is NOT leased (local wins); cache avoids a second
5
- * lease within TTL; re-lease after TTL expiry; Cohort-down / disabled fails open
6
- * to local env; unconfigured-provider error is recorded, not thrown.
7
- *
8
- * Injected fetchImpl — no network. Run: node --test lib/model-router/org-credentials.test.mjs
9
- */
10
- "use strict";
11
-
12
- import { test } from "node:test";
13
- import assert from "node:assert/strict";
14
-
15
- import { resolveOrgCredentialEnv, clearOrgCredentialCache } from "./org-credentials.mjs";
16
-
17
- // Ambient COHORT_* tokens would leak into the client config; clear them so the
18
- // tests are deterministic regardless of the host env.
19
- delete process.env.COHORT_API_TOKEN;
20
- delete process.env.COHORT_TOKEN;
21
-
22
- const BASE = "https://org.cohortapp.com";
23
-
24
- /** Enabled org config pointing the client at BASE with a bearer token. */
25
- function orgCfg() {
26
- return { org: { cohort: { enabled: true, base: BASE, token: "tok-abc" } } };
27
- }
28
-
29
- /** A catalog with a providers Map: anthropic (never brokered) + deepseek + moonshot. */
30
- function catalogWith(providers) {
31
- const map = new Map();
32
- for (const [provider, auth_env] of providers) map.set(provider, { auth_env });
33
- return { providers: map, models: [] };
34
- }
35
-
36
- /**
37
- * A fake fetch that records calls and returns a lease frame per provider. The
38
- * `byProvider` map maps provider name → { value, ttlMs?, expiresAt? } | "error".
39
- */
40
- function fakeLeaseFetch(byProvider) {
41
- const calls = [];
42
- const impl = async (url, init) => {
43
- const body = init && init.body ? JSON.parse(init.body) : {};
44
- const provider = body.provider;
45
- calls.push({ url, provider, headers: (init && init.headers) || {} });
46
- const spec = byProvider[provider];
47
- if (!spec || spec === "error") {
48
- return {
49
- ok: false,
50
- status: 404,
51
- headers: { get: () => undefined },
52
- json: async () => ({ ok: false, error: { code: "NOT_FOUND", message: "no such credential" } }),
53
- };
54
- }
55
- const result = { provider, value: spec.value };
56
- if (spec.ttlMs != null) result.ttlMs = spec.ttlMs;
57
- if (spec.expiresAt != null) result.expiresAt = spec.expiresAt;
58
- return {
59
- ok: true,
60
- status: 200,
61
- headers: { get: () => undefined },
62
- json: async () => ({ ok: true, result }),
63
- };
64
- };
65
- impl.calls = calls;
66
- return impl;
67
- }
68
-
69
- // ---------------------------------------------------------------------------
70
-
71
- test("leases a provider absent from local env and returns the overlay", async () => {
72
- const f = fakeLeaseFetch({ deepseek: { value: "sk-deepseek-leased", ttlMs: 600000 } });
73
- const out = await resolveOrgCredentialEnv({
74
- catalog: catalogWith([["anthropic", "ANTHROPIC_API_KEY"], ["deepseek", "DEEPSEEK_API_KEY"]]),
75
- cfg: orgCfg(),
76
- env: {}, // no local keys at all
77
- fetchImpl: f,
78
- cache: new Map(),
79
- now: () => 1000,
80
- });
81
- assert.equal(out.envOverlay.DEEPSEEK_API_KEY, "sk-deepseek-leased");
82
- assert.deepEqual(out.leased, ["deepseek"]);
83
- assert.deepEqual(out.errors, []);
84
- // anthropic is never brokered; only deepseek was leased.
85
- assert.equal(f.calls.length, 1);
86
- assert.equal(f.calls[0].provider, "deepseek");
87
- // The lease POSTs to the credential.lease method over /v1 with Bearer + protocol.
88
- assert.match(f.calls[0].url, /\/v1\/credential\.lease$/);
89
- assert.equal(f.calls[0].headers.authorization, "Bearer tok-abc");
90
- assert.equal(f.calls[0].headers["x-org-protocol"], "1");
91
- });
92
-
93
- test("a provider already in local env is NOT leased (local wins)", async () => {
94
- const f = fakeLeaseFetch({ deepseek: { value: "sk-org" }, moonshot: { value: "sk-moon-org" } });
95
- const out = await resolveOrgCredentialEnv({
96
- catalog: catalogWith([["deepseek", "DEEPSEEK_API_KEY"], ["moonshot", "MOONSHOT_API_KEY"]]),
97
- cfg: orgCfg(),
98
- env: { DEEPSEEK_API_KEY: "sk-local-mine" }, // local deepseek key present
99
- fetchImpl: f,
100
- cache: new Map(),
101
- now: () => 1000,
102
- });
103
- // deepseek is local-served (never leased, never in the overlay); moonshot is leased.
