fix: consolidate verified workspace issue sweep (#432)

* fix(start): use server-entry wrapper for production * fix(swarm): reconcile oneshot checkpoints and ignore phantom blockers * fix(profiles): sync editable descriptions from profile config * fix: show cron jobs across Hermes profiles * fix(capabilities): clarify dashboard-backed API detection * feat: make Conductor use native Swarm fallback Treat Workspace-native Swarm as the official Conductor fallback when the dashboard mission API is unavailable. Preserve dashboard-first dispatch, native status/cancel handling, provider-neutral setup docs, and regression coverage for gateway capability detection, swarm health, roster/profile handling, and native Conductor responses. * fix(usage-meter): reposition menu trigger for better alignment - Adjusted the position of the menu trigger in the usage meter component to enhance layout consistency and user experience. fix(chat-panel): adjust position of chat panel toggle button - Updated the positioning of the chat panel toggle button to improve visibility and accessibility by changing its bottom and right offsets. * fix(stt): wire Groq/OpenAI voice transcription into chat * Fix Workspace Kanban loopback dashboard link * fix(update): do not open historical release notes on startup * fix(chat): clear stale thinking runs reliably * fix(dashboard): trust sessions endpoint for status * fix(settings): address review — local default, OAuth lifecycle, validation * fix(dashboard): always scrape live session token from HTML * fix(chat): avoid portable history replay on bound sessions * fix(settings): remove dead smart routing controls * fix(tasks-api): guard against HTML catch-all in probeBackend The /api/hermes-tasks route was renamed to /api/claude-tasks in commit efcb7d14, but the probe logic still listed the old route as a candidate. When probed, the SPA catch-all returned a 200 HTML response instead of a 404, so probeBackend() treated it as a valid (empty) backend and then failed when the actual task fetch threw. Fixes: - probeBackend() now checks Content-Type: application/json and returns -1 for non-JSON responses, so future route renames degrade gracefully. - resolveBackend() now only selects hermes if hermesCount > 0, defaulting to claude-tasks (the active backend post rename) when hermes is absent. * fix(terminal): default cwd to ~ and fallback if path missing PTY helper chdir fails when ~/.hermes is absent (common in Docker). Default shell cwd to home; server falls back to HOME if cwd does not exist. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(settings): satisfy lint on custom providers UI * fix(docker): add -m flag to useradd so workspace home dir is created Without -m, the system account has no /home/workspace directory. The auth middleware tries to write the session store at /home/workspace/.hermes/workspace-sessions.json; mkdirSync fails with EACCES because /home/ is root-owned (755), causing the 'Failed to persist session store' warning and a 500 on every authenticated route. Adding -m causes useradd to create and chown /home/workspace correctly so the session store can be written on first login. * fix(tasks): preserve real session links and restore task launch flow * fix(launchd): install macOS plist from server-entry template * fix(docker): expose dashboard API and persist workspace volumes * fix(jobs): serialize deliver targets for cron API * Make Hermes Workspace installable as PWA * chore(deps): pin direct tanstack versions * feat: align semantic Hermes swarm agents Add semantic swarm roster metadata, profile/tool/skill docs, shared semantic worker ID validation, focused roster regression coverage, and one-shot checkpoint capture for dispatch smoke tests. * fix(conductor): pass through sessionKeyPrefix from portable spawn result sessionKeyPrefix was hardcoded to null in conductor-spawn.ts, breaking async session resolution when the dashboard backend returns a prefix. Now mirrors the sessionKey pattern and passes through the value from the spawn result. Co-authored-by: Hermes Agent * feat(swarm): bridge workspace kanban to native Hermes * fix(chat): keep portable main pinned