openclaw-master

Files (8)

• SKILL.md
• control-ui.md · references/
• docker-vm-runtime.md · references/
• docker.md · references/
• exe-dev.md · references/
• faq.md · references/
• overview.md · references/
• vps.md · references/

SKILL.md (first lines)

## Routing hints

You should route to this skill when the user has a config question that doesn't map to a more specific skill — open-ended "how do I…" questions, system doctor queries, gateway diagnostics, streaming or queue mode questions, model failover, sandbox vs tool policy, or general architecture questions. References: `architecture.md`, `streaming.md`, `queue.md`, `model-failover.md`, `sandbox-vs-tool-policy-vs-elevated.md`, `compaction.md`, `session.md`, `agent-loop.md`, `system-prompt.md`, `soul.md`, `taskflow.md`, `oauth.md`.


# OpenClaw Master Reference

Comprehensive OpenClaw knowledge base. This is the top-level skill that routes to specialized skills.

> **Scope note:** ClawDoc follows an OSS-first scope fence — internal-only modules, private runtimes, and per-host paths are never published to this repo. Each release is reviewed against this fence before tagging.

## Skill map

The 23 routable OpenClaw skills, grouped by area:

| Skill | When to use |
|-------|-------------|
| **Config + Gateway** | |
| `openclaw-config` | Config audits, gateway config keys, secrets, retry, model failover, timezone, auth profiles, authentication |
| `openclaw-gateway` | Gateway operations, HTTP/API surface (OpenAI-compatible, OpenResponses), observability (OTel, Prometheus), secrets, sandbox vs tool policy vs elevated, security audit, gateway exposure (Tailscale/runbook) |
| `openclaw-logging` | Logging configuration: log levels, log rotation, log destinations, log format, debugging log output |
| **Memory** | |
| `openclaw-memory` | Memory backends (builtin/QMD/Honcho), embedding providers, active memory, dreaming, memory search |
| **Channels** | |
| `openclaw-channels` | Discord, Telegram, WhatsApp, Slack, Signal, Matrix, iMessage (native imsg), routing, pairing, broadcast groups; historical BlueBubbles→iMessage migration guide at references/bluebubbles.md |
| **Agents + Multi-agent** | |
| `openclaw-agents` | Multi-agent setup, bindings, sandbox, tool policies, exec approvals |
| **Plugins** | |
| `openclaw-plugins` | Plugin slots, installing plugins, plugin config, context engine, plugin manifest, SDK, ClawHub registry |
| **Tools + Workflows** | |
| `openclaw-tools` | Tool reference: exec, browser, cron, sessions, subagents, ACP, slash commands, Lobster, Diff viewer, Skill Workshop, media generation |
| `openclaw-automation` | Cron jobs, hooks, tasks, webhooks, polls, standing orders, Task Flow, Clawflow, Gmail Pub/Sub |
| `openclaw-prose` | OpenProse markdown-first workflow language for multi-agent AI sessions (plugin + slash command + skill pack) |
| `openclaw-clawhub` | ClawHub public registry: discover, install, publish, and verify skills and plugins; clawhub CLI |
| **CLI + Reference** | |
| `openclaw-cli` | CLI commands: status, gateway, plugins, memory, agents, channels, config, sessions, cron, hooks, pairing, ACP, MCP, secrets, doctor, update, skills |
| `openclaw-reference` | General OpenClaw reference: token use, transcript hygiene, session compaction, rich output protocol, SecretRef credential surface, RELEASING |
| **Concepts + Help** | |
| `openclaw-concepts` | Architecture, session, compaction, streaming, bootstrap, queue modes, agent loop, system prompt, SOUL.md, Task Flow, OAuth, presence, OpenClaw SDK |
| `openclaw-help` | Help, FAQ, "how do I..." plain-language entry point, testing |
| **Install + Platforms + Start** | |
| `openclaw-install` | Install, update, migrate, uninstall. Docker, Railway, Fly, GCP, Azure, Kubernetes, Hetzner, Podman, Bun, Nix, from source |
| `openclaw-platforms` | OpenClaw on Android, iOS, macOS, Windows, Linux, Raspberry Pi, Oracle Cloud, DigitalOcean |
| `openclaw-start` | First run, bootstrapping, onboarding OpenClaw. Setup wizard, workspace bootstrap, `openclaw onboard` CLI |
| **Surfaces** | |
| `openclaw-nodes` | Pairing and using companion nodes (iOS, Android, macOS). Camera, canvas, audio, talk, location, notifications, screen |
| `openclaw-web` | Web UI: dashboard, TUI, webchat, control UI |
| **Providers** | |
| `openclaw-providers` | Model providers: OpenAI, Anthropic, Gemini, Bedrock, Ollama, DeepSeek, Groq, Together, and 60+ more |
| **Troubleshooting** | |
| `openclaw-troubleshooting` | Diagnosis, error codes, ACP, hooks, pairing, channel failures, gateway runbook |
| **Different domain — not an OpenClaw skill** | |
| `clawdoc-onboarding` | Guided first-run setup for **ClawDoc itself** (NOT OpenClaw). Walks through installing ClawDoc skills, configuring docs sync, end-to-end verification |

