@chankov/agent-fleet 2.0.10 → 2.0.11

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 (531) hide show
  1. package/.pi/agent-fleet/scripts/flow.ts +26 -8
  2. package/.pi/agent-fleet/scripts/workflows/lib/agent-phase.ts +56 -12
  3. package/.pi/agent-fleet/scripts/workflows/lib/isolated-snapshot.ts +125 -0
  4. package/.pi/agent-fleet/scripts/workflows/lib/run.ts +2 -2
  5. package/.pi/agent-fleet/scripts/workflows/lib/scout-data-boundary.ts +55 -0
  6. package/.pi/agent-fleet/scripts/workflows/lib/trace.ts +2 -2
  7. package/.pi/agent-fleet/scripts/workflows/wf-scout.ts +3 -1
  8. package/.pi/harnesses/agent-hub/README.md +53 -18
  9. package/.pi/harnesses/agent-hub/acceptance.ts +137 -0
  10. package/.pi/harnesses/agent-hub/assertion-ledger.js +15 -6
  11. package/.pi/harnesses/agent-hub/assist-profile.ts +16 -0
  12. package/.pi/harnesses/agent-hub/bounded-output.ts +26 -0
  13. package/.pi/harnesses/agent-hub/budget-recovery.ts +112 -18
  14. package/.pi/harnesses/agent-hub/capability-packs.ts +4 -1
  15. package/.pi/harnesses/agent-hub/commands/audit.ts +9 -0
  16. package/.pi/harnesses/agent-hub/commands/context.ts +1 -0
  17. package/.pi/harnesses/agent-hub/commands/retry.ts +1 -1
  18. package/.pi/harnesses/agent-hub/config/agents.ts +1 -0
  19. package/.pi/harnesses/agent-hub/config/model-profiles.ts +20 -1
  20. package/.pi/harnesses/agent-hub/context/assertions-artifacts.ts +10 -2
  21. package/.pi/harnesses/agent-hub/context/budgets.ts +2 -0
  22. package/.pi/harnesses/agent-hub/delegate.ts +19 -2
  23. package/.pi/harnesses/agent-hub/deterministic-fs.ts +127 -0
  24. package/.pi/harnesses/agent-hub/diagnostic-probe-extension.ts +26 -0
  25. package/.pi/harnesses/agent-hub/diagnostic-series-budget.ts +178 -0
  26. package/.pi/harnesses/agent-hub/dispatch-native-complete.ts +14 -5
  27. package/.pi/harnesses/agent-hub/dispatch-native-prepare.ts +74 -7
  28. package/.pi/harnesses/agent-hub/dispatch-native-spawn.ts +5 -1
  29. package/.pi/harnesses/agent-hub/dispatch-native-types.ts +15 -0
  30. package/.pi/harnesses/agent-hub/dispatch-native.ts +24 -11
  31. package/.pi/harnesses/agent-hub/filesystem-tool.ts +73 -0
  32. package/.pi/harnesses/agent-hub/helpers.ts +5 -1
  33. package/.pi/harnesses/agent-hub/hub-state-audit.js +1 -0
  34. package/.pi/harnesses/agent-hub/index.ts +145 -14
  35. package/.pi/harnesses/agent-hub/no-progress.ts +194 -64
  36. package/.pi/harnesses/agent-hub/package.json +1 -1
  37. package/.pi/harnesses/agent-hub/policy/work-mode.ts +10 -3
  38. package/.pi/harnesses/agent-hub/process-obligations.ts +57 -0
  39. package/.pi/harnesses/agent-hub/prompts/context.ts +4 -0
  40. package/.pi/harnesses/agent-hub/prompts/fragments.ts +5 -2
  41. package/.pi/harnesses/agent-hub/prompts/system-prompt.ts +1 -1
  42. package/.pi/harnesses/agent-hub/recovery-contract.ts +78 -0
  43. package/.pi/harnesses/agent-hub/research/runtime.ts +1 -0
  44. package/.pi/harnesses/agent-hub/research/spawn-run.ts +20 -5
  45. package/.pi/harnesses/agent-hub/runtime-test-check.ts +54 -0
  46. package/.pi/harnesses/agent-hub/session-audit.ts +161 -0
  47. package/.pi/harnesses/agent-hub/spawn.ts +61 -7
  48. package/.pi/harnesses/agent-hub/task-resume-contract.ts +70 -0
  49. package/.pi/harnesses/agent-hub/tool-catalog-runtime.ts +68 -0
  50. package/.pi/harnesses/agent-hub/tool-catalog-state.ts +123 -0
  51. package/.pi/harnesses/agent-hub/tool-protocol.ts +134 -0
  52. package/.pi/harnesses/agent-hub/tools/action-executors.ts +41 -5
  53. package/.pi/harnesses/agent-hub/tools/context.ts +9 -1
  54. package/.pi/harnesses/agent-hub/tools/dispatch-agent.ts +1 -1
  55. package/.pi/harnesses/agent-hub/tools/dispatch-execution.ts +176 -26
  56. package/.pi/harnesses/agent-hub/tools/run-flow.ts +122 -0
  57. package/.pi/harnesses/agent-hub/tools/set-task-tier.ts +5 -3
  58. package/.pi/harnesses/agent-hub/tools/spawn-research.ts +4 -1
  59. package/.pi/harnesses/agent-hub/tools/verification-contract.ts +5 -2
  60. package/.pi/harnesses/agent-hub/types.ts +2 -0
  61. package/.pi/harnesses/agent-hub/unknown-tool-counter.ts +116 -0
  62. package/.pi/harnesses/agent-hub/work-mode.ts +8 -2
  63. package/.pi/harnesses/agent-hub/write-isolation.ts +187 -0
  64. package/.pi/harnesses/ask-user-remote/runtime-ask.ts +31 -6
  65. package/.pi/harnesses/coms/package.json +1 -1
  66. package/.pi/harnesses/damage-control-continue/index.ts +9 -0
  67. package/.pi/harnesses/damage-control-continue/package.json +1 -1
  68. package/.pi/harnesses/lib/deterministic-handle-path.ts +23 -0
  69. package/.versions/2.0.11/.pi/agent-fleet/hermes/README.md +347 -0
  70. package/.versions/2.0.11/.pi/agent-fleet/hermes/desktop-plugins/agent-fleet-herdr/plugin.js +1291 -0
  71. package/.versions/2.0.11/.pi/agent-fleet/hermes/desktop-plugins/agent-fleet-herdr/presentation.js +672 -0
  72. package/.versions/2.0.11/.pi/agent-fleet/hermes/desktop-plugins/agent-fleet-monitor/plugin.js +19 -0
  73. package/.versions/2.0.11/.pi/agent-fleet/hermes/desktop-plugins/agent-fleet-monitor/state.js +10 -0
  74. package/.versions/2.0.11/.pi/agent-fleet/hermes/gates/gate-o-capability.md +8 -0
  75. package/.versions/2.0.11/.pi/agent-fleet/hermes/gates/gate-o-validator-fixtures.json +1 -0
  76. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/activity.py +773 -0
  77. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/coms_registry.py +134 -0
  78. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/coms_send.py +350 -0
  79. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/herdr_source.py +193 -0
  80. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/manifest.json +4 -0
  81. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/plugin_api.py +444 -0
  82. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/sessions.py +205 -0
  83. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/tasks.py +300 -0
  84. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/watch.py +819 -0
  85. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/plugin.yaml +8 -0
  86. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-monitor/dashboard/adapter.py +139 -0
  87. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-monitor/dashboard/manifest.json +11 -0
  88. package/.versions/2.0.11/.pi/agent-fleet/hermes/plugins/agent-fleet-monitor/dashboard/plugin_api.py +50 -0
  89. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-conductor/SKILL.md +84 -0
  90. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-liaison/SKILL.md +84 -0
  91. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/SKILL.md +76 -0
  92. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/manifest.json +1 -0
  93. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/scripts/watchdog.py +249 -0
  94. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/scripts/watchdog_actions.py +124 -0
  95. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/scripts/watchdog_contract.py +76 -0
  96. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/scripts/watchdog_delivery.py +10 -0
  97. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/scripts/watchdog_engine.py +128 -0
  98. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/scripts/watchdog_judgment.py +41 -0
  99. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/scripts/watchdog_recovery.py +103 -0
  100. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/scripts/watchdog_state.py +163 -0
  101. package/.versions/2.0.11/.pi/agent-fleet/hermes/skills/hub-watchdog/scripts/watchdog_transport.py +138 -0
  102. package/.versions/2.0.11/.pi/agent-fleet/hermes/watchdog-tests/__init__.py +0 -0
  103. package/.versions/2.0.11/.pi/agent-fleet/hooks/coms-stop-hook.mjs +84 -0
  104. package/.versions/2.0.11/.pi/agent-fleet/scripts/claude-code-preflight.ts +70 -0
  105. package/.versions/2.0.11/.pi/agent-fleet/scripts/coms-claude-bridge.ts +411 -0
  106. package/.versions/2.0.11/.pi/agent-fleet/scripts/coms-cli.ts +425 -0
  107. package/.versions/2.0.11/.pi/agent-fleet/scripts/coms-hermes-bridge.ts +413 -0
  108. package/.versions/2.0.11/.pi/agent-fleet/scripts/fleet-codex-client.ts +69 -0
  109. package/.versions/2.0.11/.pi/agent-fleet/scripts/fleet-codex-read.py +148 -0
  110. package/.versions/2.0.11/.pi/agent-fleet/scripts/fleet.ts +216 -0
  111. package/.versions/2.0.11/.pi/agent-fleet/scripts/flow.ts +155 -0
  112. package/.versions/2.0.11/.pi/agent-fleet/scripts/hermes-monitor-capability-runner.ts +25 -0
  113. package/.versions/2.0.11/.pi/agent-fleet/scripts/hermes-monitor-scenario.ts +474 -0
  114. package/.versions/2.0.11/.pi/agent-fleet/scripts/hermes-watchdog-capability-probe.ts +45 -0
  115. package/.versions/2.0.11/.pi/agent-fleet/scripts/install-hermes-plugin.sh +180 -0
  116. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/claude-bridge-core.ts +210 -0
  117. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/coms-envelope.ts +582 -0
  118. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/fake-herdr-server.ts +88 -0
  119. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/fleet-codex-catalog.ts +63 -0
  120. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/fleet-codex-operations.ts +190 -0
  121. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/fleet-codex-session.ts +166 -0
  122. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/fleet-codex-transport.ts +124 -0
  123. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/fleet-command.ts +252 -0
  124. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/flow-command.ts +93 -0
  125. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/flow-maintenance.ts +230 -0
  126. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/herdr-layout.ts +281 -0
  127. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/hermes-bridge-core.ts +241 -0
  128. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/hermes-monitor-events.ts +1 -0
  129. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/hermes-monitor-herdr.ts +2 -0
  130. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/hermes-monitor-model.ts +3 -0
  131. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/hermes-monitor-registry.ts +2 -0
  132. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/hermes-monitor-socket.ts +2 -0
  133. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/hermes-monitor-store.ts +2 -0
  134. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/monitor-env.ts +146 -0
  135. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/peer-launch.ts +318 -0
  136. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/persona-banner.ts +62 -0
  137. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/persona-dirs.ts +25 -0