104
- assert.equal(out.envOverlay.DEEPSEEK_API_KEY, undefined);
105
- assert.equal(out.envOverlay.MOONSHOT_API_KEY, "sk-moon-org");
106
- assert.deepEqual(out.leased, ["moonshot"]);
107
- assert.deepEqual(f.calls.map((c) => c.provider), ["moonshot"]);
108
- });
109
-
110
- test("an empty-string local key does NOT count as set — provider is leased", async () => {
111
- const f = fakeLeaseFetch({ deepseek: { value: "sk-org" } });
112
- const out = await resolveOrgCredentialEnv({
113
- catalog: catalogWith([["deepseek", "DEEPSEEK_API_KEY"]]),
114
- cfg: orgCfg(),
115
- env: { DEEPSEEK_API_KEY: " " }, // whitespace == unset
116
- fetchImpl: f,
117
- cache: new Map(),
118
- now: () => 1000,
119
- });
120
- assert.equal(out.envOverlay.DEEPSEEK_API_KEY, "sk-org");
121
- assert.deepEqual(out.leased, ["deepseek"]);
122
- });
123
-
124
- test("cache avoids a second lease within TTL", async () => {
125
- const f = fakeLeaseFetch({ deepseek: { value: "sk-cached", ttlMs: 600000 } });
126
- const cache = new Map();
127
- const catalog = catalogWith([["deepseek", "DEEPSEEK_API_KEY"]]);
128
-
129
- const first = await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: f, cache, now: () => 1000 });
130
- assert.equal(first.envOverlay.DEEPSEEK_API_KEY, "sk-cached");
131
- assert.equal(f.calls.length, 1);
132
-
133
- // Second resolve 5 minutes later — well within the 10-minute TTL → cache hit, no new lease.
134
- const second = await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: f, cache, now: () => 1000 + 5 * 60 * 1000 });
135
- assert.equal(second.envOverlay.DEEPSEEK_API_KEY, "sk-cached");
136
- assert.deepEqual(second.leased, ["deepseek"]);
137
- assert.equal(f.calls.length, 1, "no second lease within TTL");
138
- });
139
-
140
- test("re-leases after the cached lease TTL expires", async () => {
141
- let nth = 0;
142
- const f = fakeLeaseFetch({ deepseek: { value: "v", ttlMs: 600000 } });
143
- // wrap to give distinct values per call so we can prove a fresh lease landed
144
- const wrapped = async (url, init) => {
145
- nth += 1;
146
- const res = await f(url, init);
147
- const body = await res.json();
148
- body.result.value = `sk-rotated-${nth}`;
149
- return { ...res, json: async () => body };
150
- };
151
- const cache = new Map();
152
- const catalog = catalogWith([["deepseek", "DEEPSEEK_API_KEY"]]);
153
-
154
- const first = await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: wrapped, cache, now: () => 1000 });
155
- assert.equal(first.envOverlay.DEEPSEEK_API_KEY, "sk-rotated-1");
156
-
157
- // Past expiry (now far beyond TTL) → cache invalidated, fresh lease taken.
158
- const second = await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: wrapped, cache, now: () => 1000 + 999 * 60 * 1000 });
159
- assert.equal(second.envOverlay.DEEPSEEK_API_KEY, "sk-rotated-2");
160
- assert.equal(nth, 2, "expired cache forces a re-lease");
161
- });
162
-
163
- test("uses expiresAt when the server supplies it instead of ttlMs", async () => {
164
- const expiresAt = new Date(1000 + 60000).toISOString();
165
- const f = fakeLeaseFetch({ deepseek: { value: "sk-exp", expiresAt } });
166
- const cache = new Map();
167
- const catalog = catalogWith([["deepseek", "DEEPSEEK_API_KEY"]]);
168
-
169
- // Just before expiry → cache hit, no second lease.
170
- await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: f, cache, now: () => 1000 });
171
- await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: f, cache, now: () => 1000 + 30000 });
172
- assert.equal(f.calls.length, 1, "within expiresAt window → cached");
173
-
174
- // After expiry → re-lease.
175
- await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: f, cache, now: () => 1000 + 120000 });
176
- assert.equal(f.calls.length, 2, "past expiresAt → re-leased");
177
- });
178
-
179
- test("Cohort down (transport throw) fails open to local env", async () => {
180
- const thrower = async () => { throw new Error("ECONNREFUSED"); };
181
- const out = await resolveOrgCredentialEnv({
182
- catalog: catalogWith([["deepseek", "DEEPSEEK_API_KEY"]]),
183
- cfg: orgCfg(),
184
- env: { MOONSHOT_API_KEY: "sk-local-moon" }, // a local key the router will still see
185
- fetchImpl: thrower,
186
- cache: new Map(),
187
- now: () => 1000,
188
- });
189
- // No overlay added (deepseek couldn't be leased); the error is recorded, not thrown.
