@vellumai/assistant 0.12.3-staging.1 → 0.12.3-staging.2

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 (560) hide show
  1. package/ARCHITECTURE.md +43 -29
  2. package/Dockerfile +63 -0
  3. package/docs/architecture/security.md +10 -5
  4. package/docs/architecture/turn-actor.md +2 -0
  5. package/docs/desktop-browser-cli.md +45 -4
  6. package/docs/live-voice-task-outcomes.md +100 -0
  7. package/docs/mode-sessions.md +183 -0
  8. package/docs/voice-input-diagnostics.md +58 -0
  9. package/node_modules/@vellumai/ces-client/node_modules/@vellumai/service-contracts/package.json +2 -1
  10. package/node_modules/@vellumai/ces-client/node_modules/@vellumai/service-contracts/src/index.ts +1 -0
  11. package/node_modules/@vellumai/ces-client/node_modules/@vellumai/service-contracts/src/integration-categories.ts +36 -0
  12. package/node_modules/@vellumai/gateway-client/node_modules/@vellumai/service-contracts/package.json +2 -1
  13. package/node_modules/@vellumai/gateway-client/node_modules/@vellumai/service-contracts/src/index.ts +1 -0
  14. package/node_modules/@vellumai/gateway-client/node_modules/@vellumai/service-contracts/src/integration-categories.ts +36 -0
  15. package/node_modules/@vellumai/gateway-client/src/gateway-ipc-contracts.ts +20 -0
  16. package/node_modules/@vellumai/gateway-client/src/index.ts +4 -1
  17. package/node_modules/@vellumai/gateway-client/src/outbound-contract.ts +38 -7
  18. package/node_modules/@vellumai/gateway-client/src/trust-verdict-contract.ts +5 -6
  19. package/node_modules/@vellumai/service-contracts/package.json +2 -1
  20. package/node_modules/@vellumai/service-contracts/src/index.ts +1 -0
  21. package/node_modules/@vellumai/service-contracts/src/integration-categories.ts +36 -0
  22. package/node_modules/@vellumai/slack-text/src/index.ts +4 -8
  23. package/openapi.yaml +665 -14
  24. package/package.json +1 -1
  25. package/scripts/smoke-desktop-browser-cli.ts +10 -26
  26. package/scripts/smoke-desktop-window-theme.ts +2 -3
  27. package/src/__tests__/agent-loop-output-hooks.test.ts +150 -1
  28. package/src/__tests__/agent-loop-provider-error-recording.test.ts +87 -19
  29. package/src/__tests__/annotate-activity-metadata.test.ts +3 -1
  30. package/src/__tests__/annotate-risk-options.test.ts +2 -0
  31. package/src/__tests__/answered-question-persistence.test.ts +2 -0
  32. package/src/__tests__/anthropic-gateway-shared.test.ts +22 -0
  33. package/src/__tests__/approval-message-composer.test.ts +1 -118
  34. package/src/__tests__/approval-routes-http.test.ts +3 -0
  35. package/src/__tests__/assistant-event-hub-connection-history.test.ts +111 -0
  36. package/src/__tests__/assistant-event-hub-targeted.test.ts +50 -0
  37. package/src/__tests__/assistant-event-targeting.test.ts +186 -0
  38. package/src/__tests__/assistant-stream-state.test.ts +29 -9
  39. package/src/__tests__/browser-fill-credential.test.ts +11 -22
  40. package/src/__tests__/btw-routes.test.ts +4 -1
  41. package/src/__tests__/build-persisted-content.test.ts +1 -1
  42. package/src/__tests__/call-conversation-messages.test.ts +5 -0
  43. package/src/__tests__/call-recovery.test.ts +7 -2
  44. package/src/__tests__/call-routes-http.test.ts +12 -0
  45. package/src/__tests__/call-store.test.ts +7 -2
  46. package/src/__tests__/channel-retry-sweep.test.ts +9 -0
  47. package/src/__tests__/chat-reveal-guard-priming.test.ts +2 -0
  48. package/src/__tests__/checker.test.ts +11 -3
  49. package/src/__tests__/client-os-metadata-persistence.test.ts +35 -1
  50. package/src/__tests__/computer-use-skill-manifest-regression.test.ts +0 -25
  51. package/src/__tests__/computer-use-tools.test.ts +1 -233
  52. package/src/__tests__/config-model-image-provider.test.ts +39 -0
  53. package/src/__tests__/config-schema.test.ts +16 -0
  54. package/src/__tests__/config-watcher.test.ts +24 -0
  55. package/src/__tests__/context-search-agent-runner.test.ts +260 -1
  56. package/src/__tests__/context-search-fanout.test.ts +1 -1
  57. package/src/__tests__/context-search-types.test.ts +7 -7
  58. package/src/__tests__/conversation-agent-loop-disk-pressure.test.ts +8 -2
  59. package/src/__tests__/conversation-agent-loop-fatal-cleanup.test.ts +2 -0
  60. package/src/__tests__/conversation-agent-loop.test.ts +546 -16
  61. package/src/__tests__/conversation-app-control-lifecycle.test.ts +53 -1
  62. package/src/__tests__/conversation-evictor.test.ts +62 -37
  63. package/src/__tests__/conversation-history-boundary-repair-failure.test.ts +174 -0
  64. package/src/__tests__/conversation-loop-db-lock-resilience.test.ts +2 -0
  65. package/src/__tests__/conversation-process-app-control-preactivation.test.ts +2 -0
  66. package/src/__tests__/conversation-retry-route.test.ts +22 -0
  67. package/src/__tests__/conversation-routes-disk-view.test.ts +2 -0
  68. package/src/__tests__/conversation-routes-enabled-plugins.test.ts +2 -0
  69. package/src/__tests__/conversation-routes-guardian-reply.test.ts +10 -0
  70. package/src/__tests__/conversation-routes-hidden-queue.test.ts +2 -0
  71. package/src/__tests__/conversation-routes-slash-commands.test.ts +2 -0
  72. package/src/__tests__/conversation-runtime-assembly.test.ts +29 -0
  73. package/src/__tests__/conversation-surface-snapshots.test.ts +543 -0
  74. package/src/__tests__/conversation-surfaces-action-delivery.test.ts +151 -0
  75. package/src/__tests__/conversation-surfaces-point-at-budget.test.ts +63 -1
  76. package/src/__tests__/conversation-tool-setup-attribution.test.ts +23 -0
  77. package/src/__tests__/conversation-undo.test.ts +61 -0
  78. package/src/__tests__/credential-execution-client.test.ts +6 -6
  79. package/src/__tests__/credential-prompt-route.test.ts +35 -3
  80. package/src/__tests__/cu-unified-flow.test.ts +82 -1
  81. package/src/__tests__/db-mode-sessions-migration.test.ts +175 -0
  82. package/src/__tests__/deterministic-verification-control-plane.test.ts +2 -0
  83. package/src/__tests__/dm-backfill.test.ts +91 -1
  84. package/src/__tests__/document-create-route.test.ts +88 -0
  85. package/src/__tests__/document-sync-tags.test.ts +27 -0
  86. package/src/__tests__/emit-signal-routing-intent.test.ts +142 -2
  87. package/src/__tests__/events-tail-route.test.ts +25 -0
  88. package/src/__tests__/evict-conversations-for-reload.test.ts +57 -8
  89. package/src/__tests__/guardian-action-sweep.test.ts +16 -0