  138. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/runtime-dependencies.js +124 -0
  139. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/spawn-stagger.ts +101 -0
  140. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/team-project.ts +178 -0
  141. package/.versions/2.0.11/.pi/agent-fleet/scripts/lib/team-snapshot.ts +147 -0
  142. package/.versions/2.0.11/.pi/agent-fleet/scripts/package-lock.json +35 -0
  143. package/.versions/2.0.11/.pi/agent-fleet/scripts/package.json +9 -0
  144. package/.versions/2.0.11/.pi/agent-fleet/scripts/peer-banner.ts +41 -0
  145. package/.versions/2.0.11/.pi/agent-fleet/scripts/peer-launch.ts +283 -0
  146. package/.versions/2.0.11/.pi/agent-fleet/scripts/team-snapshot.ts +237 -0
  147. package/.versions/2.0.11/.pi/agent-fleet/scripts/team-up.ts +378 -0
  148. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/agent-phase.ts +177 -0
  149. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/changes.ts +61 -0
  150. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/debate.ts +149 -0
  151. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/env.ts +34 -0
  152. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/envelopes.ts +169 -0
  153. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/gates.ts +81 -0
  154. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/git.ts +96 -0
  155. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/isolated-snapshot.ts +125 -0
  156. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/merge.ts +95 -0
  157. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/model-profile.ts +22 -0
  158. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/model-visibility.ts +92 -0
  159. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/permissions.ts +130 -0
  160. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/personas.ts +57 -0
  161. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/poll.ts +86 -0
  162. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/quality.ts +64 -0
  163. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/run.ts +108 -0
  164. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/scout-data-boundary.ts +55 -0
  165. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/trace.ts +34 -0
  166. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/voices.ts +128 -0
  167. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/lib/writer-lease.ts +112 -0
  168. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/tsconfig.json +22 -0
  169. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/wf-build-test.ts +61 -0
  170. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/wf-debate.ts +87 -0
  171. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/wf-document.ts +50 -0
  172. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/wf-poll.ts +95 -0
  173. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/wf-quality.ts +20 -0
  174. package/.versions/2.0.11/.pi/agent-fleet/scripts/workflows/wf-scout.ts +40 -0
  175. package/.versions/2.0.11/.pi/agents/dispatch-policy.yaml +32 -0
  176. package/.versions/2.0.11/.pi/agents/model-profiles.yaml +83 -0
  177. package/.versions/2.0.11/.pi/agents/peers.yaml +147 -0
  178. package/.versions/2.0.11/.pi/agents/teams.yaml +41 -0
  179. package/.versions/2.0.11/.pi/agents/voices.yaml +13 -0
  180. package/.versions/2.0.11/.pi/damage-control-rules.yaml +297 -0
  181. package/.versions/2.0.11/.pi/extensions/btw/README.md +120 -0
  182. package/.versions/2.0.11/.pi/extensions/btw/index.ts +914 -0
  183. package/.versions/2.0.11/.pi/extensions/btw/package.json +6 -0
  184. package/.versions/2.0.11/.pi/extensions/btw/state.js +179 -0
  185. package/.versions/2.0.11/.pi/extensions/chrome-devtools-mcp/README.md +55 -0
  186. package/.versions/2.0.11/.pi/extensions/chrome-devtools-mcp/index.ts +93 -0
  187. package/.versions/2.0.11/.pi/extensions/chrome-devtools-mcp/package.json +6 -0
  188. package/.versions/2.0.11/.pi/extensions/compact-and-continue/README.md +42 -0
  189. package/.versions/2.0.11/.pi/extensions/compact-and-continue/index.ts +120 -0
  190. package/.versions/2.0.11/.pi/extensions/compact-and-continue/package.json +6 -0
  191. package/.versions/2.0.11/.pi/extensions/mcp-bridge/README.md +46 -0
  192. package/.versions/2.0.11/.pi/extensions/mcp-bridge/index.ts +206 -0
  193. package/.versions/2.0.11/.pi/extensions/mcp-bridge/package.json +6 -0
  194. package/.versions/2.0.11/.pi/extensions/package-lock.json +1175 -0
  195. package/.versions/2.0.11/.pi/extensions/package.json +9 -0
  196. package/.versions/2.0.11/.pi/extensions/pi-voice-stt/README.md +224 -0
  197. package/.versions/2.0.11/.pi/extensions/pi-voice-stt/config.ts +198 -0
  198. package/.versions/2.0.11/.pi/extensions/pi-voice-stt/examples/stt.json +35 -0
  199. package/.versions/2.0.11/.pi/extensions/pi-voice-stt/index.ts +248 -0
  200. package/.versions/2.0.11/.pi/extensions/pi-voice-stt/package.json +6 -0
  201. package/.versions/2.0.11/.pi/extensions/pi-voice-stt/providers.ts +214 -0
  202. package/.versions/2.0.11/.pi/extensions/pi-voice-stt/recorder.ts +158 -0
  203. package/.versions/2.0.11/.pi/harnesses/agent-hub/README.md +1252 -0
  204. package/.versions/2.0.11/.pi/harnesses/agent-hub/acceptance.ts +199 -0
  205. package/.versions/2.0.11/.pi/harnesses/agent-hub/access-approval.ts +135 -0
  206. package/.versions/2.0.11/.pi/harnesses/agent-hub/artifacts.js +146 -0
  207. package/.versions/2.0.11/.pi/harnesses/agent-hub/assertion-ledger.js +67 -0
  208. package/.versions/2.0.11/.pi/harnesses/agent-hub/assist-profile.ts +16 -0
  209. package/.versions/2.0.11/.pi/harnesses/agent-hub/backend-policy.js +174 -0
  210. package/.versions/2.0.11/.pi/harnesses/agent-hub/bounded-output.ts +26 -0
  211. package/.versions/2.0.11/.pi/harnesses/agent-hub/budget-continuation.ts +13 -0
  212. package/.versions/2.0.11/.pi/harnesses/agent-hub/budget-recovery.ts +191 -0
  213. package/.versions/2.0.11/.pi/harnesses/agent-hub/capability-confirmation.ts +47 -0
  214. package/.versions/2.0.11/.pi/harnesses/agent-hub/capability-packs.ts +183 -0
  215. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agent-model-thinking.ts +10 -0
  216. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agent-model.ts +10 -0
  217. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agent-models-substitute.ts +10 -0
  218. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agents-add.ts +9 -0
  219. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agents-drop.ts +9 -0
  220. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agents-history.ts +9 -0
  221. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agents-kill.ts +10 -0
  222. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agents-list.ts +9 -0
  223. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agents-restart.ts +10 -0
  224. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agents-save.ts +9 -0
  225. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/agents-team.ts +9 -0
  226. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/audit.ts +9 -0
  227. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/budget-continue.ts +10 -0
  228. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/compound.ts +9 -0
  229. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/coms.ts +9 -0
  230. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/context-command.ts +9 -0
  231. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/context.ts +44 -0
  232. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/debate.ts +9 -0
  233. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/dispatch-policy.ts +9 -0
  234. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/handoff.ts +10 -0
  235. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/hub-report.ts +9 -0
  236. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/models.ts +10 -0
  237. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/poll.ts +9 -0
  238. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/retry.ts +10 -0
  239. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/watchdog.ts +9 -0
  240. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/work-mode.ts +29 -0
  241. package/.versions/2.0.11/.pi/harnesses/agent-hub/commands/zoom.ts +10 -0
  242. package/.versions/2.0.11/.pi/harnesses/agent-hub/config/agents.ts +178 -0
  243. package/.versions/2.0.11/.pi/harnesses/agent-hub/config/model-profiles.ts +235 -0
  244. package/.versions/2.0.11/.pi/harnesses/agent-hub/config/overrides.ts +153 -0
  245. package/.versions/2.0.11/.pi/harnesses/agent-hub/context/agent-state.ts +30 -0
  246. package/.versions/2.0.11/.pi/harnesses/agent-hub/context/assertions-artifacts.ts +221 -0
  247. package/.versions/2.0.11/.pi/harnesses/agent-hub/context/budgets.ts +204 -0
  248. package/.versions/2.0.11/.pi/harnesses/agent-hub/context/hub-state.ts +40 -0
  249. package/.versions/2.0.11/.pi/harnesses/agent-hub/context-budget-snapshot.ts +105 -0
  250. package/.versions/2.0.11/.pi/harnesses/agent-hub/context-pressure.ts +191 -0
  251. package/.versions/2.0.11/.pi/harnesses/agent-hub/context-window.js +106 -0
  252. package/.versions/2.0.11/.pi/harnesses/agent-hub/debate-command.ts +176 -0
  253. package/.versions/2.0.11/.pi/harnesses/agent-hub/delegate.ts +463 -0
  254. package/.versions/2.0.11/.pi/harnesses/agent-hub/deterministic-fs.ts +127 -0
  255. package/.versions/2.0.11/.pi/harnesses/agent-hub/diagnostic-probe-extension.ts +26 -0
  256. package/.versions/2.0.11/.pi/harnesses/agent-hub/diagnostic-series-budget.ts +178 -0
  257. package/.versions/2.0.11/.pi/harnesses/agent-hub/dispatch-coms.ts +237 -0
  258. package/.versions/2.0.11/.pi/harnesses/agent-hub/dispatch-core.ts +27 -0
  259. package/.versions/2.0.11/.pi/harnesses/agent-hub/dispatch-native-complete.ts +163 -0
  260. package/.versions/2.0.11/.pi/harnesses/agent-hub/dispatch-native-prepare.ts +223 -0
  261. package/.versions/2.0.11/.pi/harnesses/agent-hub/dispatch-native-spawn.ts +197 -0
  262. package/.versions/2.0.11/.pi/harnesses/agent-hub/dispatch-native-types.ts +245 -0
  263. package/.versions/2.0.11/.pi/harnesses/agent-hub/dispatch-native.ts +226 -0
  264. package/.versions/2.0.11/.pi/harnesses/agent-hub/dispatch-observability.ts +225 -0
  265. package/.versions/2.0.11/.pi/harnesses/agent-hub/docs-lane.js +122 -0
  266. package/.versions/2.0.11/.pi/harnesses/agent-hub/drift-watchdog.js +218 -0
  267. package/.versions/2.0.11/.pi/harnesses/agent-hub/evidence-rules.js +96 -0
  268. package/.versions/2.0.11/.pi/harnesses/agent-hub/execution-evidence.ts +80 -0
  269. package/.versions/2.0.11/.pi/harnesses/agent-hub/external-blocker.js +85 -0
