@piagent/platform 1.5.5 → 1.6.1

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 (323) hide show
  1. package/CHANGELOG.md +166 -0
  2. package/README.md +25 -21
  3. package/README.vi.md +34 -5
  4. package/adapters/web-frontend/profile.json +2 -4
  5. package/architecture/layers.json +18 -3
  6. package/benchmarks/deep-logic-v1/SPECIALIST_PROTOCOL.vi.md +122 -0
  7. package/benchmarks/deep-logic-v1/grade.mjs +272 -10
  8. package/benchmarks/deep-logic-v1/project/apps/admin/src/billing-summary.js +3 -0
  9. package/benchmarks/deep-logic-v1/project/benchmark-contract/public-smoke.test.js +166 -0
  10. package/benchmarks/deep-logic-v1/project/package.json +1 -1
  11. package/benchmarks/deep-logic-v1/project/packages/billing/src/close-period.js +3 -0
  12. package/benchmarks/deep-logic-v1/project/packages/billing/src/plan-timeline.js +3 -0
  13. package/benchmarks/deep-logic-v1/project/test/smoke.test.js +7 -1
  14. package/benchmarks/deep-logic-v1/prompts/budgeted-context-graph.md +2 -0
  15. package/benchmarks/deep-logic-v1/prompts/fair-dependency-scheduler.md +2 -0
  16. package/benchmarks/deep-logic-v1/prompts/layered-policy-resolution.md +2 -0
  17. package/benchmarks/deep-logic-v1/prompts/resumable-stream-assembly.md +3 -1
  18. package/benchmarks/deep-logic-v1/prompts/revision-event-reconciliation.md +2 -0
  19. package/benchmarks/deep-logic-v1/prompts/temporal-usage-billing-close.md +123 -0
  20. package/benchmarks/deep-logic-v1/prompts/transactional-config-merge.md +2 -0
  21. package/benchmarks/deep-logic-v1/suite.json +51 -3
  22. package/benchmarks/deep-logic-v1/variant.mjs +22 -1
  23. package/benchmarks/production-v1/spend-control.v1.json +92 -0
  24. package/benchmarks/production-v1/suite.json +46 -3
  25. package/catalog/capabilities.json +18 -18
  26. package/docs/authority-manifest.md +1 -1
  27. package/docs/auto-delegation-policy.md +38 -73
  28. package/docs/command-reference-vietnamese.md +27 -35
  29. package/docs/context-engine.md +12 -4
  30. package/docs/context-window-policy.md +37 -1
  31. package/docs/distribution-standard.md +7 -7
  32. package/docs/en/architecture.md +27 -13
  33. package/docs/languages.json +4 -4
  34. package/docs/model-options.md +8 -5
  35. package/docs/operator-manual-vietnamese.md +27 -45
  36. package/docs/project-adapters.md +2 -2
  37. package/docs/publishing-for-teams.md +10 -0
  38. package/docs/quality-benchmark.md +427 -62
  39. package/docs/release-install-policy.md +10 -9
  40. package/docs/runtime-harness-standard.md +3 -3
  41. package/docs/runtime-policy-design.md +4 -4
  42. package/docs/runtime-quality-baseline.md +13 -6
  43. package/docs/security-threat-model.md +2 -2
  44. package/docs/subagent-orchestration-capabilities.md +19 -51
  45. package/docs/subagents-and-multiagent.md +21 -47
  46. package/docs/task-implementation-contract.md +3 -3
  47. package/docs/task-lifecycle-vietnamese.md +3 -3
  48. package/docs/team-onboarding.md +2 -2
  49. package/docs/usage-observability.md +7 -2
  50. package/docs/vi/architecture.md +27 -13
  51. package/docs-site/architecture.html +1 -1
  52. package/docs-site/benchmark.html +89 -95
  53. package/docs-site/commands.html +1 -1
  54. package/docs-site/content/benchmark.html +88 -94
  55. package/docs-site/content/en/benchmark.html +134 -90
  56. package/docs-site/content/en/index.html +19 -19
  57. package/docs-site/content/en/runtime.html +4 -4
  58. package/docs-site/content/en/whats-new.html +12 -1
  59. package/docs-site/content/en/workflows.html +2 -2
  60. package/docs-site/content/index.html +19 -19
  61. package/docs-site/content/runtime.html +4 -4
  62. package/docs-site/content/whats-new.html +12 -1
  63. package/docs-site/content/workflows.html +2 -2
  64. package/docs-site/context-engine.html +1 -1
  65. package/docs-site/en/architecture.html +1 -1
  66. package/docs-site/en/benchmark.html +135 -91
  67. package/docs-site/en/commands.html +1 -1
  68. package/docs-site/en/context-engine.html +1 -1
  69. package/docs-site/en/index.html +20 -20
  70. package/docs-site/en/mcp-approval.html +1 -1
  71. package/docs-site/en/mcp-auth.html +1 -1
  72. package/docs-site/en/mcp-servers.html +1 -1
  73. package/docs-site/en/mcp.html +1 -1
  74. package/docs-site/en/permissions.html +1 -1
  75. package/docs-site/en/profiles.html +1 -1
  76. package/docs-site/en/quickstart.html +1 -1
  77. package/docs-site/en/runtime.html +5 -5
  78. package/docs-site/en/scope.html +1 -1
  79. package/docs-site/en/security.html +1 -1
  80. package/docs-site/en/team.html +1 -1
  81. package/docs-site/en/whats-new.html +13 -2
  82. package/docs-site/en/workflows.html +3 -3
  83. package/docs-site/index.html +20 -20
  84. package/docs-site/mcp-approval.html +1 -1
  85. package/docs-site/mcp-auth.html +1 -1
  86. package/docs-site/mcp-servers.html +1 -1
  87. package/docs-site/mcp.html +1 -1
  88. package/docs-site/permissions.html +1 -1
  89. package/docs-site/profiles.html +1 -1
  90. package/docs-site/quickstart.html +1 -1
  91. package/docs-site/runtime.html +5 -5
  92. package/docs-site/scope.html +1 -1
