OpenClaw Docs Sync

Purpose

Keep a local mirror of the OpenClaw framework docs, ClawHub CLI/docs, and the Skills repository. OpenClaw's built-in QMD memory backend indexes these docs for hybrid semantic search.

When to use

Use this skill when the agent needs fresh documentation or precise references from:

• OpenClaw framework docs
• ClawHub CLI/docs
• Skills repo (all SKILL.md files)

Setup

Step 1: Run the sync script

Run the script whenever you need a fresh mirror:

# Recommended (portable)
npx tsx scripts/sync-docs.ts

# Or with Bun (fastest)
bun run scripts/sync-docs.ts

What it does:

• Mirrors OpenClaw docs/ into ~/.openclaw/docs/openclaw-docs/
• Mirrors ClawHub docs/ into ~/.openclaw/docs/clawhub-docs/
• Mirrors Skills skills/ into ~/.openclaw/docs/openclaw-skills/
• Runs security scans on downloaded content

Step 2: Add to openclaw.json

After syncing, add the path to your openclaw.json config using memory.qmd.paths:

{
  "memory": {
    "backend": "qmd",
    "citations": "auto",
    "qmd": {
      "includeDefaultMemory": true,
      "update": { "interval": "5m" },
      "paths": [
        {
          "name": "openclaw-docs",
          "path": "~/.openclaw/docs",
          "pattern": "**/*.{md,mdx}"
        }
      ]
    }
  }
}

How the agent uses synced docs

Once configured in openclaw.json, OpenClaw's QMD memory backend automatically:

1. Indexes the synced docs on startup and at regular intervals (default 5m)
2. Embeds chunks for hybrid search (BM25 + vectors + reranking)
3. Searches via memory_search tool when you ask questions
4. Retrieves exact file content via memory_get tool

The agent will cite sources like:

Source: qmd/openclaw-docs/concepts/memory.md#L45-L52

Repo structure in synced mirror

~/.openclaw/docs/
├── openclaw-docs/   # OpenClaw framework docs
├── clawhub-docs/    # ClawHub CLI docs  
└── openclaw-skills/ # All SKILL.md files from skills repo

Notes

• The sync script only ingests .md and .mdx files
• Synced docs are read-only; edit upstream repos to update
• Security scans run automatically; threats logged to ~/.openclaw/security-scan.log
• See QMD Backend docs for full configuration options