  270. package/.versions/2.0.11/.pi/harnesses/agent-hub/filesystem-tool.ts +73 -0
  271. package/.versions/2.0.11/.pi/harnesses/agent-hub/fixtures/registration-surface.json +60 -0
  272. package/.versions/2.0.11/.pi/harnesses/agent-hub/fixtures/t0a-synthetic/catalog.ts +189 -0
  273. package/.versions/2.0.11/.pi/harnesses/agent-hub/footer.ts +30 -0
  274. package/.versions/2.0.11/.pi/harnesses/agent-hub/helpers.ts +270 -0
  275. package/.versions/2.0.11/.pi/harnesses/agent-hub/hub-state-audit.js +65 -0
  276. package/.versions/2.0.11/.pi/harnesses/agent-hub/index.ts +2263 -0
  277. package/.versions/2.0.11/.pi/harnesses/agent-hub/input/completions.ts +76 -0
  278. package/.versions/2.0.11/.pi/harnesses/agent-hub/input/shortcuts.ts +20 -0
  279. package/.versions/2.0.11/.pi/harnesses/agent-hub/lifecycle/context-pressure.ts +123 -0
  280. package/.versions/2.0.11/.pi/harnesses/agent-hub/lifecycle/monitor-session.ts +104 -0
  281. package/.versions/2.0.11/.pi/harnesses/agent-hub/lifecycle/session-orchestration.ts +143 -0
  282. package/.versions/2.0.11/.pi/harnesses/agent-hub/lifecycle/turn-handlers.ts +74 -0
  283. package/.versions/2.0.11/.pi/harnesses/agent-hub/monitor-control.ts +9 -0
  284. package/.versions/2.0.11/.pi/harnesses/agent-hub/monitor-invoke-journal.ts +131 -0
  285. package/.versions/2.0.11/.pi/harnesses/agent-hub/monitor-invoke.ts +92 -0
  286. package/.versions/2.0.11/.pi/harnesses/agent-hub/monitor-lifecycle.ts +13 -0
  287. package/.versions/2.0.11/.pi/harnesses/agent-hub/monitor-publisher.ts +59 -0
  288. package/.versions/2.0.11/.pi/harnesses/agent-hub/monitor-recovery.ts +3 -0
  289. package/.versions/2.0.11/.pi/harnesses/agent-hub/monitor-runtime.ts +3 -0
  290. package/.versions/2.0.11/.pi/harnesses/agent-hub/monitor-session-bridge.ts +266 -0
  291. package/.versions/2.0.11/.pi/harnesses/agent-hub/no-progress.ts +206 -0
  292. package/.versions/2.0.11/.pi/harnesses/agent-hub/package.json +7 -0
  293. package/.versions/2.0.11/.pi/harnesses/agent-hub/peer-spawn-plan.ts +116 -0
  294. package/.versions/2.0.11/.pi/harnesses/agent-hub/policy/models.ts +153 -0
  295. package/.versions/2.0.11/.pi/harnesses/agent-hub/policy/profile-activation.ts +62 -0
  296. package/.versions/2.0.11/.pi/harnesses/agent-hub/policy/profile-runtime.ts +120 -0
  297. package/.versions/2.0.11/.pi/harnesses/agent-hub/policy/roster.ts +94 -0
  298. package/.versions/2.0.11/.pi/harnesses/agent-hub/policy/work-mode.ts +126 -0
  299. package/.versions/2.0.11/.pi/harnesses/agent-hub/poll-command.ts +176 -0
  300. package/.versions/2.0.11/.pi/harnesses/agent-hub/presentation.ts +48 -0
  301. package/.versions/2.0.11/.pi/harnesses/agent-hub/process-obligations.ts +57 -0
  302. package/.versions/2.0.11/.pi/harnesses/agent-hub/prompts/context.ts +51 -0
  303. package/.versions/2.0.11/.pi/harnesses/agent-hub/prompts/fragments.ts +118 -0
  304. package/.versions/2.0.11/.pi/harnesses/agent-hub/prompts/session-start.ts +90 -0
  305. package/.versions/2.0.11/.pi/harnesses/agent-hub/prompts/system-prompt.ts +112 -0
  306. package/.versions/2.0.11/.pi/harnesses/agent-hub/provider-semaphore.js +138 -0
  307. package/.versions/2.0.11/.pi/harnesses/agent-hub/recovery-contract.ts +78 -0
  308. package/.versions/2.0.11/.pi/harnesses/agent-hub/research/controls.ts +74 -0
  309. package/.versions/2.0.11/.pi/harnesses/agent-hub/research/runtime.ts +193 -0
  310. package/.versions/2.0.11/.pi/harnesses/agent-hub/research/spawn-run.ts +171 -0
  311. package/.versions/2.0.11/.pi/harnesses/agent-hub/research-watchdog.ts +40 -0
  312. package/.versions/2.0.11/.pi/harnesses/agent-hub/return-contract.js +327 -0
  313. package/.versions/2.0.11/.pi/harnesses/agent-hub/return-extract.js +75 -0
  314. package/.versions/2.0.11/.pi/harnesses/agent-hub/review-findings.js +102 -0
  315. package/.versions/2.0.11/.pi/harnesses/agent-hub/run-budget.js +537 -0
  316. package/.versions/2.0.11/.pi/harnesses/agent-hub/run-namespace.js +87 -0
  317. package/.versions/2.0.11/.pi/harnesses/agent-hub/runtime-test-check.ts +54 -0
  318. package/.versions/2.0.11/.pi/harnesses/agent-hub/safety-routing.ts +22 -0
  319. package/.versions/2.0.11/.pi/harnesses/agent-hub/scope-gate.js +123 -0
  320. package/.versions/2.0.11/.pi/harnesses/agent-hub/session-audit.ts +161 -0
  321. package/.versions/2.0.11/.pi/harnesses/agent-hub/session-health.js +110 -0
  322. package/.versions/2.0.11/.pi/harnesses/agent-hub/session-start.ts +42 -0
  323. package/.versions/2.0.11/.pi/harnesses/agent-hub/spawn.ts +458 -0
  324. package/.versions/2.0.11/.pi/harnesses/agent-hub/task-resume-contract.ts +70 -0
  325. package/.versions/2.0.11/.pi/harnesses/agent-hub/timeline.ts +54 -0
  326. package/.versions/2.0.11/.pi/harnesses/agent-hub/tool-catalog-runtime.ts +68 -0
  327. package/.versions/2.0.11/.pi/harnesses/agent-hub/tool-catalog-state.ts +123 -0
  328. package/.versions/2.0.11/.pi/harnesses/agent-hub/tool-protocol.ts +134 -0
  329. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/action-executors.ts +185 -0
  330. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/coms-tools.ts +139 -0
  331. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/context.ts +142 -0
  332. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/dispatch-agent.ts +82 -0
  333. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/dispatch-execution.ts +485 -0
  334. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/execution-orchestration.ts +22 -0
  335. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/fleet-tools.ts +200 -0
  336. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/herdr-executors.ts +86 -0
  337. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/run-flow.ts +122 -0
  338. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/set-task-tier.ts +31 -0
  339. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/spawn-research.ts +65 -0
  340. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/team-adjust.ts +29 -0
  341. package/.versions/2.0.11/.pi/harnesses/agent-hub/tools/verification-contract.ts +80 -0
  342. package/.versions/2.0.11/.pi/harnesses/agent-hub/types.ts +65 -0
  343. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/activity-dots.ts +16 -0
  344. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/context-budget.ts +86 -0
  345. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/detail-panel.ts +178 -0
  346. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/fleet-actions.ts +76 -0
  347. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/fleet-dashboard.ts +99 -0
  348. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/fleet-editor.ts +44 -0
  349. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/fleet-source.ts +157 -0
  350. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/grid.ts +154 -0
  351. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/history-store.ts +137 -0
  352. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/history.ts +234 -0
  353. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/pool.ts +103 -0
  354. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/running-strip.ts +2 -0
  355. package/.versions/2.0.11/.pi/harnesses/agent-hub/ui/zoom.ts +204 -0
  356. package/.versions/2.0.11/.pi/harnesses/agent-hub/unknown-tool-counter.ts +116 -0
  357. package/.versions/2.0.11/.pi/harnesses/agent-hub/version.ts +42 -0
  358. package/.versions/2.0.11/.pi/harnesses/agent-hub/work-mode-controls.ts +56 -0
  359. package/.versions/2.0.11/.pi/harnesses/agent-hub/work-mode.ts +173 -0
  360. package/.versions/2.0.11/.pi/harnesses/agent-hub/write-isolation.ts +187 -0
  361. package/.versions/2.0.11/.pi/harnesses/ask-user-remote/README.md +59 -0
  362. package/.versions/2.0.11/.pi/harnesses/ask-user-remote/index.ts +558 -0
  363. package/.versions/2.0.11/.pi/harnesses/ask-user-remote/questions.ts +161 -0
  364. package/.versions/2.0.11/.pi/harnesses/ask-user-remote/race-core.js +124 -0
  365. package/.versions/2.0.11/.pi/harnesses/ask-user-remote/runtime-ask.ts +68 -0
  366. package/.versions/2.0.11/.pi/harnesses/coms/README.md +100 -0
  367. package/.versions/2.0.11/.pi/harnesses/coms/index.ts +282 -0
  368. package/.versions/2.0.11/.pi/harnesses/coms/package.json +7 -0
  369. package/.versions/2.0.11/.pi/harnesses/coms/version.ts +42 -0
  370. package/.versions/2.0.11/.pi/harnesses/damage-control-continue/README.md +135 -0
  371. package/.versions/2.0.11/.pi/harnesses/damage-control-continue/index.ts +550 -0
  372. package/.versions/2.0.11/.pi/harnesses/damage-control-continue/package.json +7 -0
  373. package/.versions/2.0.11/.pi/harnesses/damage-control-continue/version.ts +42 -0
  374. package/.versions/2.0.11/.pi/harnesses/lib/changed-file-diagnostics.ts +642 -0
  375. package/.versions/2.0.11/.pi/harnesses/lib/coms-core-io.ts +385 -0
  376. package/.versions/2.0.11/.pi/harnesses/lib/coms-core.ts +577 -0
  377. package/.versions/2.0.11/.pi/harnesses/lib/coms-registry-entry.ts +94 -0
  378. package/.versions/2.0.11/.pi/harnesses/lib/context-budget-child-prompt.ts +180 -0
  379. package/.versions/2.0.11/.pi/harnesses/lib/context-budget-hub-prompt.ts +114 -0
  380. package/.versions/2.0.11/.pi/harnesses/lib/context-budget-view.ts +107 -0
  381. package/.versions/2.0.11/.pi/harnesses/lib/context-budget.ts +114 -0
  382. package/.versions/2.0.11/.pi/harnesses/lib/damage-control-shared.ts +197 -0
  383. package/.versions/2.0.11/.pi/harnesses/lib/deterministic-handle-path.ts +23 -0
  384. package/.versions/2.0.11/.pi/harnesses/lib/fixtures/context-budget-profiles.json +11 -0
  385. package/.versions/2.0.11/.pi/harnesses/lib/fixtures/hub-prompt-operator.legacy.txt +73 -0
  386. package/.versions/2.0.11/.pi/harnesses/lib/fixtures/hub-prompt-orchestrator.legacy.txt +73 -0
  387. package/.versions/2.0.11/.pi/harnesses/lib/fleet-dashboard-ops.ts +178 -0
  388. package/.versions/2.0.11/.pi/harnesses/lib/fleet-dashboard-view.ts +134 -0
  389. package/.versions/2.0.11/.pi/harnesses/lib/fleet-detail-view.ts +247 -0
  390. package/.versions/2.0.11/.pi/harnesses/lib/fleet-overlay.ts +26 -0
  391. package/.versions/2.0.11/.pi/harnesses/lib/fleet-panel.ts +34 -0
  392. package/.versions/2.0.11/.pi/harnesses/lib/fleet-read-model.ts +213 -0
  393. package/.versions/2.0.11/.pi/harnesses/lib/fleet-selection.ts +24 -0
  394. package/.versions/2.0.11/.pi/harnesses/lib/fleet-strip-controller.ts +74 -0
  395. package/.versions/2.0.11/.pi/harnesses/lib/fleet-strip-view.ts +132 -0
  396. package/.versions/2.0.11/.pi/harnesses/lib/fleet-transcript-store.ts +244 -0
  397. package/.versions/2.0.11/.pi/harnesses/lib/herdr-client.ts +472 -0