  93. package/docs-site/security.html +1 -1
  94. package/docs-site/team.html +1 -1
  95. package/docs-site/whats-new.html +13 -2
  96. package/docs-site/workflows.html +3 -3
  97. package/evals/fixtures/benchmark-assurance-evidence.invalid.json +1 -1
  98. package/evals/fixtures/benchmark-assurance-evidence.valid.json +1 -1
  99. package/evals/fixtures/codex-relative-efficiency-v1.invalid.json +4 -0
  100. package/evals/fixtures/codex-relative-efficiency-v1.valid.json +94 -0
  101. package/evals/fixtures/helper-request.invalid.json +1 -1
  102. package/evals/fixtures/helper-request.valid.json +1 -1
  103. package/evals/fixtures/piagent-webui/session-live-state-v1.invalid.json +18 -0
  104. package/evals/fixtures/piagent-webui/session-live-state-v1.valid.json +27 -0
  105. package/evals/fixtures/piagent-webui/subagent-tree-v1.valid.json +7 -0
  106. package/evals/fixtures/task-authority-snapshot.valid.json +4 -4
  107. package/evals/fs4-readiness-matrix.v1.json +2 -2
  108. package/evals/fs5-causal-arm.v1.json +3 -3
  109. package/evals/fs5-pilot-protocol.v1.json +3 -3
  110. package/evals/fs5-pilot-protocol.v2.json +8 -8
  111. package/evals/fs5-pilot-protocol.v3.json +8 -8
  112. package/evals/fs5-pilot-protocol.v4.json +7 -6
  113. package/evals/fs5-pilot-protocol.v5.json +9 -7
  114. package/evals/private-holdout-v1/public-exposure.v1.json +1 -1
  115. package/evals/runtime-conformance-v1/README.md +43 -0
  116. package/evals/runtime-conformance-v1/evaluate-paid-canary.mjs +198 -0
  117. package/evals/runtime-conformance-v1/lane.json +57 -0
  118. package/evals/runtime-conformance-v1/paid-canary.json +31 -0
  119. package/evals/runtime-conformance-v1/runner.mjs +456 -0
  120. package/package.json +3 -1
  121. package/packages/piagent-core/README.md +13 -4
  122. package/packages/piagent-core/benchmark/benchmark-assurance.js +4 -1
  123. package/packages/piagent-core/benchmark/benchmark-bootstrap.js +91 -9
  124. package/packages/piagent-core/benchmark/benchmark-claim-restrictions.js +38 -7
  125. package/packages/piagent-core/benchmark/benchmark-cli.js +7 -7
  126. package/packages/piagent-core/benchmark/benchmark-codex-relative-efficiency.js +698 -0
  127. package/packages/piagent-core/benchmark/benchmark-codex.js +3 -1
  128. package/packages/piagent-core/benchmark/benchmark-comparison.js +271 -2
  129. package/packages/piagent-core/benchmark/benchmark-core.js +736 -664
  130. package/packages/piagent-core/benchmark/benchmark-execution-guard.js +7 -5
  131. package/packages/piagent-core/benchmark/benchmark-forensics.js +441 -4
  132. package/packages/piagent-core/benchmark/benchmark-host-readiness.js +639 -0
  133. package/packages/piagent-core/benchmark/benchmark-matrix.js +22 -4
  134. package/packages/piagent-core/benchmark/benchmark-normalized-cost.js +317 -0
  135. package/packages/piagent-core/benchmark/benchmark-preflight.js +5 -1
  136. package/packages/piagent-core/benchmark/benchmark-process.js +6 -5
  137. package/packages/piagent-core/benchmark/benchmark-provider-free-evidence.js +345 -0
  138. package/packages/piagent-core/benchmark/benchmark-provider-wire.js +126 -0
  139. package/packages/piagent-core/benchmark/benchmark-record-validation.js +252 -0
  140. package/packages/piagent-core/benchmark/benchmark-report.js +130 -12
  141. package/packages/piagent-core/benchmark/benchmark-resume-recovery.js +14 -0
  142. package/packages/piagent-core/benchmark/benchmark-runner-policy.js +26 -0
  143. package/packages/piagent-core/benchmark/benchmark-runtime-causal.js +244 -0
  144. package/packages/piagent-core/benchmark/benchmark-stage-diagnostic.js +979 -0
  145. package/packages/piagent-core/benchmark/benchmark-statistics.js +64 -0
  146. package/packages/piagent-core/benchmark/benchmark-stop-policy.js +8 -5
  147. package/packages/piagent-core/benchmark/benchmark-suite-identity.js +31 -0
  148. package/packages/piagent-core/benchmark/benchmark-suite-runtime.js +13 -9
  149. package/packages/piagent-core/benchmark/benchmark-suite.js +149 -2
  150. package/packages/piagent-core/benchmark/benchmark-summary-support.js +380 -0
  151. package/packages/piagent-core/benchmark/benchmark-timing-diagnostics.js +971 -0
  152. package/packages/piagent-core/benchmark/benchmark-usage.js +322 -0
  153. package/packages/piagent-core/benchmark/benchmark-workflow.js +166 -0
  154. package/packages/piagent-core/capabilities/runtime-integrity.js +3 -0
  155. package/packages/piagent-core/docs/context-efficiency-improvement-plan.md +1080 -0
  156. package/packages/piagent-core/extensions/acceptance-behavior-proof.js +806 -0
  157. package/packages/piagent-core/extensions/acceptance-contract-semantics.js +83 -69
  158. package/packages/piagent-core/extensions/acceptance-error-classes.js +160 -0
  159. package/packages/piagent-core/extensions/acceptance-executable-evidence.js +12 -9
  160. package/packages/piagent-core/extensions/acceptance-receipt.js +123 -59
  161. package/packages/piagent-core/extensions/acceptance-state-machine-evidence.js +518 -0
  162. package/packages/piagent-core/extensions/acceptance-test-binding-evidence.js +60 -0