  90. package/src/__tests__/guardian-dispatch.test.ts +13 -0
  91. package/src/__tests__/headless-browser-interactions.test.ts +64 -20
  92. package/src/__tests__/heartbeat-service.test.ts +2 -59
  93. package/src/__tests__/helpers/mock-conversation.ts +21 -0
  94. package/src/__tests__/helpers/native-web-search-harness.ts +2 -0
  95. package/src/__tests__/helpers/verification-sessions-ipc-sim.ts +5 -0
  96. package/src/__tests__/host-cu-proxy.test.ts +46 -2
  97. package/src/__tests__/http-user-message-parity.test.ts +2 -0
  98. package/src/__tests__/image-credentials.test.ts +49 -0
  99. package/src/__tests__/image-service-dispatcher.test.ts +45 -0
  100. package/src/__tests__/inference-profile-session-handler.test.ts +1 -1
  101. package/src/__tests__/interrupt-on-send.test.ts +57 -43
  102. package/src/__tests__/interrupt-turn-note.test.ts +5 -6
  103. package/src/__tests__/llm-catalog-parity.test.ts +2 -2
  104. package/src/__tests__/media-generate-image.test.ts +49 -1
  105. package/src/__tests__/message-provenance-contact-id.test.ts +69 -0
  106. package/src/__tests__/migration-export-to-gcs.test.ts +142 -0
  107. package/src/__tests__/mode-session-history.test.ts +245 -0
  108. package/src/__tests__/mode-session-recovery.test.ts +222 -0
  109. package/src/__tests__/mode-session-wire.test.ts +141 -0
  110. package/src/__tests__/notification-guardian-path.test.ts +7 -0
  111. package/src/__tests__/notification-telegram-adapter.test.ts +2 -2
  112. package/src/__tests__/notification-vellum-adapter.test.ts +34 -0
  113. package/src/__tests__/oauth-provider-serializer.test.ts +12 -0
  114. package/src/__tests__/oauth-providers-routes.test.ts +1 -0
  115. package/src/__tests__/outbound-slack-persistence.test.ts +2 -0
  116. package/src/__tests__/persistence-secret-redaction.test.ts +2 -0
  117. package/src/__tests__/plugin-api-provider.test.ts +5 -0
  118. package/src/__tests__/plugin-api-resolve-credential.test.ts +97 -0
  119. package/src/__tests__/plugin-import-boundary-guard.test.ts +9 -0
  120. package/src/__tests__/process-message-display-content.test.ts +56 -0
  121. package/src/__tests__/reaction-persistence.test.ts +5 -0
  122. package/src/__tests__/request-secret-standalone-channel-gate.test.ts +25 -9
  123. package/src/__tests__/runtime-events-sse-reconnect.test.ts +1 -1
  124. package/src/__tests__/scripted-turn-metadata-persistence.test.ts +73 -0
  125. package/src/__tests__/secret-ingress-http.test.ts +2 -0
  126. package/src/__tests__/secure-keys.test.ts +43 -1
  127. package/src/__tests__/send-endpoint-busy.test.ts +163 -1
  128. package/src/__tests__/server-history-render.test.ts +91 -1
  129. package/src/__tests__/settings-routes.test.ts +50 -1
  130. package/src/__tests__/skill-script-runner-sandbox.test.ts +32 -0
  131. package/src/__tests__/subagent-fork-notifications.test.ts +1 -2
  132. package/src/__tests__/subagent-manager-notify.test.ts +1 -2
  133. package/src/__tests__/subagent-notify-parent.test.ts +146 -39
  134. package/src/__tests__/subagent-notify-stale-rebuild.test.ts +2 -4
  135. package/src/__tests__/surface-completion-in-flight-snapshot.test.ts +2 -0
  136. package/src/__tests__/thread-backfill.test.ts +53 -0
  137. package/src/__tests__/tool-preview-lifecycle.test.ts +2 -0
  138. package/src/__tests__/tool-result-metadata-plumbing.test.ts +3 -1
  139. package/src/__tests__/tool-start-timestamp.test.ts +2 -0
  140. package/src/__tests__/twilio-routes.test.ts +1 -0
  141. package/src/__tests__/web-search-backend-failure.test.ts +1 -1
  142. package/src/activation/progress-store.test.ts +12 -7
  143. package/src/agent/loop.ts +99 -41
  144. package/src/api/events/assistant-turn-start.ts +3 -0
  145. package/src/api/events/generation-handoff.ts +2 -0
  146. package/src/api/events/message-complete.ts +2 -0
  147. package/src/api/events/notification-conversation-created.ts +14 -16
  148. package/src/api/events/notification-intent.ts +4 -2
  149. package/src/api/events/tool-result.ts +85 -0
  150. package/src/api/events/tool-use-start.ts +3 -0
  151. package/src/api/events/user-message-echo.ts +3 -0
  152. package/src/api/index.ts +31 -6
  153. package/src/api/mode-session.test.ts +131 -0
  154. package/src/api/mode-session.ts +148 -0
  155. package/src/api/remember-tool.ts +15 -0
  156. package/src/api/responses/conversation-message.ts +6 -0
  157. package/src/approvals/AGENTS.md +4 -2
  158. package/src/browser/__tests__/operations.test.ts +29 -0
  159. package/src/browser/operation-meta.ts +2 -1
  160. package/src/browser/operations.ts +29 -0
  161. package/src/browser/virtual-desktop-target.ts +2 -8
  162. package/src/bundler/bundle-signer.ts +1 -3
  163. package/src/calls/__tests__/barge-in-guard.test.ts +3 -0
  164. package/src/calls/__tests__/call-controller.test.ts +94 -0
  165. package/src/calls/__tests__/call-funnel-telemetry.test.ts +204 -0
  166. package/src/calls/__tests__/voice-control-protocol.test.ts +15 -0
  167. package/src/calls/__tests__/voice-escalation-judge.test.ts +209 -0
  168. package/src/calls/__tests__/voice-leg-coordinator.test.ts +168 -0
  169. package/src/{live-voice/__tests__/live-voice-metrics.test.ts → calls/__tests__/voice-metrics.test.ts} +32 -38
  170. package/src/calls/__tests__/voice-session-bridge.test.ts +576 -122
  171. package/src/calls/barge-in-guard.ts +5 -0
  172. package/src/calls/call-controller.ts +196 -1
  173. package/src/calls/call-domain.ts +4 -0
  174. package/src/calls/call-store.ts +87 -2
  175. package/src/calls/guardian-action-sweep.ts +22 -8
  176. package/src/calls/media-stream-audio-transcode.ts +1 -3
  177. package/src/calls/typesafe-noul.ts +156 -0
  178. package/src/calls/voice-control-protocol.ts +6 -0
  179. package/src/calls/voice-escalation-judge.ts +145 -0
  180. package/src/calls/voice-escalation-target.ts +10 -0
  181. package/src/calls/voice-leg-coordinator.ts +121 -3
  182. package/src/{live-voice/live-voice-metrics.ts → calls/voice-metrics.ts} +94 -87
  183. package/src/calls/voice-session-bridge.ts +256 -112
  184. package/src/calls/voice-triage-escalate.ts +7 -0
  185. package/src/cli/__tests__/roadmap-cli.test.ts +4 -41
  186. package/src/cli/commands/__tests__/notifications.test.ts +35 -10
  187. package/src/cli/commands/browser.help.ts +4 -5
  188. package/src/cli/commands/browser.ts +0 -1
  189. package/src/cli/commands/clients.help.ts +46 -0
  190. package/src/cli/commands/clients.ts +116 -0