  398. package/.versions/2.0.11/.pi/harnesses/lib/herdr-presence.ts +399 -0
  399. package/.versions/2.0.11/.pi/harnesses/lib/hermes-monitor-events.ts +136 -0
  400. package/.versions/2.0.11/.pi/harnesses/lib/hermes-monitor-herdr.ts +28 -0
  401. package/.versions/2.0.11/.pi/harnesses/lib/hermes-monitor-model.ts +18 -0
  402. package/.versions/2.0.11/.pi/harnesses/lib/hermes-monitor-registry.ts +6 -0
  403. package/.versions/2.0.11/.pi/harnesses/lib/hermes-monitor-socket.ts +163 -0
  404. package/.versions/2.0.11/.pi/harnesses/lib/hermes-monitor-store.ts +31 -0
  405. package/.versions/2.0.11/.pi/harnesses/lib/spawned-peers.js +190 -0
  406. package/.versions/2.0.11/.pi/harnesses/lib/update-check.ts +222 -0
  407. package/.versions/2.0.11/.pi/harnesses/package-lock.json +2039 -0
  408. package/.versions/2.0.11/.pi/harnesses/package.json +10 -0
  409. package/.versions/2.0.11/.pi/prompts/af-build.md +24 -0
  410. package/.versions/2.0.11/.pi/prompts/af-code-simplify.md +22 -0
  411. package/.versions/2.0.11/.pi/prompts/af-constraints.md +31 -0
  412. package/.versions/2.0.11/.pi/prompts/af-plan.md +17 -0
  413. package/.versions/2.0.11/.pi/prompts/af-review.md +16 -0
  414. package/.versions/2.0.11/.pi/prompts/af-set-hermes-telegram.md +15 -0
  415. package/.versions/2.0.11/.pi/prompts/af-set-hermes-watchdog.md +6 -0
  416. package/.versions/2.0.11/.pi/prompts/af-ship.md +17 -0
  417. package/.versions/2.0.11/.pi/prompts/af-spec.md +15 -0
  418. package/.versions/2.0.11/.pi/prompts/af-test.md +19 -0
  419. package/.versions/2.0.11/.pi/skills/bowser/SKILL.md +145 -0
  420. package/.versions/2.0.11/.version +1 -0
  421. package/.versions/2.0.11/agents/architect.md +25 -0
  422. package/.versions/2.0.11/agents/bowser.md +28 -0
  423. package/.versions/2.0.11/agents/builder.md +56 -0
  424. package/.versions/2.0.11/agents/code-reviewer.md +210 -0
  425. package/.versions/2.0.11/agents/deep-researcher.md +33 -0
  426. package/.versions/2.0.11/agents/documenter.md +23 -0
  427. package/.versions/2.0.11/agents/orchestrator.md +23 -0
  428. package/.versions/2.0.11/agents/plan-reviewer.md +84 -0
  429. package/.versions/2.0.11/agents/planner.md +106 -0
  430. package/.versions/2.0.11/agents/releaser.md +25 -0
  431. package/.versions/2.0.11/agents/researcher.md +31 -0
  432. package/.versions/2.0.11/agents/security-auditor.md +168 -0
  433. package/.versions/2.0.11/agents/test-engineer.md +154 -0
  434. package/.versions/2.0.11/agents/web-debugger.md +44 -0
  435. package/.versions/2.0.11/agents/web-performance-auditor.md +194 -0
  436. package/.versions/2.0.11/bin/catalog/harness-runtime-closure.json +61 -0
  437. package/.versions/2.0.11/codex/fleet-session-client/SKILL.md +34 -0
  438. package/.versions/2.0.11/codex/fleet-session-client.zip +0 -0
  439. package/.versions/2.0.11/docs/ARCHITECTURE.md +595 -0
  440. package/.versions/2.0.11/docs/agent-fleet-setup.md +586 -0
  441. package/.versions/2.0.11/docs/claude-code-coms-bridge.md +134 -0
  442. package/.versions/2.0.11/docs/codex-session-bridge.md +219 -0
  443. package/.versions/2.0.11/docs/coms-hermes-bridge.md +188 -0
  444. package/.versions/2.0.11/docs/getting-started.md +126 -0
  445. package/.versions/2.0.11/docs/npm-install.md +383 -0
  446. package/.versions/2.0.11/docs/pi-extensions.md +449 -0
  447. package/.versions/2.0.11/docs/pi-setup.md +92 -0
  448. package/.versions/2.0.11/docs/skill-anatomy.md +150 -0
  449. package/.versions/2.0.11/install-manifest.json +3020 -0
  450. package/.versions/2.0.11/justfile +247 -0
  451. package/.versions/2.0.11/manifest-meta.json +970 -0
  452. package/.versions/2.0.11/references/accessibility-checklist.md +160 -0
  453. package/.versions/2.0.11/references/communication-contract.md +26 -0
  454. package/.versions/2.0.11/references/definition-of-done.md +67 -0
  455. package/.versions/2.0.11/references/fleet-coordination-patterns.md +171 -0
  456. package/.versions/2.0.11/references/observability-checklist.md +92 -0
  457. package/.versions/2.0.11/references/orchestration-patterns.md +370 -0
  458. package/.versions/2.0.11/references/performance-checklist.md +236 -0
  459. package/.versions/2.0.11/references/prompting-patterns.md +380 -0
  460. package/.versions/2.0.11/references/security-checklist.md +245 -0
  461. package/.versions/2.0.11/references/testing-patterns.md +235 -0
  462. package/.versions/2.0.11/references/workflow-authoring.md +304 -0
  463. package/.versions/2.0.11/skills/_internal/grilling.md +59 -0
  464. package/.versions/2.0.11/skills/browser-testing-with-devtools/SKILL.md +351 -0
  465. package/.versions/2.0.11/skills/code-review-and-quality/SKILL.md +396 -0
  466. package/.versions/2.0.11/skills/compound-learning/SKILL.md +137 -0
  467. package/.versions/2.0.11/skills/constraint-driven-development/SKILL.md +338 -0
  468. package/.versions/2.0.11/skills/constraint-driven-development/references/floor-guard.md +99 -0
  469. package/.versions/2.0.11/skills/context-engineering/SKILL.md +341 -0
  470. package/.versions/2.0.11/skills/deprecation-and-migration/SKILL.md +247 -0
  471. package/.versions/2.0.11/skills/designing-agents/SKILL.md +394 -0
  472. package/.versions/2.0.11/skills/designing-agents/pi-harness-authoring.md +222 -0
  473. package/.versions/2.0.11/skills/drafting-workflows/SKILL.md +143 -0
  474. package/.versions/2.0.11/skills/drafting-workflows/scripts/make-workflow.ts +177 -0
  475. package/.versions/2.0.11/skills/frontend-ui-engineering/SKILL.md +328 -0
  476. package/.versions/2.0.11/skills/git-workflow-and-versioning/SKILL.md +316 -0
  477. package/.versions/2.0.11/skills/idea-refine/SKILL.md +182 -0
  478. package/.versions/2.0.11/skills/idea-refine/examples.md +238 -0
  479. package/.versions/2.0.11/skills/idea-refine/frameworks.md +99 -0
  480. package/.versions/2.0.11/skills/idea-refine/refinement-criteria.md +113 -0
  481. package/.versions/2.0.11/skills/idea-refine/scripts/idea-refine.sh +15 -0
  482. package/.versions/2.0.11/skills/incremental-implementation/SKILL.md +302 -0
  483. package/.versions/2.0.11/skills/interview-me/SKILL.md +226 -0
  484. package/.versions/2.0.11/skills/orchestration-verification/SKILL.md +182 -0
  485. package/.versions/2.0.11/skills/peer-coms/SKILL.md +90 -0
  486. package/.versions/2.0.11/skills/performance-optimization/SKILL.md +496 -0
  487. package/.versions/2.0.11/skills/planning-and-task-breakdown/SKILL.md +289 -0
  488. package/.versions/2.0.11/skills/security-and-hardening/SKILL.md +524 -0
  489. package/.versions/2.0.11/skills/spec-driven-development/SKILL.md +295 -0
  490. package/.versions/2.0.11/skills/using-agent-skills/SKILL.md +198 -0
  491. package/.versions/2.0.11/vendor/agent-skills-upstream/LICENSE +21 -0
  492. package/.versions/2.0.11/vendor/agent-skills-upstream/README.md +430 -0
  493. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/api-and-interface-design/SKILL.md +367 -0
  494. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/browser-testing-with-devtools/SKILL.md +317 -0
  495. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/ci-cd-and-automation/SKILL.md +390 -0
  496. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/code-review-and-quality/SKILL.md +396 -0
  497. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/code-simplification/SKILL.md +331 -0
  498. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/constraint-driven-development/SKILL.md +311 -0
  499. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/constraint-driven-development/references/floor-guard.md +99 -0
  500. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/context-engineering/SKILL.md +337 -0
  501. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/debugging-and-error-recovery/SKILL.md +300 -0
  502. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/deprecation-and-migration/SKILL.md +247 -0
  503. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/documentation-and-adrs/SKILL.md +288 -0
  504. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/doubt-driven-development/SKILL.md +243 -0
  505. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/frontend-ui-engineering/SKILL.md +328 -0
  506. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/git-workflow-and-versioning/SKILL.md +355 -0
  507. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/idea-refine/SKILL.md +178 -0
  508. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/idea-refine/examples.md +238 -0
  509. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/idea-refine/frameworks.md +99 -0
  510. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/idea-refine/refinement-criteria.md +113 -0
  511. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/idea-refine/scripts/idea-refine.sh +15 -0
  512. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/incremental-implementation/SKILL.md +249 -0
  513. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/interview-me/SKILL.md +225 -0
  514. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/observability-and-instrumentation/SKILL.md +238 -0
  515. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/performance-optimization/SKILL.md +496 -0
  516. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/planning-and-task-breakdown/SKILL.md +257 -0
  517. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/security-and-hardening/SKILL.md +524 -0
  518. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/shipping-and-launch/SKILL.md +330 -0
  519. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/source-driven-development/SKILL.md +216 -0
  520. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/spec-driven-development/SKILL.md +245 -0
  521. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/test-driven-development/SKILL.md +398 -0
  522. package/.versions/2.0.11/vendor/agent-skills-upstream/skills/using-agent-skills/SKILL.md +192 -0
  523. package/CHANGELOG.md +8 -0
  524. package/README.md +1 -0
  525. package/bin/catalog/harness-runtime-closure.json +29 -0
  526. package/bin/cli.js +15 -4
  527. package/bin/lib/doctor.js +83 -1
  528. package/docs/ARCHITECTURE.md +181 -2
  529. package/docs/workflows.md +8 -2
  530. package/install-manifest.json +31 -2
  531. package/package.json +2 -2
@@ -0,0 +1,595 @@
1
+ # Agent Fleet Architecture
2
+
3
+ Agent Fleet is a Pi-centered multi-agent orchestration system. This page maps
4
+ the runtime responsibilities and where each module lives in the repository.
5
+
6
+ For the principles governing future plans and design decisions, read the
7
+ [project philosophy](PHILOSOPHY.md). That document states the intended direction;
8
+ this page describes the current architecture.