  163. package/packages/piagent-core/extensions/context-efficiency-metrics.js +302 -0
  164. package/packages/piagent-core/extensions/context-efficiency-report.js +167 -0
  165. package/packages/piagent-core/extensions/context-engine.js +149 -156
  166. package/packages/piagent-core/extensions/context-evidence.js +48 -0
  167. package/packages/piagent-core/extensions/context-import-links.js +296 -0
  168. package/packages/piagent-core/extensions/context-index-policy.js +4 -0
  169. package/packages/piagent-core/extensions/context-query-paths.js +57 -0
  170. package/packages/piagent-core/extensions/context-telemetry-integrity.js +65 -0
  171. package/packages/piagent-core/extensions/criterion-context-pack.js +435 -7
  172. package/packages/piagent-core/extensions/criterion-graph.js +82 -9
  173. package/packages/piagent-core/extensions/fallback-base-policy.ts +101 -0
  174. package/packages/piagent-core/extensions/guard-types.ts +6 -0
  175. package/packages/piagent-core/extensions/javascript-regex-evidence.js +32 -0
  176. package/packages/piagent-core/extensions/piagent-guard.ts +420 -260
  177. package/packages/piagent-core/extensions/task-contract-view.js +17 -3
  178. package/packages/piagent-core/extensions/task-lifecycle.js +39 -2
  179. package/packages/piagent-core/extensions/task-state.js +109 -142
  180. package/packages/piagent-core/extensions/verification-intelligence.js +52 -13
  181. package/packages/piagent-core/extensions/workspace-evidence-roots.js +193 -0
  182. package/packages/piagent-core/package.json +1 -1
  183. package/packages/piagent-core/policies/base-policy.json +44 -2
  184. package/packages/piagent-core/policy/authority-manifest.v1.json +3 -3
  185. package/packages/piagent-core/prompts/be-to-fe.md +2 -8
  186. package/packages/piagent-core/prompts/plan.md +1 -1
  187. package/packages/piagent-core/prompts/platform-improve.md +2 -6
  188. package/packages/piagent-core/prompts/review.md +2 -7
  189. package/packages/piagent-core/prompts/scout.md +1 -1
  190. package/packages/piagent-core/prompts/task.md +4 -3
  191. package/packages/piagent-core/runtime/context/context-delivery.ts +103 -0
  192. package/packages/piagent-core/runtime/context/context-delta-shadow.ts +53 -0
  193. package/packages/piagent-core/runtime/context/context-efficiency-format.ts +47 -0
  194. package/packages/piagent-core/runtime/context/context-evidence-qualification.ts +32 -0
  195. package/packages/piagent-core/runtime/context/prefix-telemetry.ts +47 -0
  196. package/packages/piagent-core/runtime/context/repository-memory-hints.ts +27 -0
  197. package/packages/piagent-core/runtime/hooks/agent-prompt-telemetry.ts +56 -0
  198. package/packages/piagent-core/runtime/hooks/agent-start-hook.ts +162 -102
  199. package/packages/piagent-core/runtime/hooks/completion-hook.ts +12 -12
  200. package/packages/piagent-core/runtime/hooks/input-hook.ts +17 -1
  201. package/packages/piagent-core/runtime/hooks/session-hooks.ts +22 -2
  202. package/packages/piagent-core/runtime/hooks/session-start-hook.ts +107 -3
  203. package/packages/piagent-core/runtime/hooks/tool-result-content-guards.ts +95 -0
  204. package/packages/piagent-core/runtime/hooks/tool-result-hook.ts +134 -166
  205. package/packages/piagent-core/runtime/hooks/tool-result-value-helpers.ts +66 -0
  206. package/packages/piagent-core/runtime/inspection/activity-event-adapter.ts +4 -2
  207. package/packages/piagent-core/runtime/inspection/activity-recovery.ts +104 -0
  208. package/packages/piagent-core/runtime/inspection/bounded-emission-guard.ts +286 -0
  209. package/packages/piagent-core/runtime/inspection/runtime-event-store.ts +37 -7
  210. package/packages/piagent-core/runtime/inspection/task-compaction-history.ts +1 -1
  211. package/packages/piagent-core/runtime/inspection/task-handoff-history.ts +1 -1
  212. package/packages/piagent-core/runtime/inspection/task-subagent-tree.ts +27 -4
  213. package/packages/piagent-core/runtime/inspection/tool-failure-classification.ts +70 -0
  214. package/packages/piagent-core/runtime/inspection/webui-snapshot.ts +73 -9
  215. package/packages/piagent-core/runtime/model/authenticated-catalog.ts +17 -9
  216. package/packages/piagent-core/runtime/model/openai-codex-reasoning.ts +83 -0
  217. package/packages/piagent-core/runtime/model/provider-wire-fingerprint.ts +238 -0
  218. package/packages/piagent-core/runtime/orchestration/helper-lifecycle.ts +16 -12
  219. package/packages/piagent-core/runtime/orchestration/minimal-delegation-policy.ts +71 -0
  220. package/packages/piagent-core/runtime/orchestration/owned-work-budget.ts +28 -20
  221. package/packages/piagent-core/runtime/orchestration/role-policy.ts +27 -9
  222. package/packages/piagent-core/runtime/orchestration/subagent-tool-policy.ts +329 -0
  223. package/packages/piagent-core/runtime/policy/block-remedy.ts +0 -4
  224. package/packages/piagent-core/runtime/product/activity-inspector.ts +7 -2
  225. package/packages/piagent-core/runtime/product/efficiency-metrics.ts +7 -2
