Onboarding · Day 0 → Day 7
New to OpenClaw? Start here. 12 sections covering connecting, the architecture, the agents, and your first safe interaction.
OpenClaw Team Docs
Internal team documentation for the OpenClaw multi-agent platform.
Where to start
Section titled “Where to start”How chat-driven execution works
The 8-stage flow from prompt to response, with 5 worked examples — log read, memory search, Sonnet dispatch, skill invocation, hook blocking.
Visual maps
8 SVG diagrams covering master architecture, hooks, MCPs, model routing, memory pipeline, agent dispatch, skill triggers.
Skills primer
What skills are and how slash commands + keyword auto-triggers fire workflows. Includes the trigger table summary.
Architecture snapshot
Live state of the VPS — services running, ports bound, agents active, memory + storage state.
Browse the full vault
Quartz-rendered vault browse with backlinks + graph view across the 474-page publish-allowlist subset of the Obsidian vault.
What’s NOT here
Section titled “What’s NOT here”This site is read-only orientation for the team. It doesn’t cover:
- Writing or modifying agent code
- Restarting services or modifying live infra
- Managing credentials or modifying
master.env - Running
/gsd:execute-phaseon someone else’s plan
Those operations stay with Henry (the founder/operator) until you’re explicitly authorized.
If you spot something stale, broken, or wrong — message Henry directly or post in #ops on Slack.