  191. package/src/cli/commands/credentials.help.ts +6 -1
  192. package/src/cli/commands/notifications.help.ts +16 -2
  193. package/src/cli/commands/notifications.ts +30 -8
  194. package/src/cli/commands/roadmap.help.ts +18 -21
  195. package/src/cli/commands/roadmap.ts +4 -21
  196. package/src/cli/lib/__tests__/list-installed-plugins.test.ts +5 -2
  197. package/src/cli/lib/bundled-marketplace.json +84 -34
  198. package/src/cli/lib/plugin-marketplace.ts +8 -0
  199. package/src/cli/lib/publish-plugin.ts +1 -2
  200. package/src/config/__tests__/default-provider.test.ts +3 -1
  201. package/src/config/__tests__/memory-retrospective-schema.test.ts +14 -1
  202. package/src/config/bundled-skills/computer-use/SKILL.md +1 -1
  203. package/src/config/bundled-skills/computer-use/TOOLS.json +89 -1
  204. package/src/config/bundled-skills/image-studio/SKILL.md +17 -2
  205. package/src/config/bundled-skills/image-studio/tools/media-generate-image.ts +7 -11
  206. package/src/config/bundled-skills/sequences/TOOLS.json +1 -5
  207. package/src/config/call-site-defaults.ts +9 -0
  208. package/src/config/feature-flag-registry.json +16 -0
  209. package/src/config/profile-text-generation.test.ts +2 -2
  210. package/src/config/schemas/__tests__/tools.test.ts +31 -0
  211. package/src/config/schemas/call-site-catalog.ts +14 -0
  212. package/src/config/schemas/channels.ts +0 -18
  213. package/src/config/schemas/llm.ts +3 -1
  214. package/src/config/schemas/mcp.ts +2 -1
  215. package/src/config/schemas/memory-retrospective.ts +11 -2
  216. package/src/config/schemas/services.ts +15 -6
  217. package/src/config/schemas/tools.ts +10 -0
  218. package/src/config/session-groups-gate.ts +21 -0
  219. package/src/config/skills.ts +2 -6
  220. package/src/conversations/__tests__/message-consolidation.test.ts +149 -0
  221. package/src/conversations/message-consolidation.ts +102 -5
  222. package/src/credential-execution/ces-connect.test.ts +69 -0
  223. package/src/credential-execution/ces-connect.ts +115 -0
  224. package/src/credential-execution/ces-runtime.ts +55 -93
  225. package/src/credential-execution/executable-discovery.ts +1 -11
  226. package/src/daemon/__tests__/conversation-surfaces-launch.test.ts +10 -1
  227. package/src/daemon/__tests__/inference-profile-notification.test.ts +2 -0
  228. package/src/daemon/browser-mode-session.test.ts +402 -0
  229. package/src/daemon/browser-mode-session.ts +186 -0
  230. package/src/daemon/computer-use-mode-session.test.ts +210 -0
  231. package/src/daemon/computer-use-mode-session.ts +105 -0
  232. package/src/daemon/config-watcher.ts +10 -0
  233. package/src/daemon/conversation-agent-loop-handlers.ts +58 -10
  234. package/src/daemon/conversation-agent-loop.ts +219 -25
  235. package/src/daemon/conversation-evictor.ts +5 -24
  236. package/src/daemon/conversation-history.ts +10 -0
  237. package/src/daemon/conversation-interrupt-repair.ts +4 -43
  238. package/src/daemon/conversation-interrupt.ts +7 -11
  239. package/src/daemon/conversation-messaging.ts +58 -5
  240. package/src/daemon/conversation-mode-session.test.ts +1667 -0
  241. package/src/daemon/conversation-mode-session.ts +1008 -0
  242. package/src/daemon/conversation-process.ts +5 -0
  243. package/src/daemon/conversation-runtime-assembly.ts +10 -20
  244. package/src/daemon/conversation-skill-tools.ts +1 -0
  245. package/src/daemon/conversation-store.ts +4 -4
  246. package/src/daemon/conversation-surface-snapshots.ts +135 -0
  247. package/src/daemon/conversation-surfaces.ts +113 -5
  248. package/src/daemon/conversation-tool-setup.ts +8 -3
  249. package/src/daemon/conversation.ts +143 -41
  250. package/src/daemon/handlers/config-model.ts +7 -5
  251. package/src/daemon/handlers/shared.ts +55 -16
  252. package/src/daemon/host-cu-proxy.ts +51 -22
  253. package/src/daemon/lifecycle.ts +30 -13
  254. package/src/daemon/message-protocol.ts +0 -1
  255. package/src/daemon/message-provenance.ts +28 -4
  256. package/src/daemon/mode-session-startup-recovery.ts +23 -0
  257. package/src/daemon/mode-session-tracking.ts +54 -0
  258. package/src/daemon/process-message.ts +13 -4
  259. package/src/daemon/virtual-desktop-context.ts +21 -0
  260. package/src/desktop/__tests__/fake-desktop.ts +1 -0
  261. package/src/desktop/desktop-accessibility-script.ts +88 -0
  262. package/src/desktop/desktop-accessibility.test.ts +160 -0
  263. package/src/desktop/desktop-accessibility.ts +169 -0
  264. package/src/desktop/desktop-automation-lease.test.ts +76 -68
  265. package/src/desktop/desktop-automation-lease.ts +41 -50
  266. package/src/desktop/desktop-browser-endpoint.ts +4 -0
  267. package/src/desktop/desktop-computer-use-routing.test.ts +330 -0
  268. package/src/desktop/desktop-computer-use.test.ts +449 -0
  269. package/src/desktop/desktop-computer-use.ts +502 -0
  270. package/src/desktop/desktop-dependencies.ts +47 -293
  271. package/src/desktop/desktop-display.ts +2 -2
  272. package/src/desktop/desktop-panel-config.test.ts +13 -3
  273. package/src/desktop/desktop-panel-config.ts +109 -15
  274. package/src/desktop/desktop-session-manager.test.ts +69 -21
  275. package/src/desktop/desktop-session-manager.ts +76 -22
  276. package/src/desktop/desktop-stream-bridge.test.ts +11 -122
  277. package/src/desktop/desktop-stream-bridge.ts +6 -8
  278. package/src/desktop/desktop-window-theme.ts +5 -25
  279. package/src/desktop/virtual-desktop-feature.ts +52 -2
  280. package/src/desktop/virtual-desktop-platform.test.ts +5 -9
  281. package/src/documents/document-store.ts +23 -0
  282. package/src/heartbeat/__tests__/heartbeat-service.test.ts +22 -26
  283. package/src/heartbeat/heartbeat-service.ts +5 -33
  284. package/src/ipc/__tests__/browser-ipc.test.ts +215 -6
  285. package/src/ipc/assistant-server.ts +2 -0
  286. package/src/ipc/routes/__tests__/channel-reply-ipc-routes.test.ts +111 -0
  287. package/src/ipc/routes/channel-reply-ipc-routes.ts +73 -0
  288. package/src/live-voice/__tests__/continuation-judge.test.ts +151 -0
  289. package/src/live-voice/__tests__/live-activity-reporter.test.ts +154 -15
  290. package/src/live-voice/__tests__/live-voice-flux-turn-end.test.ts +743 -31
  291. package/src/live-voice/__tests__/live-voice-look-follow-up.test.ts +191 -10
  292. package/src/live-voice/__tests__/live-voice-session-groups-gate.test.ts +87 -0