  226. package/packages/piagent-core/runtime/quality/edit-freshness-guard.ts +102 -0
  227. package/packages/piagent-core/runtime/quality/exact-output-contract.ts +22 -11
  228. package/packages/piagent-core/runtime/quality/performance-assurance-signals.ts +23 -0
  229. package/packages/piagent-core/runtime/quality/performance-assurance.ts +9 -5
  230. package/packages/piagent-core/runtime/recovery/continuation-budget.ts +1 -1
  231. package/packages/piagent-core/runtime/recovery/edit-recovery-context.ts +113 -0
  232. package/packages/piagent-core/runtime/recovery/edit-recovery-delivery.ts +42 -0
  233. package/packages/piagent-core/runtime/recovery/handoff-projection.ts +4 -1
  234. package/packages/piagent-core/runtime/recovery/pristine-task-policy-refresh.ts +91 -0
  235. package/packages/piagent-core/runtime/recovery/recovery-policy.ts +7 -4
  236. package/packages/piagent-core/runtime/recovery/resume-state.ts +112 -14
  237. package/packages/piagent-core/runtime/recovery/semantic-repair-handshake.ts +60 -13
  238. package/packages/piagent-core/runtime/recovery/semantic-repair-runtime.ts +55 -2
  239. package/packages/piagent-core/runtime/recovery/task-evidence-coverage-refresh.ts +217 -0
  240. package/packages/piagent-core/runtime/registration/context-commands.ts +82 -7
  241. package/packages/piagent-core/runtime/registration/knowledge-tools.ts +9 -5
  242. package/packages/piagent-core/runtime/registration/operator-request-intake.ts +49 -0
  243. package/packages/piagent-core/runtime/registration/policy-tools.ts +50 -15
  244. package/packages/piagent-core/runtime/registration/session-commands.ts +1 -1
  245. package/packages/piagent-core/runtime/registration/task-completion-tools.ts +61 -25
  246. package/packages/piagent-core/runtime/registration/task-evidence-tools.ts +21 -12
  247. package/packages/piagent-core/runtime/registration/task-start-guidance.ts +169 -0
  248. package/packages/piagent-core/runtime/registration/task-start-manifest.ts +54 -0
  249. package/packages/piagent-core/runtime/registration/task-start-retry-helpers.ts +37 -0
  250. package/packages/piagent-core/runtime/registration/task-start-tool.ts +108 -103
  251. package/packages/piagent-core/runtime/runtime-limits.ts +16 -2
  252. package/packages/piagent-core/runtime/session/adaptive-context-analysis.ts +461 -0
  253. package/packages/piagent-core/runtime/session/adaptive-context-governor.ts +464 -0
  254. package/packages/piagent-core/runtime/session/adaptive-context-ledger.ts +265 -0
  255. package/packages/piagent-core/runtime/session/message-signals.ts +7 -3
  256. package/packages/piagent-core/runtime/session/operator-request-carry.ts +44 -0
  257. package/packages/piagent-core/runtime/session/runtime-state.ts +161 -5
  258. package/packages/piagent-core/runtime/session/source-checkout-read-grants.ts +35 -0
  259. package/packages/piagent-core/runtime/session/system-prompt.ts +76 -14
  260. package/packages/piagent-core/runtime/session/task-contract-presentation.ts +39 -0
  261. package/packages/piagent-core/runtime/tools/apply-patch-tool.ts +500 -0
  262. package/packages/piagent-core/runtime/tools/phase-tools.ts +8 -8
  263. package/packages/piagent-core/runtime/tools/tool-groups.ts +12 -2
  264. package/packages/piagent-core/runtime/trajectory/trajectory-runtime.ts +2 -1
  265. package/packages/piagent-core/runtime/verification/exact-verifier-reuse.ts +101 -0
  266. package/packages/piagent-core/runtime/workflows/runtime-intake-compaction.ts +164 -0
  267. package/packages/piagent-core/runtime/workflows/task-intake.ts +207 -38
  268. package/packages/piagent-core/security/sensitive-data.js +13 -218
  269. package/packages/piagent-core/security/sensitive-project-file.js +42 -0
  270. package/packages/piagent-core/security/sensitive-source-expression.js +330 -0
  271. package/packages/piagent-core/security/sensitive-source-formats.js +394 -0
  272. package/packages/piagent-core/security/sensitive-text.js +234 -0
  273. package/packages/piagent-core/subagents/piagent-oracle.md +5 -3
  274. package/packages/piagent-core/subagents/piagent-planner.md +6 -4
  275. package/packages/piagent-core/subagents/piagent-reviewer.md +5 -3
  276. package/packages/piagent-core/subagents/piagent-scout.md +4 -2
  277. package/packages/piagent-core/subagents/piagent-worker.md +7 -5
  278. package/packages/piagent-webui/dist/client/assets/{MarkdownMessage-rD9WJKxs.js → MarkdownMessage-B4XehTCU.js} +1 -1
  279. package/packages/piagent-webui/dist/client/assets/index-CgOEyRPG.js +176 -0
  280. package/packages/piagent-webui/dist/client/index.html +1 -1
  281. package/packages/piagent-webui/extension/piagent-webui.ts +3 -1
  282. package/packages/piagent-webui/extension/same-session-bridge.ts +1 -1
  283. package/packages/piagent-webui/extension/session-stream-adapter.ts +33 -10
  284. package/packages/piagent-webui/gateway/gateway-events.ts +109 -5
  285. package/packages/piagent-webui/gateway/gateway-service.ts +3 -0
  286. package/packages/piagent-webui/gateway/gateway-session-stream.ts +122 -20
  287. package/packages/piagent-webui/gateway/session-command-controller.ts +18 -3