9
+
10
+ ## Runtime layers
11
+
12
+ | Layer | Role | Implementation |
13
+ | --- | --- | --- |
14
+ | **Pi Coding Agent** | Primary local runtime — runs the dispatcher and specialist subagents | `.pi/harnesses/`, `.pi/extensions/`, `.pi/agents/`, `.pi/prompts/` |
15
+ | **agent-hub** | Thin-context multi-agent harness: dispatcher + specialists + research helpers + Verification Contract | `.pi/harnesses/agent-hub/` |
16
+ | **Workflows** | Headless, code-owned phase order, retries, gates, writes enforcement, and acceptance through `just flow` | `.pi/agent-fleet/scripts/flow.ts`, `.pi/agent-fleet/scripts/workflows/` — see [workflows.md](workflows.md) |
17
+ | **Herdr** | Fleet/workspace control plane — spawns peer teams as tiled workspaces, presence via push events, snapshot/resume | [herdr.dev](https://herdr.dev); client in `.pi/harnesses/lib/herdr-client.ts`, layout in `.pi/agent-fleet/scripts/lib/herdr-layout.ts` |
18
+ | **coms** | Peer communication protocol/data plane — envelope-based P2P messaging between agents | Shared peer lifecycle in `.pi/harnesses/lib/coms-core.ts` and `coms-core-io.ts`; Pi registrations in `.pi/harnesses/coms/` and `agent-hub/`; wire helpers in `.pi/agent-fleet/scripts/lib/coms-envelope.ts` and `.pi/agent-fleet/scripts/coms-cli.ts` |
19
+ | **Claude Code bridge** | Makes an interactive Claude Code pane a bidirectional coms peer | `.pi/agent-fleet/scripts/coms-claude-bridge.ts`, `.pi/agent-fleet/hooks/coms-stop-hook.mjs`, `skills/peer-coms/` — see [claude-code-coms-bridge.md](claude-code-coms-bridge.md) |
20
+ | **Hermes bridge** | Remote human control — relays hub questions to Telegram, races phone vs. local answers, conductor/liaison skills | `.pi/agent-fleet/scripts/coms-hermes-bridge.ts`, `.pi/harnesses/ask-user-remote/`, `.pi/agent-fleet/hermes/skills/` — see [coms-hermes-bridge.md](coms-hermes-bridge.md) |
21
+ | **Hermes local monitor transport** | Local, authenticated monitor contract for Hub-owned task generations; consumers supply their own presentation | `.pi/harnesses/agent-hub/monitor-*.ts`, `.pi/harnesses/lib/hermes-monitor-{model,store,registry,socket}.ts` (with compatibility re-exports under `.pi/agent-fleet/scripts/lib/`) — see [Hermes artifacts](../.pi/agent-fleet/hermes/README.md#local-agent-hub-monitor-integration) and [watchdog limits](hermes-watchdog-supervisor.md) |
22
+ | **Hermes Desktop plugin** (`agent-fleet-herdr`) | Fleet observability surface — read-only panel of every live session joined from the coms registry, herdr presence, agent transcripts, and the monitor transport; `focus` and subagent `cancel` are its only write doors | `.pi/agent-fleet/hermes/desktop-plugins/agent-fleet-herdr/` (Electron pane), `.pi/agent-fleet/hermes/plugins/agent-fleet-herdr/dashboard/` (FastAPI backend), installed by `.pi/agent-fleet/scripts/install-hermes-plugin.sh` — see [hermes-desktop-plugins.md](hermes-desktop-plugins.md) |
23
+ | **ChatGPT Fleet session client** | Experimental ChatGPT-initiated client for an existing Pi session; no daemon or idle wake | `.pi/agent-fleet/scripts/fleet-codex-client.ts`, `.pi/agent-fleet/scripts/lib/fleet-codex-*`, opt-in feature `chatgpt-client` — see [codex-session-bridge.md](codex-session-bridge.md) |
24
+ | **Skill library** | Lifecycle workflows and quality gates every agent follows | `skills/` (native) + `vendor/agent-skills-upstream/skills/` (vendored) — see [UPSTREAM-SKILLS.md](UPSTREAM-SKILLS.md) |
25
+ | **Personas** | Reusable specialist definitions, installed verbatim | `agents/` in the package → `.pi/agents/personas/` in a workspace; `bin/lib/personas.js` |
26
+
27
+ ## One Hub runtime, independent axes
28
+
29
+ Every public Pi launch through `just fleet` loads Fleet Core and the `agent-hub`
30
+ harness. Three independent choices shape that runtime:
31
+
32
+ | Axis | Choices | What it controls |
33
+ | --- | --- | --- |
34
+ | **Work Mode** | `operator`, `orchestrator` | Whether the main agent may use direct coding tools. Operator retains the captured `read`/`bash`/`edit`/`write` and approved extension surface; orchestrator removes direct coding tools and keeps dispatch, research, assertions, `ask_user`, and available coms/Herdr tools. `/af-work-mode` and **Alt+M** switch this axis. |
35
+ | **Native roster** | empty or one entry from `.pi/agents/teams.yaml` | Which in-process Pi specialist personas `dispatch_agent` may start. Roster changes do not change work mode in a live session. |
36
+ | **Peer topology** | current terminal, Hub-only Herdr workspace, or Hub plus a `.pi/agents/peers.yaml` preset | Which separate, addressable Pi/Claude processes occupy sibling panes. Starting or spawning peers does not change work mode or the native roster. |
37
+
38
+ Bare `just fleet` therefore means operator work mode, an empty native roster, and
39
+ the current terminal. At startup an explicit `--work-mode` wins; otherwise
40
+ `--agents <roster>` implies orchestrator; otherwise operator is used. An
41
+ orchestrator startup requires a roster. `/af-work-mode` and **Alt+M** switch
42
+ only work mode in the live session, and `/af-agents-*` changes
43
+ only the native roster. Dispatch, research, nested delegation, and Verification
44
+ Contract rigor follow the **task tier** (`trivial`/`small`/`feature`/`project`),
45
+ not a session execution mode.
46
+
47
+ All Hub-owned slash commands remain registered in both work modes. Capability packs are resolved automatically from explicit task intent, work mode, tier, pending work, and compaction state—there is no activation command. Runtime **readiness** (a coms or Herdr connection) is not model-visible capability: ready-but-unrequested peer/workspace packs remain inactive. Ambiguous fleet, peer, and workspace requests are provisionally visible but require one `ask_user` confirmation before their first side effect; a rejection removes the provisional pack. Task packs persist across follow-ups and reset only through `set_task_tier(new_task: true)`, while mandatory work mode and pending-operation leases remain.
48
+
49
+ `/af-context` is a standalone, read-only full-screen budget diagnostic; it separates stable prompt cost, volatile state, active schemas, and zero-cost inactive/loaded-excluded inputs. Provider totals and cache read/write fields are authoritative where Pi supplies them; it never fabricates provider usage or combines capacity percentages across Hub, child, and peer planes. `/af-audit` is a separate read-only session audit over existing runtime-owned records: it reports allowlisted T1 recovery categories, T2 result states, T3 protocol failures, and T8 permission correlations while keeping root, child, compaction snapshot, and repeated-record identities distinct. It never copies prompts, model text, tool payloads, environment values, or transcript content, and unavailable evidence stays explicit rather than becoming a zero count. Managed specialists and research children use explicit replacement prompts/context manifests with selected persona, policy, and skill paths instead of inherited global skills/context files. Runtime gates control whether an action can proceed: without coms, Herdr, a visible target, or an active roster, the corresponding command refuses with an actionable message rather than disappearing. `--browser` and `--all-extensions` expand the captured operator surface only when those optional extensions are installed.
50
+
51
+ T12 diagnostics have been withdrawn by user decision; there is no command, runner, or replacement entry point. Historical results and accounting remain unchanged. `diagnostic-series-budget.ts` retains the shared v2 ledger, stage/per-unit ceilings, conservative input preflight, migration refusal for ambiguous history, ownership locks, and cancellation fences used by benchmark consumers. `diagnostic-probe-extension.ts` is now a side-effect-free compatibility library exposing only provider/API output-cap mapping, its types, and the transport refusal exit constant. It preserves the reviewed 2048 cap and rejects unsupported payload semantics; it does not register an extension or tool. Existing sessions must be reloaded after updating. The installer retires unchanged managed command/runner files; user-modified obsolete files are preserved and reported for manual review, never deleted automatically. Such retained files are not imported by the current Hub.
52
+
53
+ ### Context pressure and recovery
54
+
55
+ Standing prompt ceilings constrain the repeatable replacement prompt and active schemas; they do not
56
+ bound conversation messages or results accumulated inside a long tool loop. The Hub therefore has a
57
+ second, runtime pressure guard. It samples provider usage, projects a finalized tool result at
58
+ `message_end`, warns and exposes transient compaction support at **80%**, then at **90%** aborts the
59
+ pre-model continuation and waits for `agent_settled` before requesting one Pi compaction. This
60
+ ordering ensures the result is persisted before summarization and prevents an ordinary provider call
61
+ from escaping between pressure detection and recovery.
62
+
63
+ Recovery is single-flight. High-context startup or concurrent input is retained in memory and
64
+ replayed exactly once after success; failure retains it for `/compact` or a larger-context model.
65
+ `/af-context` and the status line report phase, usage, thresholds, episode, and last outcome. Session
66
+ entries store only numeric/enumerated pressure metadata, never prompts, tool payloads, credentials,
67
+ error bodies, or summaries.
68
+
69
+ A persisted native roster is also metadata, not copied configuration: only its team name is stored
70
+ and it is re-resolved against current teams and persona definitions. A stale or absent roster in a
71
+ resumed orchestrator session fails closed—work mode remains orchestrator, direct tools remain removed,
72
+ and model input is blocked until `/af-agents-team`, an explicit `--agent-team <name>` restart, or an
73
+ explicit operator selection resolves it. Explicit CLI work mode and roster selections retain
74
+ precedence.
75
+
76
+ Pi's session JSONL is the authoritative append-only record for rebuilding a session, including
77
+ compaction and Hub metadata. Pressure and roster recovery operate through Pi/Hub lifecycle APIs;
78
+ they never rewrite prior entries. Operators must not edit, truncate, reorder, or synthesize JSONL to
79
+ recover a session.
80
+
81
+ The project selected by `--project <name>` is the coms namespace for the Hub
82
+ and every standing or dynamically spawned peer. The Hub-facing peer spawn API
83
+ cannot override it. Native subagents remain children of that Hub and inherit
84
+ its repository context; peers are separate processes in their own panes.
85
+
86
+ At dispatch time, `backend: "auto"` applies
87
+ `.pi/agents/dispatch-policy.yaml`; `backend: "native"` bypasses same-name peer
88
+ substitution; and `backend: "coms"` requires a visible same-name peer and never
89
+ falls back to a native child. `.pi/agents/peers.yaml` owns how a peer runs
90
+ (`runner`, persona, model, extensions, and declared `env_file`), not whether a
91
+ dispatch uses it.
92
+
93
+ ## Fleet hierarchy
94
+
95
+ Agent Fleet is layered on purpose. Work flows **down** (delegate); evidence and status flow **up** — as compact structured returns, never raw dumps.
96
+
97
+ ```mermaid
98
+ flowchart TD
99
+ You(["You · Hermes inbound relay · Codex outbound conductor on your phone"])
100
+
101
+ subgraph HUBL["HUB — guarded runtime (bare just fleet)"]
102
+ Hub["agent-hub harness<br/>operator or orchestrator work mode · routes tasks<br/>owns the Verification Contract on disk · keeps research dumps out of its own context"]
103
+ end
104
+
105
+ subgraph TEAML["OPTIONAL NATIVE ROSTER (.pi/agents/teams.yaml)"]
106
+ Team["default: builder · test-engineer · code-reviewer · documenter<br/>also: plan · debug · frontend · security · hotfix · release · info"]
107
+ end
108
+
109
+ subgraph RESL["RESEARCH HELPERS — read-only, always available"]
110
+ Research["researcher (fast tier — simple reads)<br/>deep-researcher (deep tier — hard, cross-cutting questions)"]
111
+ end
112
+
113
+ subgraph SUBL["SUB-AGENTS — focused children, narrow tools + models"]
114
+ Subs["planner → scout · rules · risk<br/>plan-reviewer → feasibility · deps<br/>builder → recon · verifier<br/>test-engineer → coverage-scout · conventions<br/>code-reviewer → preflight · quality · perf · docs<br/>security-auditor → recon · input-sweep · secrets-sweep"]
115
+ end
116
+
117
+ You -->|task| Hub
118
+ Hub -->|"dispatch_agent — one persona = one specialist session"| Team
119
+ Hub -->|"spawn_research"| Research
120
+ Research -.->|"findings written to disk — hub gets paths, not dumps"| Hub
121
+ Team -->|"subagents: block (delegate_depth ≥ 1)"| Subs
122
+ Subs -.->|"results return to the parent agent only"| Team
123
+ Team -.->|"structured return + evidence"| Hub
124
+ Hub -.->|"one status line: Assertions: 2✓ 1○ 1✗"| You
125
+ ```
126
+
127
+ Every specialist session is one persona from [`agents/`](../agents/) (installed to `.pi/agents/personas/`) — *skills* tell each agent **how** to work; *personas* define **who** they are (see [agents.md](agents.md)).
128
+
129
+ The same idea as a tree:
130
+
131
+ ```text
132
+ hub (operator by default; orchestrator when selected)
133
+ ├── optional native roster (empty by default; example: default)
134
+ │ ├── builder → recon · verifier
135
+ │ ├── test-engineer → coverage-scout · conventions
136
+ │ ├── code-reviewer → preflight · quality · perf · docs
137
+ │ └── documenter
138
+ ├── research helpers (spawn_research, any time)
139
+ │ ├── researcher
140
+ │ └── deep-researcher
141
+ └── optional fleet peers (herdr + coms)
142
+ ├── architect / releaser / web-debugger panes
143
+ ├── Claude Code peer (coms bridge)
144
+ ├── Hermes (phone human · inbound ask_user)
145
+ └── ChatGPT Fleet session client (Desktop/Android · existing Pi session)
146
+ ```
147
+
148
+ Composition rule: **the hub (or a slash command) orchestrates; personas do not invoke other personas as peers.** Specialists may only fan out to their configured **sub-agents**. Research helpers write findings to disk; the hub resumes specialists with paths, not raw dumps.
149
+
150
+ ### Research search supervision
151
+
152
+ The shared `spawnPiAgent` seam supervises every `read`/`grep`/`find`/`ls` call from native
153
+ research helpers and nested delegate children. The supervisor tracks JSONL `toolCallId` values
154
+ independently, with a default 120-second deadline (`recon-search-timeout-s: 1..3600|off` under
155
+ `## agent-hub`). It is a per-tool watchdog; the whole-run bound is separate — the task-tier
156
+ per-run deadline (`agent-turn-timeout-s`), which terminates a hung run as `turn_timeout`.
