From openclaw
OpenClaw expert — self-hosted multi-channel AI agent gateway. Activates on any mention of OpenClaw, openclaw.json, ClawHub, pi-mono, or gateway/channel setup for WhatsApp/Telegram/Discord/Signal-style agent bots. Covers config, channels, providers, tools, CLI, sandboxing, automation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/openclaw:openclaw-expertThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
OpenClaw is a self-hosted, open-source (MIT) multi-channel gateway for AI agents: one Gateway process connects messaging surfaces (WhatsApp, Telegram, Discord, Slack, Signal, iMessage, Matrix, MS Teams, and ~20 more) to model providers (Anthropic, OpenAI, OpenRouter, Bedrock, Mistral, local models).
reference/INDEX.mdreference/automation.mdreference/channels.mdreference/cli.mdreference/concepts.mdreference/gateway-ops.mdreference/getting-started.mdreference/install.mdreference/nodes-media.mdreference/platforms.mdreference/plugins-extensions.mdreference/providers.mdreference/templates-reference.mdreference/tools.mdreference/troubleshooting.mdOpenClaw is a self-hosted, open-source (MIT) multi-channel gateway for AI agents: one Gateway process connects messaging surfaces (WhatsApp, Telegram, Discord, Slack, Signal, iMessage, Matrix, MS Teams, and ~20 more) to model providers (Anthropic, OpenAI, OpenRouter, Bedrock, Mistral, local models).
~/.openclaw/openclaw.json (strict validation, hot-reload)Reference files are large (19K–201K). Do NOT Read a whole file.
reference/INDEX.md — it maps every file to its [Source: <url>] page anchors).[Source: anchor URL to get line numbers: grep -n 'dmPolicy\|\[Source:' reference/channels.md[Source: anchor to the next ---).https://docs.openclaw.ai/<path> or https://docs.openclaw.ai/sitemap.xml.| File | Size | Covers |
|---|---|---|
reference/gateway-ops.md | 201K | Gateway config + reference, auth, security, sandboxing, networking, remote access, Tailscale, web UI/dashboard/TUI, protocols, health/doctor |
reference/channels.md | 192K | Every channel (WhatsApp, Telegram, Discord, Slack, Signal, iMessage, Matrix, Teams, BlueBubbles, Nostr, Twitch...), pairing, groups, broadcast, routing |
reference/concepts.md | 134K | Architecture, agent runtime/loop, system prompt, sessions, memory, compaction, multi-agent, streaming, retry, queue, models |
reference/troubleshooting.md | 134K | FAQ, triage, debug workflows, channel/gateway troubleshooting |
reference/tools.md | 128K | Lobster, LLM task, exec, web search, browser, sub-agents, skills, ClawHub, plugins, slash commands, elevated mode |
reference/install.md | 100K | npm/curl/Docker/Nix/Ansible/Bun install, updating, uninstalling |
reference/cli.md | 65K | All 44 CLI commands (gateway, configure, doctor, channels, agents, devices, cron, sessions, webhooks...) |
reference/platforms.md | 59K | macOS app internals, Linux, Windows/WSL2, iOS, Android, Hetzner, GCP, Fly |
reference/automation.md | 57K | Cron jobs, cron vs heartbeat, hooks, webhooks, Gmail PubSub, polls |
reference/templates-reference.md | 53K | AGENTS/SOUL/BOOT/BOOTSTRAP/IDENTITY/HEARTBEAT templates, RPC adapters, prompt caching |
reference/getting-started.md | 37K | Onboarding, CLI wizard, personal assistant setup, hubs |
reference/providers.md | 27K | Anthropic, OpenAI, OpenRouter, Bedrock, Mistral, LiteLLM, local models |
reference/nodes-media.md | 22K | Node management, voice notes, camera, talk mode, voice wake |
reference/plugins-extensions.md | 19K | Voice call plugin, Zalo personal, community plugins |
GATEWAY (daemon) port 18789 (WS+HTTP), Canvas 18793
Channels (WhatsApp/Telegram/...) → Agent Runtime (pi-mono) → Tools (exec/web/browser/skills) → Sessions (JSONL per agent)
WS API: connect → req/res + events
Config: ~/.openclaw/openclaw.json (JSON5, hot-reload)
~/.openclaw/
├── openclaw.json # Main config (JSON5, hot-reloaded)
├── credentials/ # Provider credentials (0o600)
├── workspace/ # Default agent workspace
│ ├── AGENTS.md SOUL.md TOOLS.md USER.md IDENTITY.md
│ ├── BOOTSTRAP.md MEMORY.md HEARTBEAT.md
│ ├── memory/ # Daily memory files (on-demand, NOT injected)
│ └── skills/ # Per-agent skills (highest precedence)
├── skills/ # Shared skills (all agents)
├── agents/<agentId>/sessions/<SessionId>.jsonl
└── sandboxes/ # Sandbox workspaces
npm install -g openclaw@latest && openclaw onboard --install-daemon
openclaw gateway [status] # Start / check service
openclaw configure # Interactive wizard
openclaw config get|set <path> [v] # Read/write config
openclaw doctor [--fix] # Diagnose + repair
openclaw channels list|login|status
openclaw agents list --bindings
openclaw devices list|approve|reject
openclaw logs --follow / status --all / health --verbose
openclaw security audit [--deep] [--fix]
// Minimal: ~/.openclaw/openclaw.json
{
agents: { defaults: { workspace: "~/.openclaw/workspace" } },
channels: { whatsapp: { allowFrom: ["+15555550123"] } },
}
// DM policy (any channel): dmPolicy: "pairing" | "allowlist" | "open" | "disabled"
// Models: agents.defaults.model: { primary: "anthropic/claude-sonnet-4-5", fallbacks: ["openai/gpt-5.2"] }
npx claudepluginhub nalyk/nalyk-skillsGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.