  288. package/packages/piagent-webui/gateway/session-effective-options.ts +47 -0
  289. package/packages/piagent-webui/gateway/session-live-state.ts +21 -0
  290. package/packages/piagent-webui/gateway/session-operation-lifecycle.ts +237 -0
  291. package/packages/piagent-webui/gateway/session-operation-retry-control.ts +38 -0
  292. package/packages/piagent-webui/gateway/session-operation-start.ts +5 -1
  293. package/packages/piagent-webui/gateway/session-operation-watchdog.ts +165 -0
  294. package/packages/piagent-webui/gateway/session-runtime-factory.ts +45 -0
  295. package/packages/piagent-webui/gateway/session-runtime-supervisor.ts +141 -125
  296. package/packages/piagent-webui/ownership/session-lease-store.ts +5 -2
  297. package/packages/piagent-webui/server/core-inspection-provider.ts +2 -1
  298. package/packages/piagent-webui/server/loopback-server.ts +8 -0
  299. package/packages/piagent-webui/server/transcript-projection.ts +113 -9
  300. package/packages/piagent-webui/shared/text-visibility.ts +10 -0
  301. package/schemas/codex-relative-efficiency-v1.schema.json +155 -0
  302. package/schemas/piagent-webui/catalog-v1.json +9 -0
  303. package/schemas/piagent-webui/gateway-protocol-v1.schema.json +53 -3
  304. package/schemas/piagent-webui/runtime-event-v2.schema.json +15 -5
  305. package/schemas/piagent-webui/session-live-state-v1.schema.json +76 -0
  306. package/schemas/piagent-webui/subagent-tree-v1.schema.json +11 -1
  307. package/schemas/project-profile.schema.json +1 -1
  308. package/schemas/task-contract.schema.json +27 -1
  309. package/scripts/benchmark-runner-core.mjs +367 -363
  310. package/scripts/benchmark-runner-finalization.mjs +313 -0
  311. package/scripts/benchmark-runner-host-readiness.mjs +94 -0
  312. package/scripts/benchmark-runner-support.mjs +200 -0
  313. package/scripts/benchmark-session.mjs +95 -41
  314. package/scripts/configure-model-scope.sh +4 -3
  315. package/scripts/configure-subagents.sh +42 -34
  316. package/scripts/helper-evaluation.mjs +2 -2
  317. package/scripts/install-global.sh +2 -2
  318. package/scripts/pi-usage-history.mjs +37 -10
  319. package/scripts/setup.sh +2 -2
  320. package/scripts/team-doctor.sh +1 -1
  321. package/scripts/verify-local.sh +10 -1
  322. package/templates/global/settings.json +2 -1
  323. package/packages/piagent-webui/dist/client/assets/index-BZRFw1u6.js +0 -174
@@ -21,8 +21,8 @@
21
21
  "en": "README.md",
22
22
  "vi": "README.vi.md",
23
23
  "consistentAt": {
24
- "en": "1c890fe5b2a1f53f074b54f90d4610b5ed648242",
25
- "vi": "6b7a18e5f6f18012a4d96a70b5e2815299629b05"
24
+ "en": "ec5765bc81848d3cd43806ef5a7003e8a23e3ad7",
25
+ "vi": "51d117e51003133b84540cb7ea732577f5a603d9"
26
26
  }
27
27
  },
28
28
  {
@@ -39,8 +39,8 @@
39
39
  "en": "docs/en/architecture.md",
40
40
  "vi": "docs/vi/architecture.md",
41
41
  "consistentAt": {
42
- "en": "5bcb0aa311b360408ca73425744baee71e740a9b",
43
- "vi": "b613cec25fa1737636121c11cb2c5aecb2e00375"
42
+ "en": "6a8a6f73bd6b0bde3fad00f0996426eaa101a8a9",
43
+ "vi": "580addecc766aa7317c3c49d9d740bc05f1bb43d"
44
44
  }
45
45
  },
46
46
  {
@@ -54,7 +54,10 @@ Shift+Ctrl+P # cycle ngược
54
54
  Shift+Tab # cycle thinking level
55
55
  ```
56
56
 
57
- Global setup của repo sẽ seed `enabledModels` vào `~/.pi/agent/settings.json`, nên sau OAuth anh chọn/đổi model bằng option selector của Pi.
57
+ Global setup của repo sẽ seed các preset đã kiểm chứng và một catch-all
58
+ `openai-codex/*` vào `~/.pi/agent/settings.json`. Vì các preset đứng trước glob,
59
+ chúng giữ thinking level khuyến nghị; model Codex mới trong authenticated catalog
60
+ vẫn tự xuất hiện trong scope mà không cần chờ Piagent phát hành lại.
58
61
 
59
62
  ## Kiểm tra model hiện có
60
63
 
@@ -97,7 +100,7 @@ off, minimal, low, medium, high, xhigh, max
97
100
  Không phải model nào cũng hỗ trợ mọi level. Pi sẽ clamp theo capability của model. Cách chọn:
98
101
 
99
102
  ```bash
100
- pi --model openai-codex/gpt-5.5:high
103
+ pi --model openai-codex/gpt-5.6-sol:high
101
104
  pi --model anthropic/claude-sonnet-5:xhigh
102
105
  pi --thinking medium
103
106
  ```
@@ -113,7 +116,7 @@ pi --thinking medium
113
116
  | `openai-codex/gpt-5.3-codex-spark` | fast scout | hỏi nhanh, thao tác nhỏ, chi phí thấp |
114
117
  | `openai-codex/gpt-5.4-mini` | fast/cheap | scout nhẹ, docs, simple fix |
115
118
  | `openai-codex/gpt-5.4` | balanced | task bình thường |
116
- | `openai-codex/gpt-5.5` | balanced/hard default | default mạnh cho implement |
119
+ | `openai-codex/gpt-5.5` | balanced/hard | model thế hệ trước vẫn được hỗ trợ khi có trong authenticated catalog |
117
120
  | `openai-codex/gpt-5.6-luna` | cost-sensitive/high-volume | workload nhanh, nhiều lượt, ưu tiên latency và chi phí |
118
121
  | `openai-codex/gpt-5.6-terra` | balanced | công việc hằng ngày cần cân bằng capability, speed và cost |
119
122
  | `openai-codex/gpt-5.6-sol` | frontier/quality-first | task phức tạp, long-horizon hoặc cần chất lượng cao nhất |
@@ -143,7 +146,7 @@ available nào trong authenticated provider catalog.