without breaking resolved sessions * Add Windows startup script for Hermes Workspace Document PowerShell usage for launching and restarting gateway + workspace via WSL tmux. Co-Authored-By: Oz <oz-agent@warp.dev> * fix(config): name Hermes Agent in restart notice * fix(swarm): reconcile aggregate semantic worker exports --------- Co-authored-by: Aurora release bot <release@outsourc-e.com> Co-authored-by: motoki takahashi <motokitakahashi@motokinoMac-mini.local> Co-authored-by: Vicky Wonder <vicky@openclaw.ai> Co-authored-by: Vitaliy Isikov <visikov+supagoku@gmail.com> Co-authored-by: Hermes Agent <hermes-agent@local> Co-authored-by: Nikolay Mohr <nikomohr96@gmail.com> Co-authored-by: Niko Mohr <niko@friendsfromcollege.de> Co-authored-by: wtchronos <262830926+wtchronos@users.noreply.github.com> Co-authored-by: Dak0verflow <dakotaferris@gmail.com> Co-authored-by: norema <mamadou.marone.19@gmail.com> Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: daoyuan <ludaoyuan1989@gmail.com> Co-authored-by: firemountain <firemountain@gmail.com> Co-authored-by: RAZSOC Local <razsoc@local> Co-authored-by: Waylon Kenning <waylonkenning@Waylons-MacBook-Pro.local> Co-authored-by: Kublai <kublai@kublai.local> Co-authored-by: justa <justa@local> Co-authored-by: Oz <oz-agent@warp.dev>
2026-05-13 22:34:42 -04:00
parent 372b18a8e4
commit cd6115b2fc
132 changed files with 9129 additions and 1701 deletions
--- a/swarm.yaml
+++ b/swarm.yaml
@@ -1,263 +1,498 @@
 version: 1
 workers:
-  - id: swarm6
-    name: Reviewer
-    role: Reviewer / Merge Gate
-    specialty: code review, regression detection, test coverage, merge readiness
-    model: Opus 4.7
-    mission: Catch breakage before it reaches Eric. Be the final quality gate.
-    skills:
-      - swarm-pr-worker
-      - swarm-worker-core
-    capabilities:
-      - code-review
-      - regression-analysis
-      - test-planning
-      - quality-gate
-      - merge-readiness
-    preferredTaskTypes:
-      - review
-      - qa
-      - regression
-      - verification
-      - merge-gate
-    reviewRequired: false
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm12
-    name: Triage
-    role: PR / Issues Primary
-    specialty: GitHub triage, issue reproduction, PR review prep, focused fixes
-    model: GPT-5.5
-    mission: Turn repo issues and review feedback into reproductions, plans, and tested PRs.
-    skills:
-      - swarm-pr-worker
-      - swarm-worker-core
-    capabilities:
-      - github-triage
-      - issue-reproduction
-      - pr-patching
-      - review-feedback
-      - gh-cli
-    preferredTaskTypes:
-      - pr
-      - issue
-      - bugfix
-      - github
-      - triage
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm5
-    name: Builder
-    role: Primary Builder
-    specialty: full-stack implementation across Hermes Workspace and Swarm2
-    model: GPT-5.5
-    mission: Ship focused product slices with tests and clean diffs.
-    skills:
-      - swarm-ui-worker
-      - swarm-worker-core
-    capabilities:
-      - code-editing
-      - ui-implementation
-      - full-stack-integration
-      - build-verification
-    preferredTaskTypes:
-      - implementation
-      - ui
-      - frontend
-      - integration
-      - feature
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm4
-    name: Sage
-    role: Research / AI Intelligence / X Content
-    specialty: AI/model/runtime research, technical synthesis, X post drafting
-    model: GPT-5.5
-    mission: Produce decision-grade research and copy so builders and Eric stay unblocked.
-    skills:
-      - swarm-worker-core
-    capabilities:
-      - research
-      - benchmark-planning
-      - model-analysis
-      - technical-synthesis
-      - x-content
-    preferredTaskTypes:
-      - research
-      - benchmark
-      - analysis
-      - options
-      - x-post
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm10
-    name: Sidekick
-    role: Secondary Builder / Fast Patch Lane
-    specialty: parallel implementation slices, refactors, focused bug fixes
-    model: GPT-5.5
-    mission: Take decomposed tasks and produce clean patches quickly.