  293. package/src/live-voice/__tests__/live-voice-session-manager.test.ts +5 -1
  294. package/src/live-voice/__tests__/live-voice-sight-frame-echo.test.ts +41 -0
  295. package/src/live-voice/__tests__/live-voice-sight-frame.test.ts +1017 -9
  296. package/src/live-voice/__tests__/live-voice-triage-escalate.test.ts +224 -1
  297. package/src/live-voice/__tests__/live-voice-tts-session.test.ts +101 -0
  298. package/src/live-voice/__tests__/live-voice-tts.test.ts +44 -0
  299. package/src/live-voice/__tests__/live-voice-vad.test.ts +939 -75
  300. package/src/live-voice/__tests__/microphone-idle-gate.test.ts +87 -0
  301. package/src/live-voice/__tests__/session-controls.test.ts +23 -0
  302. package/src/live-voice/__tests__/task-outcomes.test.ts +117 -0
  303. package/src/live-voice/__tests__/voice-input-diagnostics.test.ts +117 -0
  304. package/src/live-voice/camera-mode-session.test.ts +265 -0
  305. package/src/live-voice/camera-mode-session.ts +277 -0
  306. package/src/live-voice/continuation-judge.ts +123 -0
  307. package/src/live-voice/live-activity-reporter.ts +115 -6
  308. package/src/live-voice/live-voice-manager.ts +36 -1
  309. package/src/live-voice/live-voice-photo.ts +88 -7
  310. package/src/live-voice/live-voice-session-manager.ts +15 -0
  311. package/src/live-voice/live-voice-session.ts +1443 -747
  312. package/src/live-voice/live-voice-tts.ts +22 -4
  313. package/src/live-voice/microphone-idle-gate.ts +56 -0
  314. package/src/live-voice/protocol.ts +171 -4
  315. package/src/live-voice/session-controls.ts +11 -3
  316. package/src/live-voice/task-outcomes.ts +131 -0
  317. package/src/live-voice/voice-input-diagnostics.ts +117 -0
  318. package/src/mcp/__tests__/manager-tool-caps.test.ts +63 -18
  319. package/src/mcp/__tests__/tool-caps.test.ts +28 -1
  320. package/src/mcp/manager.ts +3 -0
  321. package/src/mcp/tool-caps.ts +12 -4
  322. package/src/media/__tests__/image-models.test.ts +58 -0
  323. package/src/media/image-credentials.ts +5 -0
  324. package/src/media/image-models.ts +52 -0
  325. package/src/media/image-service.ts +8 -0
  326. package/src/media/openrouter-image-service.test.ts +133 -0
  327. package/src/media/openrouter-image-service.ts +153 -0
  328. package/src/media/types.ts +1 -1
  329. package/src/messaging/providers/__tests__/transport-dispatch.test.ts +65 -11
  330. package/src/messaging/providers/discord/send.ts +4 -17
  331. package/src/messaging/providers/discord/transport.ts +2 -2
  332. package/src/messaging/providers/outlook/client.ts +16 -22
  333. package/src/messaging/providers/telegram-bot/send.test.ts +236 -1
  334. package/src/messaging/providers/telegram-bot/send.ts +10 -7
  335. package/src/messaging/providers/telegram-bot/transport.ts +3 -10
  336. package/src/messaging/providers/whatsapp/send.test.ts +169 -4
  337. package/src/monitoring/__tests__/control.test.ts +17 -5
  338. package/src/monitoring/control.ts +15 -5
  339. package/src/notifications/AGENTS.md +1 -1
  340. package/src/notifications/README.md +83 -99
  341. package/src/notifications/__tests__/broadcaster.test.ts +1 -1
  342. package/src/notifications/__tests__/channel-allowlist.test.ts +39 -0
  343. package/src/notifications/__tests__/copy-composer.test.ts +1 -1
  344. package/src/notifications/__tests__/decision-engine.test.ts +79 -2
  345. package/src/notifications/__tests__/deterministic-checks.test.ts +10 -10
  346. package/src/notifications/__tests__/home-feed-side-effect.test.ts +93 -4
  347. package/src/notifications/__tests__/notification-utils.test.ts +18 -0
  348. package/src/notifications/adapters/macos.ts +37 -46
  349. package/src/notifications/broadcaster.ts +5 -6
  350. package/src/notifications/channel-allowlist.ts +44 -0
  351. package/src/notifications/copy-composer.ts +0 -23
  352. package/src/notifications/decision-engine.ts +24 -14
  353. package/src/notifications/deterministic-checks.ts +1 -1
  354. package/src/notifications/emit-signal.ts +123 -37
  355. package/src/notifications/home-feed-side-effect.ts +9 -7
  356. package/src/notifications/notification-utils.ts +19 -0
  357. package/src/notifications/signal.ts +4 -19
  358. package/src/notifications/types.ts +15 -0
  359. package/src/oauth/connection-resolver.ts +1 -2
  360. package/src/oauth/oauth-store.ts +1 -3
  361. package/src/oauth/provider-serializer.ts +7 -0
  362. package/src/oauth/seed-providers.ts +30 -0
  363. package/src/onboarding/onboarding-events-store.ts +99 -12
  364. package/src/persistence/__tests__/client-connection-events-store.test.ts +322 -0
  365. package/src/persistence/client-connection-events-store.ts +377 -0
  366. package/src/persistence/conversation-crud.ts +55 -6
  367. package/src/persistence/conversation-mode-sessions.test.ts +424 -0
  368. package/src/persistence/conversation-mode-sessions.ts +460 -0
  369. package/src/persistence/conversation-queries.ts +6 -19
  370. package/src/persistence/llm-usage-store.ts +3 -9
  371. package/src/persistence/message-metadata.ts +25 -0
  372. package/src/persistence/migrations/381-create-conversation-mode-sessions.ts +41 -0
  373. package/src/persistence/migrations/382-create-client-connection-events.ts +40 -0
  374. package/src/persistence/migrations/__tests__/382-create-client-connection-events.test.ts +56 -0
  375. package/src/persistence/schema/conversations.ts +67 -0
  376. package/src/persistence/schema/infrastructure.ts +29 -0
  377. package/src/persistence/steps.ts +17 -0
  378. package/src/plugin-api/__tests__/plugin-name-env.test.ts +249 -0
  379. package/src/plugin-api/effective-context-window.ts +33 -0
  380. package/src/plugin-api/index.ts +24 -5
  381. package/src/plugin-api/plugin-name-env.ts +125 -0
  382. package/src/plugin-api/resolve-credential.ts +79 -20
  383. package/src/plugin-api/vision-support.test.ts +22 -0
  384. package/src/plugin-api/vision-support.ts +17 -5
  385. package/src/plugins/__tests__/active-plugin-env.test.ts +136 -0
  386. package/src/plugins/active-plugin-env.ts +51 -0
  387. package/src/plugins/defaults/index.ts +24 -0
  388. package/src/plugins/defaults/memory/__tests__/archive-removal-framing.test.ts +64 -0
  389. package/src/plugins/defaults/memory/__tests__/worker-control.test.ts +19 -6
  390. package/src/plugins/defaults/memory/context-search/agent-protocol.ts +9 -12
  391. package/src/plugins/defaults/memory/context-search/agent-runner.ts +78 -22
  392. package/src/plugins/defaults/memory/context-search/format.ts +51 -7