143
146
  | Preset | Model OpenAI Codex | Model Claude (Anthropic) | Khi dùng |
144
147
  |---|---|---|---|
145
148
  | Fast scout | `openai-codex/gpt-5.4-mini:low` | `anthropic/claude-haiku-4-5:low` | đọc nhanh, hỏi đáp, grep/scout nhẹ |
146
- | Balanced implement | `openai-codex/gpt-5.5:medium` | `anthropic/claude-sonnet-5:medium` | task source bình thường |
149
+ | Balanced implement | `openai-codex/gpt-5.6-terra:medium` | `anthropic/claude-sonnet-5:medium` | task source bình thường |
147
150
  | Cost-sensitive volume | `openai-codex/gpt-5.6-luna:medium` | `anthropic/claude-haiku-4-5:low` | workload nhanh, nhiều lượt; đo quality gate trước khi promote |
148
151
  | Balanced implement | `openai-codex/gpt-5.6-terra:high` hoặc `openai-codex/gpt-5.5:high` | `anthropic/claude-sonnet-5:high` | task source hằng ngày, cân bằng capability/cost |
149
152
  | Frontier/quality-first | `openai-codex/gpt-5.6-sol:high` | `anthropic/claude-opus-4-7:xhigh` hoặc `anthropic/claude-opus-4-8:xhigh` | architecture, long-horizon, high-risk review |
@@ -155,7 +158,7 @@ Tên model có thể đổi theo Pi model catalog. Khi không chắc, ưu tiên
155
158
  Global setup mặc định:
156
159
 
157
160
  ```bash
158
- piagent-model-scope --preset full --default-model openai-codex/gpt-5.5:high
161
+ piagent-model-scope --preset full --default-model openai-codex/gpt-5.6-sol:high
159
162
  ```
160
163
 
161
164
  Scope này ghi vào settings:
@@ -19,12 +19,12 @@ Support matrix release hiện tại: macOS Apple Silicon + Bash và Linux x64 +
19
19
  ```bash
20
20
  node --version # >= 22.19.0
21
21
  npm install -g --ignore-scripts @earendil-works/pi-coding-agent@0.84.1
22
- npm install -g --ignore-scripts @piagent/platform@1.5.5
22
+ npm install -g --ignore-scripts @piagent/platform@1.6.1
23
23
  piagent-install --stable --dry-run
24
24
  piagent-install --stable
25
25
  ```
26
26
 
27
- Khi seed `.pi/settings.json` cho team/repo cần audit lặp lại, giữ package source dạng pinned tag như `git:github.com/Vt-mmm/piagent@v1.5.5`. Máy cá nhân có thể dùng `git:github.com/Vt-mmm/piagent` để theo latest.
27
+ Khi seed `.pi/settings.json` cho team/repo cần audit lặp lại, giữ package source dạng pinned tag như `git:github.com/Vt-mmm/piagent@v1.6.1`. Máy cá nhân có thể dùng `git:github.com/Vt-mmm/piagent` để theo latest.
28
28
 
29
29
  Nếu đang ở source checkout của platform, dùng helper theo channel để preview trước khi đổi:
30
30
 
@@ -100,7 +100,7 @@ install package once
100
100
 
101
101
  ```bash
102
102
  npm install -g --ignore-scripts @earendil-works/pi-coding-agent@0.84.1
103
- npm install -g --ignore-scripts @piagent/platform@1.5.5
103
+ npm install -g --ignore-scripts @piagent/platform@1.6.1
104
104
  piagent-install --stable --dry-run
105
105
  piagent-install --stable
106
106
  ```
@@ -283,7 +283,7 @@ Không phải daily default. Dùng khi muốn tạo sẵn `.pi` files cho projec
283
283
  bash /path/to/piagent/scripts/setup.sh /path/to/project \
284
284
  --project-only \
285
285
  --profile auto \
286
- --package-source git:github.com/Vt-mmm/piagent@v1.5.5 \
286
+ --package-source git:github.com/Vt-mmm/piagent@v1.6.1 \
287
287
  --mcp-preset core \
288
288
  --subagents-preset safe
289
289
  ```
@@ -673,7 +673,8 @@ piagent-benchmark
673
673
  piagent-benchmark --production --dry-run
674
674
  ```
675
675
 
676
- Lệnh mặc định tự chạy 4 scenario trên Raw Pi và Piagent, mỗi bên 3 lần; token,
676
+ Lệnh mặc định tự chạy 4 scenario trên Piagent và controlled `codex-cli` bằng
677
+ Luna/medium, mỗi bên 3 lần; token,
677
678
  cost, model và thinking được đọc từ Pi session thay vì nhập tay. Runner chỉ cho
678
679
  phép kết luận tiết kiệm token khi safety đạt `10/10` và các gate quality,
679
680
  reliability, workflow của suite đều pass. Production yêu cầu các điểm tổng hợp
@@ -688,9 +689,10 @@ profile/lifecycle/difficulty và cận trên 95% của paired token ratio. Seed/
688
689
  động không đi vào agent workspace; root seed chỉ nằm trong private report để
689
690
  tái lập bằng `--seed`.