-    skills:
-      - swarm-ui-worker
-      - swarm-worker-core
-    capabilities:
-      - code-editing
-      - refactor
-      - parallel-implementation
-      - bug-fix
-      - build-verification
-    preferredTaskTypes:
-      - implementation
-      - refactor
-      - backend
-      - frontend
-      - patch
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm2
-    name: Foundation
-    role: Backend Foundation / Runtime Contracts
-    specialty: ClaudeSwarm runtime contracts, API design, plugin foundations, state
-    model: GPT-5.4
-    mission: Build the backend substrate that keeps swarm state, dispatch, sessions, and plugins coherent.
-    skills:
-      - swarm-dev-runtime
-      - swarm-worker-core
-    capabilities:
-      - api-design
-      - runtime-contracts
-      - plugin-foundation
-      - state-management
-    preferredTaskTypes:
-      - backend
-      - runtime
-      - api
-      - infrastructure
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm3
-    name: Mirror
-    role: Swarm Control Plane / Main-Session Mirror
-    specialty: control-plane state, card parity, main-agent continuity
-    model: GPT-5.4
-    mission: Keep the swarm control plane operationally useful and visually canonical.
-    skills:
-      - swarm-orchestrator
-      - swarm-worker-core
-    capabilities:
-      - orchestration
-      - main-session-continuity
-      - control-plane-state
-    preferredTaskTypes:
-      - orchestration
-      - coordination
-      - handoff
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm8
-    name: Watch
-    role: Ops / Health / Lifecycle
-    specialty: gateway health, cron, runtime/process status, lifecycle sweeps
-    model: GPT-5.4
-    mission: Keep the local swarm/gateway environment stable and observable.
-    skills:
-      - swarm-dev-runtime
-      - swarm-worker-core
-    capabilities:
-      - ops-health
-      - gateway-status
-      - runtime-monitoring
-      - cron
-      - lifecycle-sweep
-    preferredTaskTypes:
-      - ops
-      - health
-      - monitoring
-      - lifecycle
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm7
-    name: Scribe
-    role: Docs / Specs / Handoffs / Memory Hygiene
-    specialty: handoffs, specs, README/docs, skill documentation, memory curation
-    model: GPT-5.5
-    mission: Preserve context and make the swarm understandable across sessions.
-    skills:
-      - swarm-worker-core
-    capabilities:
-      - documentation
-      - handoffs
-      - spec-writing
-      - runbooks
-      - memory-curation
-    preferredTaskTypes:
-      - docs
-      - handoff
-      - spec
-      - runbook
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm9
-    name: Lab
-    role: PC1 Experiments / Local-Model R&D / BenchLoop
-    specialty: PC1 experiments, local model testing, model runtime comparisons
-    model: GPT-5.5
-    mission: Explore fast without destabilizing canonical paths.
-    skills:
-      - swarm-ui-worker
-      - swarm-worker-core
-    capabilities:
-      - prototype
-      - experiment
-      - local-model
-      - benchmark-experiment
-      - rapid-build
-    preferredTaskTypes:
-      - prototype
-      - experiment
-      - hackathon
-      - benchmark
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm11
-    name: QA
-    role: Secondary QA / Reviewer Support
-    specialty: secondary review, smoke verification, expected-vs-actual checks
-    model: GPT-5.4
-    mission: Review parallel outputs and protect quality gates.
-    skills:
-      - swarm-pr-worker
-      - swarm-worker-core
-    capabilities:
-      - secondary-review
-      - quality-gate
-      - regression-analysis
-      - smoke-verification
-    preferredTaskTypes:
-      - review
-      - qa
-      - verification
-      - smoke
-    maxConcurrentTasks: 1
-    acceptsBroadcast: true
-  - id: swarm1
-    name: Overflow
-    role: PR / Issues Overflow / Clawsuite (offline for now)
-    specialty: parked broken lane, keep out of rotation until worker runtime is repaired
-    model: GPT-5.5
-    mission: Stay out of active rotation until the swarm1 worker runtime is repaired, then absorb overflow from PR/issues lanes and cover Clawsuite.