157
+ On timeout or caller cancellation it owns and terminates the child's process group (SIGTERM,
158
+ then SIGKILL after a bounded grace), has a separate settlement timer for missing `close`/pipe
159
+ drain, and reports timeout separately from cancellation. Research helpers and nested delegates
160
+ are each given safe process-group ownership; delegates forward parent termination so no detached
161
+ child is orphaned. Full pattern catalog: [references/orchestration-patterns.md](../references/orchestration-patterns.md).
162
+
163
+ ### Task-tier budgets
164
+
165
+ The hub enforces per-user-turn budgets in code (`run-budget.js`): the current **task tier**
166
+ caps `dispatch_agent` calls, `spawn_research` calls, and active time per turn, sets the
167
+ per-run deadline above, and controls nested delegation (`off` at trivial/small). Exhausted
168
+ budgets make the dispatch tools refuse and request one Yes/No `ask_user` confirmation; Yes
169
+ renews the turn in the same tool loop. A normal new user message also opens a fresh turn
170
+ window.
171
+ Specialist context pressure is measured over input + cacheRead + cacheWrite against **that
172
+ agent's own** model window, resolved from pi's model registry with the source recorded
173
+ (`context-window.js`) — measuring a 49k local model against the dispatcher's window is what
174
+ made readings like "315%" unactionable; anything over 100% now emits a one-time diagnostic
175
+ naming the window and where it came from. Specialist sessions are recycled (fresh spawn
176
+ instead of `-c` resume) after `session-recycle-runs` runs, at ≥60% measured context, and
177
+ unconditionally at a full window; a resumed session whose *projected* prompt would overflow
178
+ is recycled before the spawn rather than after the run. Requests to one provider are capped
179
+ per process (`provider-semaphore.js`: 2 in flight for `custom/*` by default, unlimited
180
+ elsewhere, `AGENT_HUB_PROVIDER_LIMITS` to override) — the cap is per level of the delegation
181
+ tree, and a nested spawn reuses its parent's permit so it can never wait on its own ancestor.
182
+ Configured under `## agent-hub` (`max-dispatches-per-turn`,
183
+ `max-research-per-turn`, `turn-wall-time-s`, `agent-turn-timeout-s`, `session-recycle-runs`,
184
+ `run-history-keep`) as ceilings (`min` with the tier). A leftover `mode:` key is ignored with
185
+ a warning.
186
+
187
+ A per-message allowance cannot bound a task, so a second envelope sits above the turn one:
188
+ the **task budget** (`run-budget.js`, `3×` the turn envelope) counts dispatches, research
189
+ runs and **active** time across the WHOLE task and is *not* reset by a user message. Both turn
190
+ and task active time exclude `ask_user` waits — billing human idle false-stops a normal steered
191
+ session and teaches people to reset reflexively. The auto-research pipe is exempt from the turn
192
+ budget but charged against the task envelope, so it cannot smuggle 8 helpers per dispatch past
193
+ the outer bound.
194
+
195
+ Either envelope stops before another dispatch/research call and asks one localized Yes/No
196
+ `ask_user` question. A confirmed turn continuation renews its counters/clock in the same tool
197
+ loop. A confirmed task continuation opens one audited tranche, also renews the turn, and
198
+ preserves task tier, assertions, capability packs, label, blockers, and progress. No typed
199
+ `continue` message or slash command is required. `set_task_tier` with `new_task: true` is only
200
+ for genuinely different work and clears task identity/state. This is the guardrail the
201
+ post-mortem run never hit: every steering message reopened the turn window,
202
+ so one workspace stayed open 47 hours on a change that took 13 minutes in a narrow one.
203
+
204
+ ### Runtime acceptance and recovery
205
+
206
+ Writable specialist operations (`write`, `edit`, or `bash`) receive the same T2 minimal
207
+ acceptance contract at every task tier. The trigger is the actual operation/tool contract and
208
+ current assertion ledger, not user-language matching or the model's tier label. Read-only
209
+ operations remain lightweight. Language heuristics only expose the fuller verification pack;
210
+ they are supplementary and do not decide whether changed work needs evidence. Pending
211
+ capability leases survive a new-task transition, while stale task packs and assertion state
212
+ reset through the explicit lifecycle path.
213
+
214
+ The runtime-owned result schema is `agent-fleet.runtime-result/v1`. It records orthogonal
215
+ `execution` (`completed | failed | pending`), `changes` (`changed | unchanged | unknown`, with
216
+ `certain | uncertain | not_observed` attribution), `verification`
217
+ (`passed | failed | missing | stale | unsupported`), and `acceptance`
218
+ (`accepted | not_accepted`) dimensions, bound to task identity, before/after revision, command,
219
+ integer exit status, and evidence references. A changed result and failed verification can
220
+ therefore both be true. Compatibility presentation remains
221
+ `not_available | deliverable_failed | needs_verification | accepted` for the Hub and
222
+ `accepted | rejected` for flows.
223
+
224
+ Only native Pi children support trusted runtime checks. The Hub injects the code-owned
225
+ `runtime-test-check.ts` observer only when the native persona has `bash`; the observer wraps the
226
+ existing bash implementation and preserves the normal damage-control `tool_call` approval and
227
+ bash error-event semantics. The parent accepts a check only when producer schema, observed bash
228
+ start, exact declared command, task, coverage metadata, exit status, and unchanged inspected
229
+ revision agree. This proves exact command execution, not semantic adequacy: even a declared
230
+ `true` command can pass, so test quality remains subject to independent or human review.
231
+ Coms peers cannot produce trusted checks, and `manual`/`runtime-ui` requirements currently have
232
+ no trusted producer; all three paths fail closed as missing or unsupported.
233
+
234
+ The observer and trusted event-correlation path have an observed compatibility dependency on
235
+ `@earendil-works/pi-coding-agent@0.84.2`: they rely on its `createBashToolDefinition`, `tool_result`
236
+ hook, `args.path`, and JSON-event passthrough contracts. This records the currently verified API
237
+ surface rather than changing or independently pinning the package version; Pi upgrades must
238
+ re-verify these integration points.
239
+
240
+ Native runs also retain a bounded record of actual tool starts/ends by tool-call identity. T3
241
+ uses those code-owned events only to correlate a terminal, unfenced pseudo-write claim with the
242
+ readback of an explicitly declared deliverable. It never executes tool-shaped text. Fenced
243
+ XML/JSON documentation examples are excluded; a real matching write event with failed readback
244
+ remains a T2 verification/effect failure rather than being relabelled as a protocol failure.
245
+
246
+ A mismatch emits `agent-fleet.tool-protocol-diagnostic/v1` with category
247
+ `tool_protocol_error`, task-or-controlled-probe origin, effective backend/model/tool
248
+ configuration, the claimed effect, matching-event and deliverable facts, `none | partial`
249
+ effects, this-run-only conclusion, and evidence references. The T2 runtime-result dimensions
250
+ remain unchanged and orthogonal. For a real task, the retained diagnostic artifact is the only
251
+ input accepted by the task-generation-bound `establishEffects` recovery port; it records what
252
+ already happened so unchanged retries cannot replay partial effects. It does not authorize a
253
+ retry by itself and does not turn a task failure into a probe result or a universal model
254
+ judgment. The withdrawn T12 runner is not part of this path; historical evidence remains archived.
255
+
256
+ T5's opt-in `filesystem` tool performs deterministic inventory, byte excerpts, artifact readback, and
257
+ local-file-only source snapshots without another model run. It is registered in the Hub and native
258
+ children but appears in their effective tool surface only while the inherited complete profile has
259
+ `assist.deterministic-tools: true`; stale or forged calls are refused at execution too. Profile and
260
+ work-mode changes rebuild the truthful T4 catalog, and compaction reconciles that effective surface.
261
+ Orchestrator mode may use this narrow inspection/managed-snapshot exception, but still has no generic
262
+ bash, edit, or write tools. Native specialist, research, and nested-delegate explicit tool caps remain
263
+ authoritative: the assist flag cannot append `filesystem` outside a declared cap.
264
+ Opaque handles bind canonical paths,
265
+ content hashes/fingerprints, and offsets; continuation fails on stale bytes or directory state,
266
+ and canonical-root checks reject symlink escape. Snapshot bytes and provenance metadata are
267
+ stored separately under the owning current session, marked untrusted, and never interpreted as analysis
268
+ or fetched from a network origin. Snapshot destinations are runtime-owned and cannot be supplied by the
269
+ caller; this narrow artifact creation path grants no general write or shell authority. Every source
270
+ path—including the path resolved from a readback handle—passes through the existing damage-control
271
+ `zeroAccessPaths` hook before execution, preserving its exemptions and interactive/headless approval
272
+ semantics; a denial occurs before source bytes or snapshot artifacts are created. Handle decoding is
273
+ single-sourced in `.pi/harnesses/lib/deterministic-handle-path.ts`, so damage-control retains this guard
274
+ when installed with its declared shared-library companion but without agent-hub. Inventory names and
275
+ all other operation results are marked untrusted. Because snapshot has a managed artifact side effect,
276
+ `filesystem` is not classified as replay-safe for automatic mid-run model fallback.
277
+ `assist.deterministic-tools` and `assist.bounded-output` are independent. Enabling
278
+ `bounded-output` alone bounds read-only tool results inside native specialists, research helpers,
279
+ and nested delegates before the next model continuation, and bounds parent reply transport too;
280
+ `deterministic-tools` controls the deterministic filesystem assistance without being a prerequisite
281
+ for output bounds. The fixed bounds are 500 inventory entries per page, 64 KiB per content/reply,
282
+ and 180 Unicode characters per preview; truncation always names a retained full-output path and
283
+ retrieval handle. With `bounded-output` off, legacy transport is preserved regardless of the
284
+ `deterministic-tools` setting. Inventory records an escaping or unavailable symlink as a denied
285
+ entry without traversing it or disclosing its target outside the authorized inventory.
286
+ T5 does not consume `assist.write-isolation`; the separate T6c native launch boundary does.
287
+
288
+ Recovery remains the single T1 contract: no automatic retry, queue, model fallback, or new
289
+ numeric allowance. Busy work is refused before execution accounting or history mutation and is
290
+ not recorded as no-progress failure; indeterminate failures do not authorize retry. Operator
291
+ cancellation fences the same agent identity until a fresh, one-use authorization is consumed.
292
+ Other agent identities—including the anonymous, read-only research actor—remain independent,
293
+ and a task reset does not erase the cancellation fence. Native session reuse is bound to an exact,
294
+ normalized current-task contract (task identity, instructions, scope, deliverables, input artifacts,
295
+ effective model, and permissions). A different/narrower task starts with a fresh session and
296
+ replacement manifest; only an exact continuation, including the auto-research continuation path,
297
+ receives Pi resume transport. This remains context hygiene rather than confinement.
298
+
299
+ When `assist.write-isolation` is explicitly true, dispatch snapshots that effective setting before
300
+ routing and forces a native child. Explicit coms is refused because remote isolation cannot be
301
+ claimed. The complete Pi process and descendants run under Linux bubblewrap or macOS Seatbelt;
302
+ the repository/root view is read-only except for existing exact relative files, expressly allowed
303
+ existing recursive directories, and separate current-run runtime/artifact/temp roots. Glob scope,
304
+ absolute/escaping scope, symlinked grants, an empty writable grant set, absent or wrong-platform
305
+ backends, and sandbox launch failure all fail closed without an unsandboxed retry. Damage-control remains the policy/audit/
306
+ approval overlay, but an approval cannot widen the OS boundary already created for a run; a wider
307
+ path requires a new explicitly authorized dispatch contract. Network, model, and read access are
308
+ otherwise unchanged. Turning the profile flag off does not dismantle an
309
+ active process boundary; the next run takes a new snapshot. Opt-out retains the legacy spawn path.
