一键在 Manus 中运行任何 Skill

openclaw-docs

星标3

分支0

更新时间2026年3月24日 07:13

Complete OpenClaw documentation (347 pages). Use whenever the user mentions OpenClaw, openclaw, Clawd, or any topic related to OpenClaw: installation, gateway configuration, channel setup (WhatsApp, Telegram, Discord, Slack, Signal, iMessage, IRC, Matrix), CLI commands, providers, plugins, automation (hooks, cron, webhooks, standing orders), sandboxing, multi-agent routing, voice/audio nodes, compaction, or the control dashboard. Also trigger for self-hosted AI gateways, WhatsApp/Telegram/Discord bots in an OpenClaw context, or references to docs.openclaw.ai. Never guess or web-search when this skill is available — the docs are the source of truth. Covers 20 topic areas.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

Jamessfks

Jamessfks/easycalw

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

OpenClaw Documentation Reference

You have the complete docs.openclaw.ai documentation (347 pages) available locally. Use these files to answer questions accurately — never guess or web-search when the answer is in these docs. The docs are the source of truth; always prefer them over your training data for OpenClaw-specific details.

What is OpenClaw?

OpenClaw is an open-source, self-hosted AI gateway that bridges messaging platforms (WhatsApp, Telegram, Discord, iMessage, Slack, Signal, IRC, Matrix, etc.) with AI coding agents. It provides multi-channel messaging, agent orchestration, plugin extensibility, automation (hooks, cron, webhooks), and a browser-based control dashboard.

How to Find Answers

All docs live under ${CLAUDE_SKILL_DIR}/docs/ organized by section. Follow this lookup strategy — start specific, broaden if needed:

If you know the topic area, go straight to the right section directory (see table below)
If you need to find something by keyword, use Grep on ${CLAUDE_SKILL_DIR}/docs/
If you need to browse what's available, read ${CLAUDE_SKILL_DIR}/INDEX.md (full manifest of all 347 docs)

Section Directory

Intent	Path	What's Inside
Install OpenClaw	`docs/install/`	npm, Docker, Nix, source, cloud deployments (GCP, Azure, Hetzner, Fly, Railway, etc.)
Configure the gateway	`docs/gateway/`	Configuration guide, reference, examples, sandboxing, secrets, security, Tailscale
Set up messaging channels	`docs/channels/`	WhatsApp, Telegram, Discord, iMessage, Slack, Signal, IRC, Matrix, LINE, Teams, etc.
Use CLI commands	`docs/cli/`	Full CLI reference (47 commands)
Add AI providers	`docs/providers/`	OpenAI, Anthropic, Google, Ollama, Bedrock, Groq, OpenRouter, xAI, etc. (37 providers)
Use tools & skills	`docs/tools/`	Browser, exec, slash commands, MCP, skills, sub-agents, web search, PDF, TTS
Understand concepts	`docs/concepts/`	Agents, workspaces, multi-agent routing, compaction, memory, sessions, streaming
Build plugins	`docs/plugins/`	Plugin architecture, bundles, building custom plugins, community plugins
Deploy to platforms	`docs/platforms/`	macOS, iOS, Android, Linux, Windows, Raspberry Pi
Set up automation	`docs/automation/`	Hooks, cron jobs, polls, webhooks, standing orders, Gmail PubSub
Security	`docs/security/`	Threat model (MITRE ATLAS), formal verification, contributing
Get help	`docs/help/`	FAQ, debugging, environment variables, troubleshooting
Reference	`docs/reference/`	Memory config, test framework, API costs, templates (AGENTS.md, SOUL.md, etc.)
Getting started	`docs/start/`	Getting started, onboarding, bootstrapping, showcase
Web dashboard	`docs/web/`	Control UI, dashboard setup, WebChat, TUI
Node types	`docs/nodes/`	Audio, camera, voice wake, talk mode, media understanding
Diagnostics	`docs/diagnostics/`	Feature flags and diagnostic tools
Codelabs	`docs/_codelabs/`	Tutorials (ADK streaming, multi-agent A2A)
Other	`docs/_root/`	Logging, networking, auth semantics, CI, date-time, OpenProse