-    skills:
-      - swarm-pr-worker
-      - swarm-worker-core
-    capabilities:
-      - github-triage
-      - small-patches
-      - issue-reproduction
-      - clawsuite
-    preferredTaskTypes:
-      - pr
-      - issue
-      - small-fix
-      - clawsuite
-    maxConcurrentTasks: 1
-    acceptsBroadcast: false
+- id: orchestrator
+  name: Orchestrator
+  role: Swarm Orchestrator / Greenlight Gate
+  specialty: mission routing, task decomposition, handoffs, proof contracts, human approval gates
+  model: GPT-5.5
+  mission: Decompose missions into safe, proof-bearing work and route to the right specialist while preserving human greenlight control.
+  profile: orchestrator
+  modes:
+  - plan
+  tools:
+  - todo
+  - kanban
+  - delegation
+  - terminal
+  - file
+  - gbrain
+  - session_search
+  - cronjob
+  - skills
+  - clarify
+  - web
+  skills:
+  - orchestrator-core
+  - gstack-for-hermes
+  - gbrain
+  - kanban-orchestrator
+  - subagent-driven-development
+  - writing-plans
+  - requesting-code-review
+  - workspace-dispatch
+  capabilities:
+  - orchestration
+  - decomposition
+  - routing
+  - proof-contracts
+  - greenlight-gate
+  preferredTaskTypes:
+  - orchestration
+  - planning
+  - routing
+  - coordination
+  - handoff
+  greenlightRequiredFor:
+  - merge
+  - publish
+  - destructive
+  - external-send
+  - credential-change
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: orchestrator:plan
+- id: km-agent
+  name: KM Agent
+  role: RAZSOC / GBrain Knowledge Steward
+  specialty: RAZSOC, GBrain, Obsidian, TaskNotes, graph health, durable knowledge capture, drift audits
+  model: GPT-5.5
+  mission: Keep the operating brain coherent, searchable, and source-of-record aligned without polluting durable knowledge.
+  profile: km-agent
+  modes:
+  - health
+  - curate
+  tools:
+  - gbrain
+  - file
+  - terminal
+  - session_search
+  - skills
+  - todo
+  - cronjob
+  - web
+  skills:
+  - km-agent-core
+  - gbrain
+  - obsidian-markdown
+  - obsidian-cli
+  - obsidian-bases
+  - json-canvas
+  - gstack-for-hermes
+  capabilities:
+  - gbrain
+  - razsoc
+  - obsidian
+  - tasknotes
+  - drift-audit
+  - knowledge-curation
+  preferredTaskTypes:
+  - knowledge
+  - curation
+  - brain-health
+  - drift
+  - documentation
+  greenlightRequiredFor:
+  - delete
+  - purge
+  - bulk-edit
+  - source-of-record-change
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: km:health
+- id: builder
+  name: Builder
+  role: Scoped Implementation Agent
+  specialty: focused implementation, tests, small diffs, integration fixes
+  model: GPT-5.5
+  mission: Ship scoped product/code slices with tests, minimal diffs, and clear verification evidence.
+  profile: builder
+  modes:
+  - task
+  tools:
+  - terminal
+  - file
+  - browser
+  - web
+  - gbrain
+  - session_search
+  - skills
+  - todo
+  skills:
+  - builder-core
+  - gstack-for-hermes
+  - test-driven-development
+  - systematic-debugging
+  - github-pr-workflow
+  - requesting-code-review
+  - codebase-inspection
+  capabilities:
+  - implementation
+  - code-editing
+  - tests
+  - integration
+  - build-verification
+  preferredTaskTypes:
+  - implementation
+  - bugfix
+  - feature
+  - refactor
+  - integration
+  greenlightRequiredFor:
+  - merge
+  - push
+  - destructive
+  - external-write
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: builder:task
+- id: reviewer
+  name: Reviewer
+  role: Independent Review / Merge Gate
+  specialty: security review, logic review, regression detection, quality gates
+  model: GPT-5.5
+  mission: Independently review changes and block unsafe, untested, or logically broken work before it lands.