310
+ The sandbox never rolls back or deletes user edits, and it cannot protect an allowlisted file from
311
+ a concurrent writer. Exact-file grants support direct writes to the existing inode but intentionally
312
+ block temp-file-plus-rename atomic saves; callers must explicitly grant the containing directory for
313
+ that behavior, and the runtime never widens the grant automatically. Sandboxed spawn uses exactly
314
+ three pipe stdio entries and rejects extra inherited descriptors, closing the known writable-FD
315
+ bypass. Linux deliberately preserves the Hub-owned process group instead of adding a PID namespace,
316
+ and `/dev/tty` remains narrowly available; confinement therefore assumes host ptrace and terminal
317
+ injection controls prevent same-uid cross-process escape (for example Linux Yama ptrace restrictions
318
+ and disabled legacy TIOCSTI). Linux enforcement is exercised on a real bubblewrap host. Seatbelt
319
+ canonicalizes only trusted runtime/artifact/temp aliases such as `/tmp` → `/private/tmp`, while user
320
+ scope grants retain strict symlink rejection; its only device write exceptions are literal
321
+ `/dev/null` and `/dev/tty`, never a blanket `/dev` grant. The Seatbelt profile has contract coverage
322
+ but still requires a Darwin runtime check before T6c is fully verified.
323
+
324
+ T4 adds a code-owned effective-tool catalog without creating another recovery policy. Every
325
+ application of work-mode tools compares the actual active catalog before and after
326
+ `setActiveTools` and can emit `agent-fleet.tool-catalog-delta/v1`: removed, added/available,
327
+ active substitutes with explicit limitations, `permissionExpansion: false`, and SHA-256 catalog
328
+ identities. The delta is persisted as a trusted session entry and included in the next system
329
+ prompt. Orchestrator mode still excludes direct `write`, `edit`, and `bash`; a substitute is only
330
+ an already-active path such as `dispatch_agent` or read-only inspection, and is never invoked by
331
+ the catalog machinery.
332
+
333
+ The runtime latches the effective catalog at `before_agent_start`. Actual assistant tool-call
334
+ blocks are compared to that originating-turn catalog on the real `message_end` path, so a
335
+ next-turn mode switch cannot charge an unknown-tool refusal for a call that was legal when the
336
+ turn began; text blocks are ignored and no pseudo-tool text is executed. On `session_compact`,
337
+ the runtime reconciles the restored `agent-fleet.tool-catalog-state/v1` identity with Pi's current
338
+ active tools, persists restore evidence, and retains the task-scoped unknown-tool counter. Catalog
339
+ errors are recorded fail-closed but cannot skip the existing compaction settlement/deferred-input
340
+ lifecycle.
341
+
342
+ Each unavailable tool refusal is keyed by normalized task and tool, while normalized arguments
343
+ and catalog identity record whether the request or availability was reevaluated. The
344
+ operator-approved limit is exactly three counted refusals, including the first. Mode switches,
345
+ compaction, prose, and real catalog changes never refill it; only the explicit new-task lifecycle
346
+ resets it. A real `unknown_tool` dispatch failure is stamped by the Hub with the latched,
347
+ code-owned catalog identity; a result-supplied identity is ignored. A later successfully persisted
348
+ effective-catalog event can establish `toolStateChanged` evidence for the existing T1
349
+ `unknown_tool` `recoveryDecision`, but still requires explicit reinvocation and existing general
350
+ budgets; it never schedules an automatic retry.
351
+
352
+ On top of the tier envelope sit several qualitative guardrails. **Task triage**: the dispatcher
353
+ classifies the current TASK via the `set_task_tier` tool (`trivial`/`small`/`feature`/`project`)
354
+ and that classification *is* the budget. The tier is task-scoped and **ratcheted** — it survives
355
+ the user's next message, lowering is free, raising needs a stated `reason` — because a
356
+ turn-scoped tier reset to `feature` on every correction. Three refusals enforce
357
+ proportionality in code rather than prose: a duplicate-dispatch guard (near-identical
358
+ re-dispatches within a turn), a **tier persona gate** (`planner`, `plan-reviewer`,
359
+ `architect`, `security-auditor`, `deep-researcher` are refused at trivial/small — each opens
360
+ a document/finding loop), and a **review round cap** (review dispatches per task, by tier).
361
+ Skipping triage assumes `small`, not `feature`: the tier latches for the task, and the case
362
+ where it was never declared is exactly the case where proportionality was not being
363
+ considered. The review **finding** budget is the one thing deliberately left advisory — the
364
+ hub counts blocking findings (`review-findings.js`) and reports an over-budget return, but
365
+ never reclassifies one, because no rule it can evaluate separates an invented manifest from
366
+ a leaked credential; the round cap is what carries the enforcement.
367
+ **Process obligations (T11)** are task-scoped and independent of that budget tier. The
368
+ `set_task_tier` tool accepts explicit `risk` (`unknown | low | high`) and process `scope`
369
+ (`read-only | small | wide`); neither is inferred from task prose. Legacy/omitted risk is
370
+ `unknown`, risk changes require a reason, and scope expansion requires a reason plus an
371
+ explicit risk reassessment. Unknown permits read-only research but cannot accept changed
372
+ work. Confirmed low-risk small work uses the existing T2 minimal acceptance path; high risk
373
+ adds a separate review stage, while wide work adds plan and review stages. These obligations
374
+ ratchet within a task, so lowering spend cannot erase them. Completed stage records name an
375
+ artifact and revision but only prove that the stage occurred—not that model-authored review
376
+ is semantically adequate.
377
+
378
+ The same code-owned state is consumed by dispatch acceptance in operator and orchestrator
379
+ modes, appended as `agent-hub-process-state`, restored from the session log, and retained
380
+ through compaction and mode changes. Only `set_task_tier(new_task: true)` creates fresh
381
+ unknown-risk process state. `/af-audit` exports allowlisted satisfied/open/unsupported
382
+ obligation statuses and explains that the budget tier limits spend rather than correctness;
383
+ reasons and task prose are not copied. This extends the existing task/acceptance/audit paths
384
+ and is not a second workflow engine.
385
+
386
+ **Docs lane** (`docs-lane.js`): a dispatch whose whole declared `scope` is documentation
387
+ refuses review personas (overridable with `review_reason`) and tells the dispatcher not to
388
+ open a review gate; an absent scope is never the lighter lane. **External-blocker stop**
389
+ (`external-blocker.js`): a specialist emits `EXTERNAL_BLOCKED: …` when it needs something
390
+ outside the fleet's reach (account, permission, credential, telemetry destination), and the
391
+ hub refuses the next dispatch with an owner-escalation packet until the human is addressed —
392
+ the alternative, observed, is hours spent approximating the missing fact with scripts,
393
+ manifests and fixtures while the assertion still ends UNPROVEN. **Drift watchdog**
394
+ (`drift-watchdog.js`): armed dispatches are
395
+ observed in-flight from the JSON event stream — deterministic rules (out-of-scope writes
396
+ against the declared `scope` globs, tool-call loops, consecutive failures, tool-call cap)
397
+ escalate to a one-shot cheap LLM judge whose DRIFTING/STUCK verdict terminates the run as
398
+ `drift_stop` (exit 125, partial output preserved); enabled per hub/agent/dispatch
399
+ (`watchdog` key, `/af-watchdog`, `watchdog` param). Orchestrator work mode auto-arms the
400
+ watchdog when the hub setting is `auto`/`on`; a dispatch `watchdog: false` cannot disarm it
401
+ there. Hub or per-agent `off` remains the opt-out. Two rules about scope: the session's
402
+ own `artifacts/`, `findings/`, and `delegations/` subtrees are implicitly in scope (the
403
+ deliverable protocol *orders* specialists to write there, and the judge is told so), and the
404
+ `scope` rule is non-terminal — it reports a drift advisory on the result and never stops a
405
+ run by itself, matching the post-run scope gate, which reverts nothing. **Dynamic teams**: `/af-agents-add`,
406
+ `/af-agents-drop`, `/af-agents-save` restructure the roster live (the system prompt rebuilds
407
+ every turn), and the gated `team_adjust` tool lets the dispatcher itself adjust the roster
408
+ when nested delegation is on (feature/project), with user notification. `/af-hub-report` accounts each turn's dispatches,
409
+ tokens (billed = input + cacheRead + cacheWrite), recycles, drift stops, and refusals.
410
+
411
+ ## Runtime stack (tools the fleet sits on)
412
+
413
+ ```mermaid
414
+ flowchart TD
415
+ AF["<b>Agent Fleet</b><br/>agent-hub · personas · skills · coms · bridges · CLI"]
416
+ AF -->|primary runtime| PI["<b>pi</b><br/>coding agent — loads harnesses,<br/>extensions, prompts, personas"]
417
+ AF -->|control plane| HERDR["<b>herdr</b><br/>tiled peer workspaces,<br/>presence, snapshot/resume"]
418
+ AF -->|coms peer only| CC["<b>Claude Code</b><br/>bidirectional peer via<br/>the coms bridge<br/><i>not an install target</i>"]
419
+ AF -->|remote human| HERMES["<b>Hermes</b><br/>hub questions relayed to your phone,<br/>plus the Desktop fleet panel"]
420
+ AF -->|ChatGPT-initiated session client| CODEX["<b>ChatGPT Fleet session client</b><br/>existing Pi session via<br/>opt-in chatgpt-client (experimental)"]
421
+ ```
422
+
423
+ ### External dependencies
424
+
425
+ These are the external systems Agent Fleet assumes or integrates with — not npm packages, but the **runtime stack** the fleet operates on top of.
426
+
427
+ | Dependency | Role | Required? |
428
+ | --- | --- | --- |
429
+ | **[pi](https://github.com/badlogic/pi-mono)** (or your pi install) | Primary coding-agent runtime; loads harnesses, extensions, prompts, and personas | Yes for `just fleet` |
430
+ | **[herdr](https://herdr.dev)** | Workspace control plane: Hub/peer panes, presence push events, team snapshot/resume | Required for `--herdr` or `--peers`; optional for bare `just fleet` |
431
+ | **[Claude Code](https://docs.anthropic.com/en/docs/claude-code)** | First-class coms peer via the [coms bridge](claude-code-coms-bridge.md) — cross-model review and analysis. Never an install target: it hosts no skills, commands, or personas | Optional peer |
432
+ | **Hermes** | Remote human-in-the-loop (Telegram relay for hub questions — [coms-hermes-bridge](coms-hermes-bridge.md)) and the Desktop fleet panel ([hermes-desktop-plugins](hermes-desktop-plugins.md), needs v0.19.0+ and the Desktop app) | Optional |
433
+ | **ChatGPT Desktop/Android** | Experimental opt-in Fleet session client for an existing Pi session; Node 22.18+, Python 3.10+, local herdr/coms access — [runbook](codex-session-bridge.md) | Optional experimental feature `chatgpt-client` |
434
+ | **[addyosmani/agent-skills](https://github.com/addyosmani/agent-skills)** | Upstream skill library (manually vendored) | Bundled (vendored) |
435
+ | **[disler/pi-vs-claude-code](https://github.com/disler/pi-vs-claude-code)** | Source inspiration / MIT port origin for pi harnesses | Design lineage (ported in-repo) |
436
+ | **LLM providers** | Models per persona (`model:` / `models:` in agent frontmatter) — e.g. OpenAI Codex, GitHub Copilot, Ollama, … | Yes (at least one provider your agents can call) |
437
+ | **Chrome DevTools MCP** / **Playwright Agent CLI** | Browser verify (`browser-testing-with-devtools`) and headless automation (`bowser`) | Optional, feature-specific |
438
+ | **Node.js + npm** | CLI (`npx @chankov/agent-fleet`), package install, `just` recipes | Yes for install & tooling |
439
+
440
+ The existing `agent-fleet doctor` command also performs a read-only manifest preflight. It reports exact selected installation files and manifest-declared tool probes as environment readiness, distinguishes missing installation from unknown or unavailable platforms, and names the existing explicit remediation path. Bare doctor never repairs or installs; `doctor --fix` remains the only repair activation, and package/tool installation remains behind its existing explicit consent path.