690
691
 
691
- Production runner mặc định retry tối đa hai lần, cách nhau 60 giây, chỉ khi process chết trước
692
- khi ghi usage. Timeout hoặc task đã tiêu token rồi fail vẫn là reliability failure thật,
693
- không được retry. Report hiện số retry và giữ chi tiết local trong
692
+ Production/deep runner mặc định không retry và release gate yêu cầu đúng zero
693
+ retry cùng zero unknown provider-attempt usage. Override retry chỉ dùng để chẩn
694
+ đoán recovery và sẽ làm claim fail. Timeout hoặc task đã tiêu token rồi fail vẫn
695
+ là reliability failure thật. Report hiện số retry và giữ chi tiết local trong
694
696
  `infrastructure-attempts.jsonl`. Dùng `--scenarios <id,id> --repeats 1` để debug
695
697
  rubric; subset không thể vượt production release gate.
696
698
 
@@ -922,13 +924,13 @@ Default safe config:
922
924
 
923
925
  - `toolDescriptionMode: compact`;
924
926
  - `asyncByDefault: false`;
925
- - `parallel.concurrency: 3`;
926
- - `parallel.maxTasks: 6`;
927
+ - `parallel.concurrency: 1`;
928
+ - `parallel.maxTasks: 1`;
927
929
  - `maxSubagentDepth: 1`;
928
- - `maxSubagentSpawnsPerSession: 32`;
930
+ - `maxSubagentSpawnsPerSession: 1`;
929
931
  - scheduled runs off;
930
932
  - worktree base stable;
931
- - intercom bridge on.
933
+ - wait/intercom bridge off; builtin và worker disabled.
932
934
 
933
935
  ### Piagent agents
934
936
 
@@ -936,13 +938,13 @@ Default safe config:
936
938
  |---|---|---|
937
939
  | `piagent-scout` | Map source/spec read-only. | No write. |
938
940
  | `piagent-planner` | Plan implementation + verify gates. | No write. |
939
- | `piagent-worker` | Implement approved bounded task. | Write in scope. |
941
+ | `piagent-worker` | Compatibility metadata only; Piagent config disables it. | Không được dispatch. |
940
942
  | `piagent-reviewer` | Review diff/tests/scope. | Review-first. |
941
943
  | `piagent-oracle` | Risk/architecture challenge. | No write. |
942
944
 
943
945
  ### Auto-delegation
944
946
 
945
- Với `/workflow task`, `/workflow be-to-fe`, `/workflow platform-improve`, `/workflow plan`, `/workflow review`, parent agent dùng solo-first orchestration: lập task tree/review lenses trước, rồi chỉ spawn subagent khi có phần việc độc lập và đáng token. Alias cũ như `/task` vẫn giữ cùng policy.
947
+ Với `/workflow task`, `/workflow be-to-fe`, `/workflow platform-improve`, `/workflow plan`, `/workflow review`, parent agent tự suy luận và implement. Helper mặc định tắt. Khi operator bật rõ, runtime chỉ được dispatch một helper read-only, context fresh, nếu có ít nhất hai lane độc lập và projected net token saving từ 30% sau handoff/merge. Worker, retry, nested và parallel helper đều tắt. Alias cũ như `/task` giữ cùng policy.
946
948
 
947
949
  Kiểm tra nhanh policy:
948
950
 
@@ -952,13 +954,12 @@ Kiểm tra nhanh policy:
952
954
 
953
955
  Không cần tự gọi `/run` cho task bình thường. Chỉ dùng `/run` khi muốn ép rõ role hoặc debug.
954
956
 
955
- Parent nên spawn khi:
957
+ Helper chỉ đủ điều kiện khi đồng thời:
956
958
 
957
- - cần scout codebase rộng;
958
- - cần map BE/spec read-only trước khi FE implement;
959
- - cần reviewer độc lập;
960
- - cần context builder cho task lớn;
961
- - cần oracle/risk challenge.
959
+ - operator đã opt-in;
960
+ - có ít nhất hai workstream độc lập và helper không phụ thuộc output tiếp theo của parent;
961
+ - context chuyển giao fresh, tối đa 2.048 token và không inherit history;
962
+ - tổng token parent + helper + handoff + merge được dự báo thấp hơn solo ít nhất 30%.
962
963
 
963
964
  Không nên spawn khi:
964
965
 
@@ -973,27 +974,15 @@ Không nên spawn khi:
973
974
  ```text
974
975
  /run piagent-scout "Map listing page state flow. Read-only."
975
976
  /run piagent-planner "Plan FE implementation from this backend contract."
976
- /run piagent-worker "Implement the approved plan. Do not touch backend."
977
977
  /run piagent-reviewer "Review current diff for correctness, tests, and scope drift."
978
978
  /run piagent-oracle "Challenge this architecture decision before implementation."
979
979
  ```
980
980
 
981
- Parallel read-only review:
982
-
983
- ```text
984
- /parallel piagent-reviewer "Review correctness" -> piagent-reviewer "Review tests" -> piagent-reviewer "Review scope drift"
985
- ```
986
-
987
- Chain:
988
-
989
- ```text
990
- /chain piagent-scout "Scout target area" -> piagent-planner "Plan from {previous}" -> piagent-worker "Implement from {previous}" -> piagent-reviewer "Review implementation"
991
- ```
981
+ `/parallel`, `/chain` và writer role là capability upstream của package `pi-subagents`, nhưng Piagent config chủ động chặn các surface đó để tránh nhân context và token.
992
982
 
993
983
  Background:
994
984
 
995
985
  ```text
996
- /run piagent-scout "Map this module" --bg
997
986
  /subagents-fleet
998
987
  ```
999
988
 
@@ -1005,26 +994,19 @@ subagent({ action: "status", view: "fleet" })
1005
994
  subagent({ action: "status", id: "<run-id>", view: "transcript" })
1006
995
  subagent({ action: "steer", id: "<run-id>", message: "Focus only on tests." })
1007
996
  subagent({ action: "stop", id: "<run-id>" })
1008
- subagent({ action: "resume", id: "<run-id>", message: "Continue after this clarification." })
1009
997
  ```
1010
998
 
1011
999
  Output file mode để giảm parent context:
1012
1000
 
1013
1001
  ```text
1014
1002
  /run scout[output=context.md,outputMode=file-only] "Map target area"
1015
- /chain scout[output=context.md,as=context] "Scan" -> planner[reads=context.md] "Plan from {outputs.context}"
1016
1003
  ```
1017
1004
 
1018
1005
  ### Worktree isolation
1019
1006
 
1020
- Chỉ bật writer parallel/worktree khi:
1021
-
1022
- - repo là Git repo;
1023
- - worktree clean;
1024
- - write sets không overlap;
1025
- - parent review/merge outputs.