## Quick reference

### Docs URLs (always current)
| Topic | URL |
|---|---|
| Config reference | https://docs.openclaw.ai/gateway/configuration-reference |
| CLI commands | https://docs.openclaw.ai/cli/index |
| Channels | https://docs.openclaw.ai/channels/index |
| Providers | https://docs.openclaw.ai/providers/index |
| Tools | https://docs.openclaw.ai/tools/index |
| Troubleshooting | https://docs.openclaw.ai/help/troubleshooting |
| OpenClaw docs | https://docs.openclaw.ai |
| ClawHub (skills) | https://clawhub.ai |

### Usage quick reference
| Situation | Action |
|---|---|
| Config changes | Use gateway tool (`config.get` / `config.patch` / `config.apply`) |
| Skill installation | `clawhub` CLI or clone to `~/.openclaw/skills/` |
| Bot mention in Discord | `<@bot-id>` not plain `@name` |
| `requireMention: true` | Respond only when pinged |
| `requireMention: false` | Always open in that channel |
| Bot-to-bot messaging | Requires `allowBots: "mentions"` on receiving side |

### Config location
```
~/.openclaw/openclaw.json
```

This is the user-level config path for typical OpenClaw installs (Homebrew, pip, source). If you are running OpenClaw inside a container image that sets `OPENCLAW_HOME=/data/.openclaw`, the config path is `/data/.openclaw/openclaw.json` instead. Check `openclaw doctor` to see which path your install uses.

### Logs
```
/tmp/openclaw/openclaw-YYYY-MM-DD.log
```

### Gateway management
```bash
openclaw status              # check status
openclaw gateway restart    # restart (confirm first!)
ps aux | grep openclaw-gateway  # is it running?
```

### Validate config
```bash
python3 -m json.tool ~/.openclaw/openclaw.json > /dev/null && echo "valid"
```

### Key memory fix (Honcho)
```json5
plugins: { slots: { memory: "openclaw-honcho" } }
```

### Key config paths
- `agents.defaults.workspace` — agent working directory
- `agents.defaults.model` — primary + fallback models
- `agents.defaults.memorySearch` — embedding provider
- `memory.backend` — `"builtin"` or `"qmd"`
- `plugins.slots.memory` — memory plugin override
- `channels.discord.token` — bot token
- `tools.elevated.enabled` — elevated exec

### Common error fixes
- **"plugin disabled (memory slot)"** → set `plugins.slots.memory`
- **"Port already in use"** → gateway already running, don't restart

From Hashi-Ai-Dev/openclaw-clawdoc, MIT.

More from Hashi-Ai-Dev/openclaw-clawdoc