Frequently Needed Docs

Task	Direct Path
Getting started	`${CLAUDE_SKILL_DIR}/docs/start/getting-started.md`
Install via Docker	`${CLAUDE_SKILL_DIR}/docs/install/docker.md`
Install via npm	`${CLAUDE_SKILL_DIR}/docs/install/npm.md`
Configuration reference	`${CLAUDE_SKILL_DIR}/docs/gateway/configuration-reference.md`
Configuration examples	`${CLAUDE_SKILL_DIR}/docs/gateway/configuration-examples.md`
WhatsApp setup	`${CLAUDE_SKILL_DIR}/docs/channels/whatsapp.md`
Discord setup	`${CLAUDE_SKILL_DIR}/docs/channels/discord.md`
Telegram setup	`${CLAUDE_SKILL_DIR}/docs/channels/telegram.md`
Slack setup	`${CLAUDE_SKILL_DIR}/docs/channels/slack.md`
CLI reference	`${CLAUDE_SKILL_DIR}/docs/cli/index.md`
Multi-agent routing	`${CLAUDE_SKILL_DIR}/docs/concepts/multi-agent.md`
Plugin architecture	`${CLAUDE_SKILL_DIR}/docs/plugins/architecture.md`
Sandboxing	`${CLAUDE_SKILL_DIR}/docs/gateway/sandboxing.md`
System prompt	`${CLAUDE_SKILL_DIR}/docs/concepts/system-prompt.md`
FAQ	`${CLAUDE_SKILL_DIR}/docs/help/faq.md`
Onboarding (CLI)	`${CLAUDE_SKILL_DIR}/docs/start/wizard.md`

Search Patterns

# Keyword search across all docs
Grep pattern="whatsapp" path="${CLAUDE_SKILL_DIR}/docs/" -i

# Search within a specific section
Grep pattern="memory" path="${CLAUDE_SKILL_DIR}/docs/gateway/"

# List all docs in a section
Glob pattern="*.md" path="${CLAUDE_SKILL_DIR}/docs/providers/"

Common Cross-References

Many questions touch topics that span multiple docs. When a user asks about one of these topics, also read the related doc — the combination gives a much better answer than either alone.

When the user asks about...	Also read...
Cron jobs / scheduled tasks	`docs/automation/cron-vs-heartbeat.md` (heartbeat is often confused with cron)
Sandboxing / code isolation	`docs/gateway/sandbox-vs-tool-policy-vs-elevated.md` (the three controls)
Multi-agent setup	`docs/cli/agents.md` (CLI for managing agents) + `docs/gateway/configuration-examples.md`
Channel setup (any)	`docs/gateway/configuration-examples.md` (copy-paste config patterns)
Hooks / webhooks	`docs/automation/standing-orders.md` (standing orders complement hooks)
Model providers / failover	`docs/concepts/model-failover.md` + `docs/concepts/model-providers.md`
Session management	`docs/concepts/compaction.md` + `docs/concepts/session-pruning.md`
WhatsApp groups	`docs/channels/group-messages.md` + `docs/channels/groups.md`
Troubleshooting (any channel)	`docs/channels/troubleshooting.md` + `docs/gateway/troubleshooting.md`
Discord bot not responding	`docs/channels/discord.md` (check Message Content Intent)
Memory / context	`docs/concepts/memory.md` + `docs/concepts/context.md` + `docs/reference/memory-config.md`
Remote access / SSH	`docs/gateway/remote.md` + `docs/gateway/remote-gateway-readme.md`

Answering Guidelines

Always cite the doc filename: for every major topic you reference, include the doc path at least once (e.g., "According to docs/gateway/sandboxing.md..."). This is how the user verifies your answer and finds more detail. Don't just use the information — name where it came from.
Include source URLs: each doc file has a source_url in its YAML frontmatter. Include these URLs so the user can read the full page on docs.openclaw.ai.
Be precise with CLI commands: OpenClaw has 47 CLI commands. When the user asks about a command, read the specific CLI doc (e.g., docs/cli/gateway.md for openclaw gateway) rather than paraphrasing from memory.
Cross-reference using the table above: many topics span multiple docs. Check the cross-reference table and read all relevant docs before answering — the user benefits from a complete picture.
Don't fabricate config fields or flags: if you're unsure about a config key or CLI flag, look it up in the configuration reference or the specific CLI doc. The docs are exhaustive — if a field isn't documented, it probably doesn't exist.

Source

All documentation was scraped from https://docs.openclaw.ai/ on 2026-03-21. Each file includes YAML frontmatter with title, source_url, and section for traceability.