441
+
442
+ ## Repository module map
443
+
444
+ ### Agent Hub composition
445
+
446
+ `.pi/harnesses/agent-hub/index.ts` owns mutable Hub state and composes the extracted concerns without merging their context contracts:
447
+
448
+ ```text
449
+ agent-hub/
450
+ ├── index.ts # bounded composition root; constructs root-owned mutable ports
451
+ ├── types.ts # shared Hub agent/state contracts
452
+ ├── config/ # agent/frontmatter, teams, profiles, dispatch policy, overrides
453
+ ├── context/ # state, budgets, assertions/artifacts, session-health factories
454
+ ├── commands/ # 21 typed /af-* command registrars
455
+ ├── tools/ # 16 tools plus dispatch/research execution orchestration
456
+ ├── dispatch-core.ts # facade over native, coms, and observability dispatch
457
+ ├── dispatch-*.ts # native preparation/spawn/completion and backend adapters
458
+ ├── acceptance.ts # agent-fleet.runtime-result/v1 and minimal acceptance evaluation
459
+ ├── runtime-test-check.ts # code-owned native bash observer for trusted runtime checks
460
+ ├── tool-protocol.ts # T3 event/readback correlation and shared diagnostic schema
461
+ ├── recovery-contract.ts # shared typed failure/recovery policy
462
+ ├── policy/ # model, roster, capability, and work-mode decisions
463
+ ├── research/ # helper runtime, spawn transitions, and controls
464
+ ├── ui/ # grid, dashboard, detail, context, pool, zoom, and history
465
+ ├── input/ # shortcuts and completion derivation
466
+ ├── lifecycle/ # turn, pressure, monitor, session, and shutdown orchestration
467
+ ├── prompts/ # data-oriented system/session templates + HubPromptContext
468
+ ├── session-start.ts # typed ordered session_start step runner
469
+ ├── presentation.ts # marker, name, thinking, and delegate-path helpers
470
+ ├── timeline.ts # bounded transcript/timeline append helpers
471
+ ├── monitor-*.ts # local monitor transport and recovery primitives
472
+ └── *.test.{ts,js} # registration, loader, parity, and behavior contracts
473
+ ```
474
+
475
+ Cross-harness coms lifecycle remains shared in `.pi/harnesses/lib/coms-core.ts` and `coms-core-io.ts`; it is not owned by the Hub composition root.
476
+
477
+ ```text
478
+ .pi/ # Pi runtime: harnesses, extensions, agents config, prompts
479
+ .pi/agent-fleet/ # The fleet runtime, mirrored verbatim into a workspace:
480
+ scripts/ # CLI helpers, bridges, team + one-off peer launchers (pure logic in scripts/lib/)
481
+ scripts/workflows/ # deterministic flow graphs + phase/gate/permission/quality runtime
482
+ hermes/ # Hermes-facing skills/integration assets
483
+ hooks/ # the coms bridge Stop hook
484
+ docs/ # the one doc that installs into a workspace (workflows.md)
485
+ skills/ # Agent Fleet-native skills (shadow vendored names)
486
+ agents/ # Personas/subagents; installed to a workspace's .pi/agents/personas/
487
+ codex/ # Canonical Codex conductor contract (runtime copy lives outside checkout)
488
+ systemd/user/ # Owned user-unit template for Codex remote control
489
+ vendor/agent-skills-upstream/ # Manually imported upstream skills (pinned SHA)
490
+ bin/ # npm CLI: init/update/doctor + the installer engine
491
+ references/ # Supplementary checklists, installed with the skills that cite them
492
+ docs/ # This file, setup guides, bridge references, vendoring policy
493
+ ```
494
+
495
+ The fleet runtime keeps its *source* layout identical to its *installed* layout.
496
+ That is not cosmetic: `companion:harness-runtime-closure` and its siblings copy
497
+ with `preserveLayout`, so a relative import written here resolves the same way in
498
+ a workspace — including the ones reaching from the stationary `.pi/harnesses/`
499
+ into the movable `.pi/agent-fleet/scripts/`. `bin/test/path-boundaries.test.js`
500
+ holds that invariant.
501
+
502
+ Reserved for future modules (do not repurpose these paths):
503
+
504
+ ```text
505
+ apps/dashboard/ # future dashboard: Kanban state, Herdr workspaces, peer status
506
+ packages/fleet-core/ # future extracted core orchestration library
507
+ packages/herdr-bridge/ # future Herdr integration package
508
+ packages/hermes-bridge/ # future Hermes integration package
509
+ ```
510
+
511
+ ## Design rules
512
+
513
+ - **Workflow engine remains single-sourced.** `just flow` owns deterministic phase order, retries, executable gates, post-hoc writes enforcement, and one flow acceptance decision. It uses the clean spawn/context seams and keeps its existing clean-tree/branch defaults; normal traces live under `.pi/flow-sessions/`. Flow branches retain source/result metadata, while reserved `cleanup` and `merge` delegate safe removal and squash integration to Worktrunk. The initial Hub `run_flow` adapter calls the same exported dispatcher directly and permits only read-only `scout`: it runs in a stable detached snapshot with no flow branch, stores trace/manifest in the current Hub session, and has one Hub budget/cancellation owner. Snapshot overlay operations validate plain parents before mutation and never follow a HEAD symlink leaf. The native process and descendants use `confineNativeChild`, with only the session-owned phase directory writable; the actual `read`/`grep`/`find`/`ls` data paths are separately limited to the snapshot while damage-control remains loaded. Runtime/model libraries and provider configuration remain readable for Pi startup, so this is intentionally not a global no-read claim. Duplicate identity is idempotent; stale/busy calls are uncharged. All setup/result exceptions settle as structured failures, release active ownership, and preserve honest charged/started state. Flow acceptance never promotes parent assertions or closes T11 obligations. Writable/committing flows remain outside this adapter pending a separate review.
514
+ - **Thin dispatcher context.** Nothing lands persistently in the dispatcher's
515
+ context if it can live on disk or in a one-line status. Research findings,
516
+ the Verification Contract ledger, and team snapshots are all disk-first.
517
+ - **A harness fault must never look like a specialist fault.** The hub passes
518
+ `--session <file>` on every run, so one corrupt session file used to fail a
519
+ persona in ~1s with no output — indistinguishable from a bad agent, and
520
+ unrecoverable by drop + re-add. Unusable session files are now validated and
521
+ quarantined (`session-health.js`) with the reset named in the result. Same rule
522
+ for the return contract: a report the parser cannot read gets one cheap
523
+ read-only extraction pass (`return-extract.js`) before its assertions are
524
+ written off as unproven, and extracted evidence is always labelled as weaker
525
+ than declared evidence. The converse also holds: a run that errored or timed
526
+ out writes to `artifacts/failures/`, never `returns/` — an error stub filed
527
+ as a return reads as a specialist verdict and gets acted on as one.
528
+ - **A pool status field beats reading the screen.** `coms_list` publishes each
529
+ peer's `pane_id` and `status` (`idle`/`working`/`booting`), and
530
+ `herdr_spawn_peer` waits for the peer to register and returns `peer_ready`
531
+ rather than a bare pane id. A spawned peer boots idle and does nothing until
532
+ addressed, so peers spawned and never sent to are named at turn end and in
533
+ `/af-hub-report` (`spawned-peers.js`); closing stays the human's call.
534
+ - **A declared requirement carries its origin.** Every assertion in the ledger
535
+ names its source (`assertion-ledger.js`), and the open ledger is soft-capped
536
+ at 8 — an id nobody can trace back to a plan line costs a dispatch and an
537
+ ASK_USER cycle to re-derive.
538
+ - **Pre-flight validation is free.** Anything the hub can reject before spawning
539
+ — an unresolvable artifact path, an unknown research persona, a heavy persona
540
+ at a low tier, a reviewer on a docs-only scope — is refused without spending a
541
+ budget slot. Artifact paths also resolve across artifact kinds when the name is
542
+ unique, since the hub writes every auto-return under `returns/` while
543
+ dispatchers reasonably guess `reviews/`.
544
+ - **Evidence is archived, never overwritten.** Session start moves the previous
545
+ session's `artifacts/` into an immutable `.pi/agent-sessions/runs/<runId>/`
546
+ namespace with a read-only `meta.json` and an appended `runs/index.json`
547
+ (`run-namespace.js`), retained per `run-history-keep` (default 10). The old
548
+ behaviour — delete `artifacts/` at start, name returns by a per-session counter
549
+ — collided two ways at once and made a post-mortem record eleven specialist
550
+ returns and two reviews as NOT RECOVERABLE. A failed archive leaves the
551
+ artifacts in place: a stale tree is recoverable, a deleted one is not.
552
+ - **Herdr owns panes, presence, and lifecycle; coms owns messages.** Herdr
553
+ topology recipes (`--herdr` or `--peers`) require a running server and refuse
554
+ with an actionable message otherwise. Bare Fleet does not require Herdr.
555
+ `--no-coms` leaves direct/native execution available but disables peer
556
+ messaging, peer-backed dispatch, and handoff from the Hub.
557
+ - **External agents are peers, not plugins.** Claude Code (and future CLI
558
+ agents) join the fleet through bridge adaptors that speak coms envelopes —
559
+ the fleet core stays agent-agnostic. Hermes remains the inbound `ask_user`/
560
+ Telegram route; the experimental Codex conductor is outbound-initiated only,
561
+ approval-gated, and restricted to listed peers through the validated wrapper.
562
+ - **Hermes monitor presentation is outside the fleet core.** Agent Fleet exposes owner-only
563
+ local monitor operations for Hub-owned state; a consumer owns its UI and lifecycle. The
564
+ worktree contains additive monitor/event/invoke code, but that implementation and its local
565
+ tests are not proof of a durable external identity or live delivery contract. `invoke`, where
566
+ available, is Hub-owned and queues dispatcher work rather than exposing tools directly.
567
+ - **Packaged Hermes source is opt-in, never auto-installed.** The npm tarball carries the
568
+ `hub-watchdog` skill (`.pi/agent-fleet/hermes/skills/`) plus the backend and Desktop monitor plugin source
569
+ (`.pi/agent-fleet/hermes/plugins/`, `.pi/agent-fleet/hermes/desktop-plugins/`) as runtime-only source. Shipping that source
570
+ makes it available to an operator; installing it into a Hermes profile is always an explicit
571
+ action through `agent-fleet set-hermes-watchdog` (skill) or the consumer's own flow (plugins).
572
+ Nothing is enabled, launched, or configured by installing the package.
573
+ - **Watchdog delivery is capability-gated and currently unproven.** No checked-in
574
+ Gate O live artifact proves Hermes origin identity, updates, reconnect, or two-chat isolation,
575
+ so its supported posture is journal-only/dormant: no delivery, steering, or surgical use. It
576
+ never manages services, gateways, Herdr, or shell commands. Local runtime evidence — including
577
+ a real foreground watcher against a disposable Hub UDS — is `synthetic-local` and proves none of
578
+ those capabilities; see [the watchdog runbook](hermes-watchdog-supervisor.md).
579
+ - **External conductor contracts are advisory.** Pi damage-control wraps Pi
580
+ tool calls, not Hermes or Codex processes; human approvals and their
581
+ contracts reduce risk but do not provide an OS command allowlist.
582
+ - **Destructive fleet verbs are damage-control-guarded.** Specialists cannot
583
+ spawn/close herdr panes; the human confirms destructive actions.
584
+ - **Native-over-vendored skills.** The skill catalogue resolves `skills/`
585
+ first, then the vendored upstream import; upstream updates are explicit
586
+ maintainer actions ([UPSTREAM-SKILLS.md](UPSTREAM-SKILLS.md)).
587
+
588
+ ## History
589
+
590
+ Agent Fleet began as a fork of `addyosmani/agent-skills` and was split into a
591
+ standalone repository in July 2026, with upstream demoted to vendored content.
592
+ The one-time migration record, including the history-filtering commands, lives
593
+ in [MIGRATION-agent-fleet.md](MIGRATION-agent-fleet.md). The product
594
+ requirements that drove the split were a one-off planning document and are no
595
+ longer maintained.