+  profile: reviewer
+  modes:
+  - gate
+  tools:
+  - terminal
+  - file
+  - web
+  - gbrain
+  - session_search
+  - skills
+  skills:
+  - reviewer-core
+  - requesting-code-review
+  - github-code-review
+  - systematic-debugging
+  - gstack-for-hermes
+  - gbrain
+  - codebase-inspection
+  capabilities:
+  - code-review
+  - security-review
+  - regression-analysis
+  - quality-gate
+  - merge-readiness
+  preferredTaskTypes:
+  - review
+  - qa
+  - regression
+  - verification
+  - merge-gate
+  greenlightRequiredFor:
+  - approve-merge
+  - external-comment
+  - destructive
+  reviewRequired: false
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: reviewer:gate
+- id: qa
+  name: QA
+  role: Browser / Workflow / CLI Smoke Verification
+  specialty: browser QA, workflow smoke tests, expected-vs-actual checks, regression reproduction
+  model: GPT-5.5
+  mission: Verify user-visible behavior with concrete smoke evidence and concise bug reports.
+  profile: qa
+  modes:
+  - smoke
+  tools:
+  - browser
+  - terminal
+  - file
+  - vision
+  - gbrain
+  - session_search
+  - skills
+  - web
+  skills:
+  - qa-core
+  - browser-harness-power-use
+  - dogfood
+  - gstack-for-hermes
+  capabilities:
+  - browser-qa
+  - smoke-verification
+  - regression-reproduction
+  - cli-verification
+  - evidence-capture
+  preferredTaskTypes:
+  - qa
+  - smoke
+  - browser
+  - verification
+  - regression
+  greenlightRequiredFor:
+  - destructive
+  - external-write
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: qa:smoke
+- id: researcher
+  name: Researcher
+  role: Brain-first Research / Bounded Autoresearch
+  specialty: GBrain-first lookup, external research, synthesis, source trails, bounded research loops
+  model: GPT-5.5
+  mission: Produce decision-grade research with brain-first context, external verification, and explicit uncertainty.
+  profile: researcher
+  modes:
+  - quick
+  - autoresearch
+  tools:
+  - gbrain
+  - web
+  - browser
+  - terminal
+  - file
+  - vision
+  - session_search
+  - skills
+  - todo
+  skills:
+  - researcher-core
+  - gbrain
+  - autoresearch
+  - browser-harness-power-use
+  - gstack-for-hermes
+  - researcher-quick
+  - researcher-autoresearch
+  - arxiv
+  - youtube-content
+  - polymarket
+  capabilities:
+  - research
+  - synthesis
+  - source-verification
+  - gbrain-query
+  - autoresearch
+  preferredTaskTypes:
+  - research
+  - analysis
+  - options
+  - source-review
+  - market-map
+  greenlightRequiredFor:
+  - publish
+  - external-send
+  - long-running-loop
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: researcher:quick
+- id: ops-watch
+  name: Ops Watch
+  role: Local Infra / Runtime Health Watch
+  specialty: gateway health, cron, MCP, workspace services, local process status, boring reliability
+  model: GPT-5.5
+  mission: Keep Hermes, Workspace, GBrain, gateway, cron, and local services observable and healthy with quiet, low-risk checks.
