Documentation Index
This is the authoritative file map for the first-party OpenClaw workspace docs.
If a doc is not listed here as active, do not assume it is canonical. Check the audit register first.
Anti-drift rule:
- when code or config changes materially, update the appropriate existing canonical
.mdfile in the same change set and reference the relevant paths where useful
Capability-truth rule:
- all declared non-template agents remain in scope for full capability uplift
- do not let current partial runtime truth silently become the documented end state
- when agent capability, task promotion, service truth, or operator exposure changes, review the full canonical capability pack, not only the single file you touched
Authority Order
Use this order when determining documentation truth:
- active runtime code and live config
- canonical anchors (
../OPENCLAW_CONTEXT_ANCHOR.md,../README.md, this index, current KB truth docs) - supporting docs and runbooks
- historical or snapshot docs
If a supporting or historical document conflicts with runtime code, it is not canonical.
Canonical Navigation
| File | Purpose |
|---|---|
| README.md | docs entrypoint |
| NAVIGATION.md | role-based routes |
| SUMMARY.md | current docs status summary |
| operations/DOCUMENT_AUDIT.md | canonical vs historical classification |
Runtime And Operations
| File | Purpose |
|---|---|
| OPERATOR_GUIDE.md | operator-facing runtime behavior |
| GOVERNANCE_REPO_HYGIENE.md | cleanup and protection policy |
| operations/KNOWLEDGE_MIRROR_POLICY.md | policy for mirrored upstream knowledge sources |
| operations/public-release.md | sanitized public mirror workflow |
| operations/SPRINT_TO_COMPLETION.md | Orchestrator-first Sprint To Completion subordinate execution tracker for the root anchor |
| operations/deployment.md | production operations checklist |
| operations/backup-recovery.md | recovery guidance |
Guides
| File | Purpose |
|---|---|
| guides/configuration.md | orchestrator_config.json and env settings |
| guides/running-agents.md | agent execution guidance |
| guides/monitoring.md | monitoring and health checks |
| guides/adding-tasks.md | extending the task surface |
Technical References
| File | Purpose |
|---|---|
| concepts/architecture.md | technical system explanation |
| architecture/AGENT_CAPABILITY_MODEL.md | target maturity model for agent capability and promotion gating |
| architecture/AGENT_CAPABILITY_IMPLEMENTATION_MATRIX.md | concrete per-agent implementation and promotion matrix for the full declared agent set |
| architecture/AGENT_ADAPTATION_PLAN.md | implementation-ready adaptation plan for strengthening the current agent portfolio using proven role patterns without changing the current runtime architecture |
| architecture/OPERATOR_SURFACE_CAPABILITY_MATRIX.md | current runtime/operator exposure matrix for routes, tasks, and agents |
| architecture/OPERATOR_CONSOLE_AUDIT_AND_SPEC.md | backend-first audit and redesign spec for the private operator console |
| architecture/DOCUMENTATION_SITE_INFORMATION_ARCHITECTURE.md | published docs-site structure and downstream publishing model |
| reference/api.md | API and code-facing reference |
| reference/task-types.md | task allowlist reference |
| reference/state-schema.md | runtime state summary |
| WEBHOOK_SIGNING_CONTRACT.md | webhook HMAC contract |
Mandatory Working Sets
Use the smallest relevant pack below before making material changes.
Agent Capability / Task Promotion
Read in this order:
- ../OPENCLAW_CONTEXT_ANCHOR.md
- architecture/AGENT_CAPABILITY_MODEL.md
- architecture/AGENT_CAPABILITY_IMPLEMENTATION_MATRIX.md
- architecture/AGENT_ADAPTATION_PLAN.md when the work is specifically about adapting proven external role patterns into the current agent set without adding new task lanes
- architecture/OPERATOR_SURFACE_CAPABILITY_MATRIX.md
- reference/task-types.md
- reference/api.md
Use this pack when:
- promoting tasks into operator-facing profiles
- changing agent readiness or capability claims
- changing
/operatororoperator-s-consoletask exposure - updating the documented understanding of what OpenClaw has actually built so far versus what is still partial
Operator Surface / UX Wiring
Read in this order:
- ../OPENCLAW_CONTEXT_ANCHOR.md
- architecture/OPERATOR_SURFACE_CAPABILITY_MATRIX.md
- reference/api.md
- reference/task-types.md
- OPERATOR_GUIDE.md
Use this pack when:
- wiring new backend routes into
/operator - wiring or trimming routes in
operator-s-console - deciding whether a route belongs in a private operator surface
Start And Troubleshooting
| File | Purpose |
|---|---|
| start/getting-started.md | docs-local onboarding |
| start/quickstart.md | short checklist |
| start/architecture-overview.md | non-technical overview |
| troubleshooting/common-issues.md | common fixes |
| troubleshooting/debugging.md | deeper debugging |
Historical Or Snapshot Docs
These are not canonical, but they are still useful as historical evidence:
- ../SPRINT_HARDENING_BASELINE.md
- operations/DOCUMENTATION_COMPLETE.md
- operations/IMPLEMENTATION_COMPLETE.md
- operations/OPERATOR_S_CONSOLE_CUTOVER_BLUEPRINT.md
- operations/orchestrator_documentation.md
- operations/orchestrator-status.md
- operations/orchestrator_workflow_plan.md
- operations/PRD_GOVERNANCE_REMEDIATION.md
../orchestrator/docs/hostile-audit/*should also be treated as historical audit evidence, not current runtime truth.
Retired proof-surface docs were removed from main; use Git history if you need the old milestone ingest or proof-lane runbooks.
Root Workspace Companions
Published Docs Site
The published docs site is scaffolded from canonical repo docs, not maintained as a separate doc tree.
- local dev:
npm run docs:site:dev - production build:
npm run docs:site:build