1007
+ Piagent không bật writer parallel/worktree; parent là writer duy nhất.
1026
1008
 
1027
- Default solo/internal: một `piagent-worker`, parallel read-only reviewers/scouts.
1009
+ Default: parent là writer duy nhất; không worker, không parallel helper.
1028
1010
 
1029
1011
  ### Watchdog
1030
1012
 
@@ -1042,8 +1024,8 @@ Watchdog là optional adversarial reviewer ở cuối turn, không bật mặc
1042
1024
 
1043
1025
  | Command | Dùng để |
1044
1026
  |---|---|
1045
- | `npm install -g --ignore-scripts @piagent/platform@1.5.5` | Cài terminal helper `piagent-*` từ release tag hiện tại. |
1046
- | `pi install git:github.com/Vt-mmm/piagent@v1.5.5` | Install pinned release cho reproducible team setup. |
1027
+ | `npm install -g --ignore-scripts @piagent/platform@1.6.1` | Cài terminal helper `piagent-*` từ release tag hiện tại. |
1028
+ | `pi install git:github.com/Vt-mmm/piagent@v1.6.1` | Install pinned release cho reproducible team setup. |
1047
1029
  | `pi install git:github.com/Vt-mmm/piagent` | Install latest platform package cho máy cá nhân/sandbox. |
1048
1030
  | Cài exact Pi host của release, rồi `npm install -g --ignore-scripts @piagent/platform@X.Y.Z` và `piagent-install --stable` | Full update: đồng bộ host, terminal helper và Pi package. Mỗi release pin một Pi host chính xác; lấy đúng version của release đang cài trong [release/install policy](release-install-policy.md). |
1049
1031
  | Cài exact Pi host ghi trong release cũ, rồi helper `vPREVIOUS` và `piagent-install --stable` | Full rollback; đánh giá lại dependency findings của host cũ trước khi hạ version. |
@@ -1155,7 +1137,7 @@ Mở lại Pi session sau khi install.
1155
1137
  Cài lại terminal helper đúng release rồi kiểm tra `PATH`:
1156
1138
 
1157
1139
  ```bash
1158
- npm install -g --ignore-scripts @piagent/platform@1.5.5
1140
+ npm install -g --ignore-scripts @piagent/platform@1.6.1
1159
1141
  command -v piagent-install
1160
1142
  ```
1161
1143
 
@@ -66,8 +66,8 @@ Adapter mô tả project cho Pi core. Core không biết domain/project cụ th
66
66
  Path policy fields have distinct meanings:
67
67
 
68
68
  - `protectedPaths`: block read/write path-tool access and protect these paths from shell parsing when `shellProtectedPaths` is omitted.
69
- - `readOnlyPaths`: allow `read`/`grep`/`find`/`ls`, but block `write`/`edit` and shell access.
70
- - `shellProtectedPaths`: block shell access only. Do not rely on this field to block write/edit; `profile-doctor` and `team-doctor` warn when a path is present only here.
69
+ - `readOnlyPaths`: allow `read`/`grep`/`find`/`ls`, but block `write`/`edit`/`apply_patch` and shell access.
70
+ - `shellProtectedPaths`: block shell access only. Do not rely on this field to block `write`/`edit`/`apply_patch`; `profile-doctor` and `team-doctor` warn when a path is present only here.
71
71
 
72
72
  Glob trong các field này neo từ project root, nên `backend/**` chỉ khớp backend nằm ở root. Với monorepo, phải liệt kê đúng vị trí thật (`packages/api/**`, `services/*/**`). Không dùng `**/api/**`: nó khớp luôn `packages/web/src/api/` — thư mục HTTP client của frontend — và biến chính phần được phép sửa thành read-only.
73
73
 
@@ -98,6 +98,16 @@ pi install /path/to/piagent
98
98
  Inspector; projection Task Changes / Working Tree / Staged có bounded diff,
99
99
  provenance và enforce protected path; session lease bền, recovery khi restart
100
100
  hoặc owner cũ treo, doctor có repair tường minh.
101
+ - `1.6.1`: patch ổn định runtime và chi phí — transcript chỉ hiện response khi
102
+ durable task đã terminal, Activity phân loại đúng retry/helper failure, review
103
+ repo ngoài chỉ nhận read grant session-scoped cho checkout chính xác, và direct
104
+ subagent bị khóa solo-first với một helper read-only có budget hữu hạn. Benchmark
105
+ thêm gate provider-free/Codex-relative nhưng không đưa ra claim chi phí mới.
106
+ - `1.6.0`: context evidence khóa đúng turn/session/task run; WebUI có canonical
107
+ live state, watchdog và Stop an toàn qua reload/multi-tab; chat chỉ hiện kết
108
+ quả thành công còn retry/lỗi nằm trong Activity; deep-logic benchmark có bài
109
+ temporal billing nhiều package và accounting fail-closed. Release này không
110
+ đưa ra claim token mới vì production benchmark đã được operator dừng.
101
111
  - `1.5.5`: hotfix WebUI command execution — approval nhận đúng opaque tool-call
102
112
  ID, lệnh tìm kiếm không còn bị nhận nhầm thành migration, Stop hủy approval
103
113
  đang chờ, activity chỉ báo running sau khi command thật sự chạy, và restart