  393. package/src/plugins/defaults/memory/context-search/limits.ts +10 -12
  394. package/src/plugins/defaults/memory/context-search/recall-input.ts +38 -0
  395. package/src/plugins/defaults/memory/context-search/search.ts +8 -14
  396. package/src/plugins/defaults/memory/context-search/sources/conversations.ts +1 -0
  397. package/src/plugins/defaults/memory/context-search/types.ts +1 -11
  398. package/src/plugins/defaults/memory/graph/__tests__/handle-remember-v2.test.ts +11 -4
  399. package/src/plugins/defaults/memory/graph/tool-handlers.ts +12 -21
  400. package/src/plugins/defaults/memory/graph/tools.ts +35 -64
  401. package/src/plugins/defaults/memory/src/memory-eval-routes.ts +2 -3
  402. package/src/plugins/defaults/memory/src/memory-item-routes.test.ts +5 -0
  403. package/src/plugins/defaults/memory/src/memory-item-routes.ts +2 -1
  404. package/src/plugins/defaults/memory/substrate/prompts/consolidation.ts +2 -2
  405. package/src/plugins/defaults/memory/tools.test.ts +124 -9
  406. package/src/plugins/defaults/memory/tools.ts +38 -11
  407. package/src/plugins/defaults/memory/v3/__tests__/pool-log-store.test.ts +165 -27
  408. package/src/plugins/defaults/memory/v3/__tests__/pool-select.test.ts +626 -27
  409. package/src/plugins/defaults/memory/v3/__tests__/shadow-plugin.test.ts +46 -27
  410. package/src/plugins/defaults/memory/v3/orchestrate.ts +78 -38
  411. package/src/plugins/defaults/memory/v3/pool-log-store.ts +65 -79
  412. package/src/plugins/defaults/memory/v3/pool-select.test.ts +14 -5
  413. package/src/plugins/defaults/memory/v3/pool-select.ts +683 -44
  414. package/src/plugins/defaults/memory/worker-control.ts +15 -5
  415. package/src/plugins/defaults/task-progress/__tests__/registration.test.ts +70 -0
  416. package/src/plugins/defaults/task-progress/__tests__/task-progress-context.test.ts +380 -0
  417. package/src/plugins/defaults/task-progress/hooks/post-compact.ts +19 -0
  418. package/src/plugins/defaults/task-progress/hooks/user-prompt-submit.ts +24 -0
  419. package/src/plugins/defaults/task-progress/package.json +14 -0
  420. package/src/plugins/defaults/task-progress/src/apply-task-progress-context.ts +93 -0
  421. package/src/plugins/defaults/task-progress/src/format-task-progress-context.ts +67 -0
  422. package/src/providers/__tests__/context-overflow-error.test.ts +10 -0
  423. package/src/providers/__tests__/search-provider-catalog.test.ts +18 -0
  424. package/src/providers/anthropic/client.ts +10 -1
  425. package/src/providers/anthropic-gateway-shared.ts +6 -0
  426. package/src/providers/fetch-provider-catalog.ts +11 -0
  427. package/src/providers/gemini/client.ts +25 -8
  428. package/src/providers/inference/adapter-factory.ts +1 -1
  429. package/src/providers/jev/client.test.ts +10 -0
  430. package/src/providers/jev/client.ts +20 -4
  431. package/src/providers/model-catalog.ts +4 -3
  432. package/src/providers/openai/__tests__/chat-completions-extra-body.test.ts +94 -1
  433. package/src/providers/openai/__tests__/chat-completions-provider-reasoning.test.ts +106 -1
  434. package/src/providers/openai/__tests__/chat-template-flatten.test.ts +24 -0
  435. package/src/providers/openai/__tests__/google-thought-signature.test.ts +1 -2
  436. package/src/providers/openai/chat-completions-provider.ts +100 -29
  437. package/src/providers/openai/google-thought-signature.ts +0 -6
  438. package/src/providers/openai/responses-provider.ts +23 -1
  439. package/src/providers/ratelimit.ts +1 -3
  440. package/src/providers/search-provider-catalog.ts +24 -0
  441. package/src/providers/speech-to-text/__tests__/deepgram-flux-frames.test.ts +24 -0
  442. package/src/providers/speech-to-text/__tests__/deepgram-flux-realtime.test.ts +7 -1
  443. package/src/providers/speech-to-text/__tests__/vellum-managed-realtime.test.ts +50 -1
  444. package/src/providers/speech-to-text/deepgram-flux-frames.ts +11 -0
  445. package/src/providers/speech-to-text/vellum-managed-flux-realtime.ts +13 -0
  446. package/src/providers/types.ts +5 -0
  447. package/src/routes/control.ts +7 -5
  448. package/src/runtime/AGENTS.md +4 -2
  449. package/src/runtime/actor-trust-resolver.ts +3 -3
  450. package/src/runtime/agent-wake.ts +1 -1
  451. package/src/runtime/approval-message-composer.ts +3 -69
  452. package/src/runtime/assistant-event-hub.ts +73 -75
  453. package/src/runtime/assistant-event-publish-options.ts +7 -0
  454. package/src/runtime/assistant-event-targeting.ts +79 -0
  455. package/src/runtime/assistant-stream-state.ts +13 -100
  456. package/src/runtime/background-job-runner.ts +5 -5
  457. package/src/runtime/channel-retry-sweep.ts +1 -0
  458. package/src/runtime/guardian-action-message-composer.ts +8 -161
  459. package/src/runtime/guardian-reply-router.ts +28 -32
  460. package/src/runtime/http-types.ts +6 -0
  461. package/src/runtime/message-composer-types.ts +3 -61
  462. package/src/runtime/migrations/__tests__/vbundle-import-parity.test.ts +67 -1
  463. package/src/runtime/migrations/vbundle-builder.ts +24 -1
  464. package/src/runtime/migrations/vbundle-import-analyzer.ts +13 -4
  465. package/src/runtime/migrations/vbundle-import-policy.ts +10 -0
  466. package/src/runtime/migrations/vbundle-importer.ts +4 -1
  467. package/src/runtime/migrations/vbundle-streaming-importer.ts +4 -1
  468. package/src/runtime/migrations/vbundle-validator.ts +1 -0
  469. package/src/runtime/pending-interactions-mode-session.test.ts +62 -0
  470. package/src/runtime/pending-interactions.ts +29 -0
  471. package/src/runtime/routes/__tests__/client-history-routes.test.ts +104 -0
  472. package/src/runtime/routes/__tests__/image-generation-routes.test.ts +97 -0
  473. package/src/runtime/routes/__tests__/inference-profiles-routes.test.ts +1 -1
  474. package/src/runtime/routes/__tests__/roadmap-routes.test.ts +29 -6
  475. package/src/runtime/routes/browser-context.ts +31 -11
  476. package/src/runtime/routes/browser-routes.ts +43 -4
  477. package/src/runtime/routes/btw-routes.ts +9 -4
  478. package/src/runtime/routes/client-routes.ts +108 -0
  479. package/src/runtime/routes/conversation-management-routes.ts +4 -1
  480. package/src/runtime/routes/conversation-routes.ts +113 -27
  481. package/src/runtime/routes/credential-prompt-routes.ts +4 -3
  482. package/src/runtime/routes/credential-routes.ts +1 -2