190
- assert.deepEqual(out.envOverlay, {});
191
- assert.deepEqual(out.leased, []);
192
- assert.equal(out.errors.length, 1);
193
- assert.equal(out.errors[0].provider, "deepseek");
194
- });
195
-
196
- test("org plane disabled → empty overlay, no lease attempts (fail-open)", async () => {
197
- const f = fakeLeaseFetch({ deepseek: { value: "x" } });
198
- const out = await resolveOrgCredentialEnv({
199
- catalog: catalogWith([["deepseek", "DEEPSEEK_API_KEY"]]),
200
- cfg: { org: { cohort: { enabled: false, base: BASE } } },
201
- env: {},
202
- fetchImpl: f,
203
- cache: new Map(),
204
- });
205
- assert.deepEqual(out.envOverlay, {});
206
- assert.deepEqual(out.leased, []);
207
- assert.deepEqual(out.errors, []);
208
- assert.equal(f.calls.length, 0, "no lease attempted when org plane is off");
209
- });
210
-
211
- test("a provider the org hasn't configured records an error and is omitted (router sees missing_credential)", async () => {
212
- // deepseek configured; moonshot not (lease → NOT_FOUND).
213
- const f = fakeLeaseFetch({ deepseek: { value: "sk-ds" } });
214
- const out = await resolveOrgCredentialEnv({
215
- catalog: catalogWith([["deepseek", "DEEPSEEK_API_KEY"], ["moonshot", "MOONSHOT_API_KEY"]]),
216
- cfg: orgCfg(),
217
- env: {},
218
- fetchImpl: f,
219
- cache: new Map(),
220
- now: () => 1000,
221
- });
222
- assert.equal(out.envOverlay.DEEPSEEK_API_KEY, "sk-ds");
223
- assert.equal(out.envOverlay.MOONSHOT_API_KEY, undefined, "unconfigured provider omitted → router's missing_credential");
224
- assert.deepEqual(out.leased, ["deepseek"]);
225
- assert.equal(out.errors.length, 1);
226
- assert.equal(out.errors[0].provider, "moonshot");
227
- assert.equal(out.errors[0].reason, "NOT_FOUND");
228
- });
229
-
230
- test("reads providers from catalog model rows when no providers Map is present", async () => {
231
- const f = fakeLeaseFetch({ deepseek: { value: "sk-rows" } });
232
- const catalog = {
233
- models: [
234
- { provider: "anthropic", auth_env: "ANTHROPIC_API_KEY", id: "claude" },
235
- { provider: "deepseek", auth_env: "DEEPSEEK_API_KEY", id: "deepseek-chat" },
236
- { provider: "deepseek", auth_env: "DEEPSEEK_API_KEY", id: "deepseek-reasoner" }, // dup → one lease
237
- ],
238
- };
239
- const out = await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: f, cache: new Map(), now: () => 1000 });
240
- assert.equal(out.envOverlay.DEEPSEEK_API_KEY, "sk-rows");
241
- assert.deepEqual(out.leased, ["deepseek"]);
242
- assert.equal(f.calls.length, 1, "duplicate rows collapse to one lease");
243
- });
244
-
245
- test("accepts a loadCatalog-style result (catalog nested under .catalog)", async () => {
246
- const f = fakeLeaseFetch({ deepseek: { value: "sk-nested" } });
247
- const inner = catalogWith([["deepseek", "DEEPSEEK_API_KEY"]]);
248
- const result = { catalog: inner, models: [], byRef: new Map() };
249
- const out = await resolveOrgCredentialEnv({ catalog: result, cfg: orgCfg(), env: {}, fetchImpl: f, cache: new Map(), now: () => 1000 });
250
- assert.equal(out.envOverlay.DEEPSEEK_API_KEY, "sk-nested");
251
- });
252
-
253
- test("clearOrgCredentialCache forces a re-lease against the module singleton", async () => {
254
- const f = fakeLeaseFetch({ deepseek: { value: "sk-singleton", ttlMs: 600000 } });
255
- const catalog = catalogWith([["deepseek", "DEEPSEEK_API_KEY"]]);
256
- clearOrgCredentialCache();
257
- // No injected cache → uses the module singleton.
258
- await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: f, now: () => 1000 });
259
- await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: f, now: () => 1000 });
260
- assert.equal(f.calls.length, 1, "singleton cache hit on the second resolve");
261
- clearOrgCredentialCache();
262
- await resolveOrgCredentialEnv({ catalog, cfg: orgCfg(), env: {}, fetchImpl: f, now: () => 1000 });
263
- assert.equal(f.calls.length, 2, "cache cleared → re-leased");
264
- clearOrgCredentialCache();
265
- });