| name | openclaw-docs |
| description | Use when the user asks about OpenClaw documentation sources, how to access crawled docs, where to find official documentation, how the documentation sync system works, or when referencing OpenClaw product docs for any task.
|
OpenClaw Documentation
The plugin maintains a local cache of official OpenClaw documentation from two upstream sources. This enables all skills and agents to reference authoritative, up-to-date product docs.
Documentation Sources
1. docs.openclaw.ai (Primary)
The official documentation site. Provides a machine-readable index at:
https://docs.openclaw.ai/llms.txt
This llms.txt file lists ~431 pages using Markdown link syntax:
- [Cron Jobs](https://docs.openclaw.ai/automation/cron-jobs)
- [Telegram Setup](https://docs.openclaw.ai/channels/telegram)
Categories: automation, channels, cli, concepts, gateway, install, platforms, plugins, providers, reference, security, start, tools, web
2. GitHub Repository
Source code and docs at https://github.com/openclaw/openclaw/tree/main/docs.
Raw content available at:
https://raw.githubusercontent.com/openclaw/openclaw/main/docs/<path>
Useful for: release notes, contribution guides, source-level documentation not on the docs site.
Local Storage
Two-Tier Architecture
plugins/openclaw/
docs/ # Committed (metadata only)
docs-manifest.json # Tracks fetched pages, timestamps, hashes
INDEX.md # Categorized index with fetch status
FETCH-CONFIG.md # Source URLs, priority tiers, staleness rules
.crawled/ # Gitignored (regenerable content)
docs.openclaw.ai/ # Fetched from docs site
automation/
cron-jobs.md
webhooks.md
channels/
telegram.md
whatsapp.md
cli/
commands.md
...
github/ # Fetched from GitHub
docs/
start/
cli/
Why two tiers? The docs/ directory is committed to git so marketplace installs know what documentation exists. The .crawled/ directory is gitignored because it contains regenerable content that can be re-fetched with /oc-docs sync.
Accessing Docs
To read a cached documentation page:
Read .crawled/docs.openclaw.ai/<category>/<filename>.md
Examples:
- Cron jobs:
Read .crawled/docs.openclaw.ai/automation/cron-jobs.md
- Telegram setup:
Read .crawled/docs.openclaw.ai/channels/telegram.md
- CLI commands:
Read .crawled/docs.openclaw.ai/cli/commands.md
To find what's available:
- Check
docs/INDEX.md for a categorized listing
- Check
docs/docs-manifest.json for detailed metadata
Priority Tiers
| Tier | Categories | Pages | When Fetched |
|---|
| 1 | automation, channels, cli, concepts, gateway, security | ~148 | Default /oc-docs sync |
| 2 | install, providers, tools, platforms, reference, start | ~124 | --full or --category |
| 3 | web, plugins, root pages | ~13 | On explicit request only |
When .crawled/ Is Empty
If .crawled/ doesn't exist or has no content, suggest the user run:
/oc-docs sync
This fetches Tier 1 documentation (~148 core pages). For all pages:
/oc-docs sync --full
How Other Skills Should Use Docs
When a skill needs authoritative OpenClaw documentation:
- Check if the relevant page exists in
.crawled/docs.openclaw.ai/<category>/
- If it exists, read it for accurate, up-to-date information
- If it doesn't exist, fall back to the skill's embedded
references/ content
- If
.crawled/ is entirely empty, suggest /oc-docs sync before proceeding
This ensures skills provide the most current information while gracefully degrading when docs haven't been synced.