  483. package/src/runtime/routes/desktop-setup-routes.test.ts +9 -18
  484. package/src/runtime/routes/desktop-setup-routes.ts +4 -6
  485. package/src/runtime/routes/documents-routes.ts +51 -0
  486. package/src/runtime/routes/email-routes.ts +3 -7
  487. package/src/runtime/routes/events-routes.ts +75 -54
  488. package/src/runtime/routes/identity-routes.ts +1 -5
  489. package/src/runtime/routes/image-generation-routes.ts +8 -14
  490. package/src/runtime/routes/inbound-message-handler.ts +35 -38
  491. package/src/runtime/routes/inbound-stages/backfill-sender-provenance.test.ts +168 -0
  492. package/src/runtime/routes/inbound-stages/backfill-sender-provenance.ts +96 -0
  493. package/src/runtime/routes/inbound-stages/background-dispatch.test.ts +38 -9
  494. package/src/runtime/routes/inbound-stages/background-dispatch.ts +2 -5
  495. package/src/runtime/routes/inbound-stages/reaction-intercept.ts +2 -0
  496. package/src/runtime/routes/llm-context-normalization.ts +1 -7
  497. package/src/runtime/routes/migration-routes.ts +73 -2
  498. package/src/runtime/routes/notification-routes.ts +10 -0
  499. package/src/runtime/routes/oauth-providers.ts +7 -0
  500. package/src/runtime/routes/plugins-routes.ts +6 -0
  501. package/src/runtime/routes/roadmap-routes.ts +47 -8
  502. package/src/runtime/routes/secret-routes.ts +1 -1
  503. package/src/runtime/routes/settings-routes.ts +10 -4
  504. package/src/runtime/routes/watcher-routes.ts +1 -3
  505. package/src/runtime/trust-class.test.ts +7 -6
  506. package/src/runtime/trust-class.ts +9 -17
  507. package/src/schedule/__tests__/worker-control.test.ts +36 -5
  508. package/src/schedule/worker-control.ts +15 -5
  509. package/src/security/secure-keys.ts +53 -103
  510. package/src/skills/clawhub.ts +3 -8
  511. package/src/skills/managed-store.ts +1 -5
  512. package/src/stt/types.ts +4 -0
  513. package/src/subagent/notify.ts +54 -34
  514. package/src/subagent/parent-notification.ts +6 -0
  515. package/src/telemetry/__tests__/phone-call-funnel.test.ts +72 -0
  516. package/src/telemetry/phone-call-funnel.ts +144 -0
  517. package/src/tools/__tests__/tool-schema-root-combinator-guard.test.ts +81 -0
  518. package/src/tools/ask-question/ask-question-tool.ts +6 -4
  519. package/src/tools/browser/browser-execution.ts +22 -59
  520. package/src/tools/browser/cdp-client/__tests__/cdp-dom-helpers.test.ts +561 -39
  521. package/src/tools/browser/cdp-client/cdp-dom-helpers.ts +435 -39
  522. package/src/tools/browser/cdp-client/cdp-inspect-client.ts +1 -8
  523. package/src/tools/client-os.ts +15 -0
  524. package/src/tools/computer-use/skill-proxy-bridge.ts +1 -1
  525. package/src/tools/computer-use/target.ts +42 -0
  526. package/src/tools/document/document-tool.ts +13 -18
  527. package/src/tools/execution-target.ts +14 -16
  528. package/src/tools/host-terminal/host-shell.ts +4 -1
  529. package/src/tools/network/__tests__/web-fetch-tinyfish.test.ts +250 -0
  530. package/src/tools/network/__tests__/web-search.test.ts +203 -0
  531. package/src/tools/network/web-fetch.ts +274 -7
  532. package/src/tools/network/web-search.ts +468 -6
  533. package/src/tools/permission-checker.ts +3 -3
  534. package/src/tools/policy-context.ts +3 -1
  535. package/src/tools/skills/load.ts +2 -1
  536. package/src/tools/skills/sandbox-runner.ts +11 -2
  537. package/src/tools/skills/skill-script-runner.ts +7 -0
  538. package/src/tools/skills/skill-tool-factory.ts +15 -1
  539. package/src/tools/terminal/__tests__/safe-env.test.ts +11 -0
  540. package/src/tools/terminal/safe-env.ts +4 -0
  541. package/src/tools/terminal/shell.ts +2 -0
  542. package/src/tools/tool-approval-handler.ts +5 -1
  543. package/src/tools/tool-types.ts +3 -3
  544. package/src/tools/types.ts +8 -2
  545. package/src/util/__tests__/worker-process-command.test.ts +5 -4
  546. package/src/util/abort-reasons.ts +18 -26
  547. package/src/util/canonicalize-identity.test.ts +21 -0
  548. package/src/util/canonicalize-identity.ts +16 -0
  549. package/src/util/errors.ts +13 -3
  550. package/src/util/provider-error-patterns.ts +3 -0
  551. package/src/util/worker-process.ts +77 -35
  552. package/src/workspace/migrations/157-add-desktop-files-launcher.ts +94 -0
  553. package/src/workspace/migrations/__tests__/156-extract-workspace-mcp-json.test.ts +4 -10
  554. package/src/workspace/migrations/__tests__/157-add-desktop-files-launcher.test.ts +113 -0
  555. package/src/workspace/migrations/registry.ts +2 -0
  556. package/src/__tests__/computer-use-window-schema.test.ts +0 -24
  557. package/src/api/events/heartbeat-alert.ts +0 -20
  558. package/src/daemon/message-types/web-activity.ts +0 -65
  559. package/src/desktop/desktop-dependencies.test.ts +0 -131
  560. package/src/tools/computer-use/definitions.ts +0 -607
@@ -0,0 +1,449 @@
1
+ import { describe, expect, mock, spyOn, test } from "bun:test";
2
+
3
+ import { run as click } from "../config/bundled-skills/computer-use/tools/computer-use-click.js";
4
+ import * as configLoader from "../config/loader.js";
5
+ import { HostCuProxy } from "../daemon/host-cu-proxy.js";
6
+ import {
7
+ DesktopAutomationLease,
8
+ desktopAutomationLease,
9
+ } from "./desktop-automation-lease.js";
10
+ import {
11
+ executeDesktopComputerUse,
12
+ performDesktopComputerUse,
13
+ planDesktopComputerUse,
14
+ } from "./desktop-computer-use.js";
15
+ import * as sessionManager from "./desktop-session-manager.js";
16
+
17
+ function driver(userGuidance?: string) {
18
+ return {
19
+ input: mock(async (_args: string[], _signal?: AbortSignal) => {}),
20
+ capture: mock(async (_signal: AbortSignal) => ({
21
+ userGuidance,
22
+ screenshot: "anBlZw==",
23
+ screenshotWidthPx: 1440,
24
+ screenshotHeightPx: 810,
25
+ screenWidthPt: 1440,
26
+ screenHeightPt: 810,
27
+ })),
28
+ };
29
+ }
30
+
31
+ const perform = async (
32
+ toolName: string,
33
+ input: Record<string, unknown>,
34
+ signal: AbortSignal,
35
+ backend: Parameters<typeof performDesktopComputerUse>[2],
36
+ ) =>
37
+ performDesktopComputerUse(
38
+ planDesktopComputerUse(toolName, input),
39
+ signal,
40
+ backend,
41
+ );
42
+
43
+ const signal = () => new AbortController().signal;
44
+
45
+ describe("virtual desktop computer use", () => {
46
+ test("observes without input and returns pixels on repeated observations", async () => {
47
+ const backend = driver();
48
+ for (let i = 0; i < 2; i++) {
49
+ const result = await perform(