+  profile: ops-watch
+  modes:
+  - health
+  tools:
+  - terminal
+  - cronjob
+  - file
+  - gbrain
+  - skills
+  - session_search
+  - web
+  skills:
+  - ops-watch-core
+  - gbrain
+  - hermes-agent
+  - systematic-debugging
+  - webhook-subscriptions
+  capabilities:
+  - ops-health
+  - gateway-status
+  - runtime-monitoring
+  - cron
+  - mcp-health
+  - lifecycle-sweep
+  preferredTaskTypes:
+  - ops
+  - health
+  - monitoring
+  - lifecycle
+  - incident-triage
+  greenlightRequiredFor:
+  - restart-service
+  - destructive
+  - config-change
+  - credential-change
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: ops:health
+- id: maintainer
+  name: Maintainer
+  role: Upstream Dependency / Patch Hygiene
+  specialty: upstream tracking, local patch hygiene, dependency updates, PR/issue follow-through
+  model: GPT-5.5
+  mission: Keep local forks and upstream dependencies healthy without losing local patches or dirty-worktree context.
+  profile: maintainer
+  modes:
+  - check
+  tools:
+  - terminal
+  - file
+  - web
+  - browser
+  - gbrain
+  - session_search
+  - skills
+  skills:
+  - maintainer-core
+  - github-repo-management
+  - github-pr-workflow
+  - github-issues
+  - github-code-review
+  - gbrain
+  - gstack-for-hermes
+  - hermes-agent
+  capabilities:
+  - upstream-tracking
+  - dependency-maintenance
+  - patch-hygiene
+  - git-review
+  - issue-tracking
+  preferredTaskTypes:
+  - maintenance
+  - upstream
+  - dependency
+  - patch
+  - release-notes
+  greenlightRequiredFor:
+  - merge
+  - push
+  - destructive
+  - fork-reset
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: maintainer:check
+- id: strategist
+  name: Strategist
+  role: Wedges / Bets / Kill Criteria
+  specialty: operating plans, gstack-style wedges, strategy reviews, decision framing, kill criteria
+  model: GPT-5.5
+  mission: Turn ambiguity into crisp wedges, bets, constraints, and kill criteria without over-planning.
+  profile: strategist
+  modes:
+  - review
+  tools:
+  - gbrain
+  - web
+  - session_search
+  - file
+  - skills
+  - todo
+  - clarify
+  skills:
+  - strategist-core
+  - gstack-for-hermes
+  - gbrain
+  - writing-plans
+  - polymarket
+  capabilities:
+  - strategy
+  - planning
+  - ideation
+  - decision-framing
+  - kill-criteria
+  preferredTaskTypes:
+  - strategy
+  - planning
+  - review
+  - ideation
+  - prioritization
+  greenlightRequiredFor:
+  - publish
+  - external-send
+  - irreversible-decision
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: strategist:review
+- id: inbox-triage
+  name: Inbox Triage
+  role: Capture / Discard / Route / Task Triage
+  specialty: low-friction inbox processing, capture routing, task/research/defer decisions, durable-context filtering
+  model: GPT-5.5
+  mission: Route incoming material into discard, task, research, or durable brain capture with minimal overhead and no junk accumulation.
+  profile: inbox-triage
+  modes:
+  - triage
+  tools:
+  - gbrain
+  - web
+  - file
+  - session_search
+  - todo
+  - skills
+  - terminal
+  skills:
+  - inbox-triage-core
+  - gbrain
+  - obsidian-markdown
+  - gstack-for-hermes
+  - defuddle
+  - youtube-content
+  capabilities:
+  - inbox-triage
+  - capture-routing
+  - task-routing
+  - knowledge-filtering
+  - source-review
+  preferredTaskTypes:
+  - triage
+  - inbox
+  - capture
+  - task-routing
+  - reading-queue
+  greenlightRequiredFor:
+  - delete
+  - purge
+  - publish
+  - external-send
+  maxConcurrentTasks: 1
+  acceptsBroadcast: true
+  plugins: []
+  pluginToolsets: []
+  mcpServers:
+  - gbrain
+  wrapper: inbox:triage