50
+ "computer_use_observe",
51
+ {},
52
+ signal(),
53
+ backend,
54
+ );
55
+ expect(result.screenshot).toBe("anBlZw==");
56
+ }
57
+ expect(backend.input).not.toHaveBeenCalled();
58
+ expect(backend.capture).toHaveBeenCalledTimes(2);
59
+ });
60
+
61
+ test("runs a sequence in order and captures once after its last action", async () => {
62
+ const backend = driver();
63
+ await perform(
64
+ "computer_use_sequence",
65
+ {
66
+ actions: [
67
+ { action: "click", x: 30, y: 40 },
68
+ { action: "key", key: "ctrl+l" },
69
+ { action: "type", text: "--shell $(example)" },
70
+ ],
71
+ },
72
+ signal(),
73
+ backend,
74
+ );
75
+ expect(backend.input.mock.calls.map(([args]) => args)).toEqual([
76
+ [
77
+ "mousemove",
78
+ "30",
79
+ "40",
80
+ "click",
81
+ "--repeat",
82
+ "1",
83
+ "--delay",
84
+ "100",
85
+ "1",
86
+ ],
87
+ ["key", "--clearmodifiers", "ctrl+l"],
88
+ ["type", "--clearmodifiers", "--delay", "0", "--", "--shell $(example)"],
89
+ ]);
90
+ expect(backend.capture).toHaveBeenCalledTimes(1);
91
+ });
92
+
93
+ for (const clickType of ["double", "right"] as const) {
94
+ test(`the bundled click wrapper executes a ${clickType} click`, async () => {
95
+ const backend = driver();
96
+ const proxy = new HostCuProxy(10);
97
+ await click(
98
+ { click_type: clickType, x: 30, y: 40 },
99
+ {
100
+ workingDir: "/tmp",
101
+ conversationId: "conv-123",
102
+ trustClass: "guardian",
103
+ proxyToolResolver: (name, input) =>
104
+ proxy.executeLocal(name, input, () =>
105
+ perform(name, input, signal(), backend),
106
+ ),
107
+ },
108
+ );
109
+ expect(backend.input.mock.calls[0][0]).toEqual([
110
+ "mousemove",
111
+ "30",
112
+ "40",
113
+ "click",
114
+ "--repeat",
115
+ clickType === "double" ? "2" : "1",
116
+ "--delay",
117
+ "100",
118
+ clickType === "right" ? "3" : "1",
119
+ ]);
120
+ expect(backend.capture).toHaveBeenCalledTimes(1);
121
+ });
122
+ }
123
+
124
+ test("invalid input does not claim a lease or consume the CU budget", async () => {
125
+ const proxy = new HostCuProxy(1);
126
+ const run = spyOn(desktopAutomationLease, "runBrowser");
127
+ try {
128
+ await expect(
129
+ executeDesktopComputerUse(
130
+ "computer_use_click",
131
+ { element_id: -1 },
132
+ {
133
+ workingDir: "/tmp",
134
+ conversationId: "conv-123",
135
+ trustClass: "guardian",
136
+ },
137
+ proxy,
138
+ ),
139
+ ).rejects.toThrow("element_id must be an integer");
140
+ expect(run).not.toHaveBeenCalled();
141
+ expect(proxy.stepCount).toBe(0);
142
+ expect(proxy.actionHistory).toHaveLength(0);
143
+ } finally {
144
+ run.mockRestore();
145
+ }
146
+ });
147
+
148
+ test("an execution failure stops a sequence and returns the resulting screen", async () => {
149
+ const backend = driver();
150
+ backend.input.mockRejectedValueOnce(new Error("Input failed"));
151
+ const result = await perform(
152
+ "computer_use_sequence",
153
+ {
154
+ actions: [
155
+ { action: "key", key: "ctrl+l" },
156
+ { action: "type", text: "example.com" },
157
+ ],
158
+ },
159
+ signal(),
160
+ backend,
161
+ );
162
+ expect(backend.input).toHaveBeenCalledTimes(1);
163
+ expect(result.executionError).toBe("Input failed");
164
+ expect(result.screenshot).toBe("anBlZw==");
165
+ });
166
+
167
+ test("done releases only its conversation without starting the desktop", async () => {
168
+ const release = spyOn(
169
+ desktopAutomationLease,
170
+ "releaseForConversation",
171
+ ).mockImplementation(() => {});
172
+ const run = spyOn(desktopAutomationLease, "runBrowser");
173
+ try {
174
+ const proxy = new HostCuProxy(1);
175
+ proxy.recordAction("computer_use_observe", {});
176
+ const result = await executeDesktopComputerUse(
177
+ "computer_use_done",
178
+ { summary: "Finished" },
179
+ {
180
+ workingDir: "/tmp",
181
+ conversationId: "conv-123",
182
+ trustClass: "guardian",
183
+ },
184
+ proxy,
185
+ );
186
+ expect(result).toEqual({ content: "Finished", isError: false });
187
+ expect(release).toHaveBeenCalledWith("conv-123");
188
+ expect(run).not.toHaveBeenCalled();
189
+ expect(proxy.stepCount).toBe(0);
190
+ } finally {
191
+ release.mockRestore();
192
+ run.mockRestore();
193
+ }
194
+ });
195
+
196
+ test("tool deadline cancels queued startup before it can perform an action", async () => {
197
+ const config = configLoader.getConfig();
198
+ const getConfig = spyOn(configLoader, "getConfig").mockReturnValue({
199
+ ...config,
200
+ timeouts: { ...config.timeouts, toolExecutionTimeoutSec: 0.01 },
201
+ });
202
+ const run = spyOn(desktopAutomationLease, "runBrowser").mockImplementation(
203
+ async (context) => {
204
+ return new Promise((_, reject) =>
205
+ context.signal!.addEventListener(
206
+ "abort",
207
+ () => reject(context.signal!.reason),
208
+ { once: true },
209
+ ),
210
+ );
211
+ },
212
+ );
213
+ try {
214
+ await expect(
215
+ executeDesktopComputerUse(
216
+ "computer_use_click",
217
+ { x: 1, y: 2, observation_id: "observation-123" },
218
+ {
219
+ workingDir: "/tmp",
220
+ conversationId: "conv-123",
221
+ trustClass: "guardian",
222
+ },
223
+ new HostCuProxy(10),
224
+ ),
225
+ ).rejects.toThrow("Computer use timed out");
226
+ expect(run.mock.calls[0][0].signal?.aborted).toBe(true);
227
+ } finally {
228
+ getConfig.mockRestore();
229
+ run.mockRestore();
230
+ }
231
+ });
232
+
233
+ test("validates the whole sequence before any input", async () => {
234
+ const backend = driver();
235
+ await expect(
236
+ perform(
237
+ "computer_use_sequence",
238
+ {
239
+ actions: [
240
+ { action: "click", x: 30, y: 40 },
241
+ { action: "click", element_id: -1 },
242
+ ],
243
+ },
244
+ signal(),
245
+ backend,
246
+ ),
247
+ ).rejects.toThrow("element_id must be an integer");
248
+ expect(backend.input).not.toHaveBeenCalled();
249
+ expect(backend.capture).not.toHaveBeenCalled();
250
+ });
251
+
252
+ for (const [tool, input] of [
253
+ ["computer_use_observe", { capture_window_id: 12 }],
254
+ ["computer_use_run_applescript", { script: "return 1" }],
255
+ ["computer_use_click", { x: 1440, y: 10 }],
256
+ ["computer_use_key", { key: "Return click 1" }],
257
+ ["computer_use_scroll", { direction: "down", amount: 100 }],
258
+ ] as const) {
259
+ test(`rejects unsupported input for ${tool} before acting`, async () => {
260
+ const backend = driver();
261
+ await expect(perform(tool, input, signal(), backend)).rejects.toThrow();
262
+ expect(backend.input).not.toHaveBeenCalled();
263
+ expect(backend.capture).not.toHaveBeenCalled();
264
+ });
265
+ }
266
+
267
+ test("cancellation releases a held drag button and prevents capture", async () => {
268
+ const backend = driver();
269
+ const abort = new AbortController();
270
+ backend.input.mockImplementation(async (_args, actionSignal) => {
271
+ if (actionSignal) {
272
+ abort.abort();
273
+ actionSignal.throwIfAborted();
274
+ }
275
+ });
276
+ await expect(
277
+ perform(
278
+ "computer_use_drag",
279
+ {
280
+ x: 30,
281
+ y: 40,
282
+ to_x: 300,
283
+ to_y: 400,
284
+ },
285
+ abort.signal,
286
+ backend,
287
+ ),
288
+ ).rejects.toThrow();
289
+ expect(backend.input.mock.calls.at(-1)).toEqual([["mouseup", "1"]]);
290
+ expect(backend.capture).not.toHaveBeenCalled();
291
+ });
292
+
293
+ test("an already-cancelled request does not actuate or capture", async () => {
294
+ const backend = driver();
295
+ const abort = new AbortController();
296
+ abort.abort();
297
+ await expect(
298
+ perform("computer_use_click", { x: 30, y: 40 }, abort.signal, backend),
299
+ ).rejects.toThrow();
300
+ expect(backend.input).not.toHaveBeenCalled();
301
+ expect(backend.capture).not.toHaveBeenCalled();
302
+ });
303
+
304
+ test("uses the CU step budget and image result format and resets on done", async () => {
305
+ const proxy = new HostCuProxy(1);
306
+ const capture = driver().capture;
307
+ const execute = () => capture(signal());
308
+ const result = await proxy.executeLocal(
309
+ "computer_use_observe",
310
+ {},
311
+ execute,
312
+ );
313
+ expect(result.contentBlocks).toEqual([
314
+ {
315
+ type: "image",
316
+ source: {
317
+ type: "base64",
318
+ media_type: "image/jpeg",
319
+ data: "anBlZw==",
320
+ },
321
+ },
322
+ ]);
323
+ expect(result.content).toContain("1440x810 px");
324
+ expect(proxy.actionHistory).toHaveLength(1);
325
+ expect(
326
+ (await proxy.executeLocal("computer_use_click", { x: 1, y: 2 }, execute))
327
+ .isError,
328
+ ).toBe(true);
329
+ expect(capture).toHaveBeenCalledTimes(1);
330
+ proxy.endTask("conv-123");
331
+ expect(
332
+ (await proxy.executeLocal("computer_use_observe", {}, execute)).isError,
333
+ ).toBe(false);
334
+ });
335
+ });
336
+
337
+ test("native dispatch consumes observations while preserving CU history, images, and loop warnings", async () => {
338
+ const warning = "Accessibility is unavailable. Use screenshot coordinates.";
339
+ const backend = driver(warning);
340
+ const manager = {
341
+ browser: { release: async () => {} },
342
+ acquireAutomationSlot: () => ({ ok: true }),
343
+ releaseAutomationSlot: () => {},
344
+ ensureDesktopRunning: async () => {},
345
+ } as unknown as sessionManager.DesktopSessionManager;
346
+ const lease = new DesktopAutomationLease({
347
+ enabled: () => true,
348
+ ready: () => true,
349
+ manager: () => manager,
350
+ notify: async () => {},
351
+ });
352
+ const run = spyOn(desktopAutomationLease, "runBrowser").mockImplementation(
353
+ lease.runBrowser.bind(lease),
354
+ );
355
+ const record = spyOn(
356
+ desktopAutomationLease,
357
+ "recordObservation",
358
+ ).mockImplementation(lease.recordObservation.bind(lease));
359
+ const getManager = spyOn(
360
+ sessionManager,
361
+ "getDesktopSessionManager",
362
+ ).mockReturnValue(manager);
363
+ const context = {
364
+ workingDir: "/tmp",
365
+ conversationId: "conv-123",
366
+ sourceActorPrincipalId: "user-123",
367
+ trustClass: "guardian" as const,
368
+ };
369
+ const proxy = new HostCuProxy(20);
370
+ const call = (name: string, input: Record<string, unknown> = {}) =>
371
+ executeDesktopComputerUse(
372
+ name,
373
+ { target: "assistant-desktop", ...input },
374
+ context,
375
+ proxy,
376
+ backend,
377
+ );
378
+ const id = (result: { content: string }) =>
379
+ result.content.match(/observation_id: ([a-f0-9-]+)/)![1]!;
380
+ try {
381
+ await expect(call("computer_use_click", { x: 30, y: 40 })).rejects.toThrow(
382
+ "observation_id",
383
+ );
384
+ expect(run).not.toHaveBeenCalled();
385
+ expect(proxy.stepCount).toBe(0);
386
+ const observed = await call("computer_use_observe");
387
+ expect(observed.contentBlocks?.[0]).toMatchObject({ type: "image" });
388
+ expect(observed.content).toContain(`USER GUIDANCE: ${warning}`);
389
+ expect(observed.content).toContain("pass the latest observation_id");
390
+ const firstId = id(observed);
391
+ let result = await call("computer_use_click", {
392
+ x: 30,
393
+ y: 40,
394
+ observation_id: firstId,
395
+ });
396
+ expect(id(result)).not.toBe(firstId);
397
+ const steps = proxy.stepCount;
398
+ await expect(
399
+ call("computer_use_click", { x: 30, y: 40, observation_id: firstId }),
400
+ ).rejects.toThrow("stale");
401
+ expect(backend.input).toHaveBeenCalledTimes(1);
402
+ expect(proxy.stepCount).toBe(steps);
403
+ for (let i = 0; i < 3; i++) {
404
+ result = await call("computer_use_click", {
405
+ x: 30,
406
+ y: 40,
407
+ observation_id: id(result),
408
+ });
409
+ }
410
+ expect(result.content).toContain("repeated");
411
+ await lease.runBrowser(context, async () => ({
412
+ content: "browser action",
413
+ isError: false,
414
+ }));
415
+ await expect(
416
+ call("computer_use_key", { key: "Return", observation_id: id(result) }),
417
+ ).rejects.toThrow("stale");
418
+ const refreshed = await call("computer_use_observe");
419
+ const sequenced = await call("computer_use_sequence", {
420
+ observation_id: id(refreshed),
421
+ actions: [
422
+ { action: "key", key: "ctrl+l" },
423
+ { action: "type", text: "example.com" },
424
+ ],
425
+ });
426
+ expect(backend.input.mock.calls.at(-1)?.[0]).toContain("example.com");
427
+ backend.input.mockRejectedValueOnce(new Error("Input failed"));
428
+ const failed = await call("computer_use_key", {
429
+ key: "Return",
430
+ observation_id: id(sequenced),
431
+ });
432
+ expect(failed.isError).toBe(true);
433
+ expect(failed.contentBlocks?.[0]).toMatchObject({ type: "image" });
434
+ expect(failed.content).not.toContain("observation_id:");
435
+ expect(failed.content).toContain("Call computer_use_observe");
436
+ expect(lease.isActive).toBe(false);
437
+ const recovered = await call("computer_use_observe");
438
+ expect(id(recovered)).toBeTruthy();
439
+ } finally {
440
+ await lease.runBrowser(
441
+ context,
442
+ async () => ({ content: "", isError: false }),
443
+ true,
444
+ );
445
+ run.mockRestore();
446
+ record.mockRestore();
447
+ getManager.mockRestore();
448
+ }
449
+ });