https://github.com/hashi-ai-dev/openclaw-clawdoc
ClawDoc — OpenClaw configuration expert & system doctor. An agent that fully understands, audits, patches, fixes, and manages any OpenClaw config or issue.
https://github.com/hashi-ai-dev/openclaw-clawdoc
ai-agents chatbot configuration-management documentation knowledge-base openclaw
Last synced: about 2 hours ago
JSON representation
ClawDoc — OpenClaw configuration expert & system doctor. An agent that fully understands, audits, patches, fixes, and manages any OpenClaw config or issue.
- Host: GitHub
- URL: https://github.com/hashi-ai-dev/openclaw-clawdoc
- Owner: Hashi-Ai-Dev
- License: mit
- Created: 2026-04-14T12:56:45.000Z (2 months ago)
- Default Branch: master
- Last Pushed: 2026-06-23T16:54:23.000Z (3 days ago)
- Last Synced: 2026-06-23T18:28:25.198Z (3 days ago)
- Topics: ai-agents, chatbot, configuration-management, documentation, knowledge-base, openclaw
- Language: Python
- Homepage: https://github.com/Hashi-Ai-Dev/openclaw-clawdoc
- Size: 6.48 MB
- Stars: 7
- Watchers: 1
- Forks: 1
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- Funding: .github/FUNDING.yml
- License: LICENSE
- Code of conduct: CODE_OF_CONDUCT.md
- Audit: AUDIT.md
- Codeowners: .github/CODEOWNERS
- Security: SECURITY.md
Awesome Lists containing this project
README
# ClawDoc
**The OpenClaw knowledge-base agent for config audits, troubleshooting, and grounded system fixes.**
[](https://github.com/openclaw/openclaw)
[](#skill-tree)
[](#reference-docs)
[](LICENSE)
🧠 **Same idea, different agent:** [HermitDoc](https://github.com/Hashi-Ai-Dev/HermitDoc) — the ClawDoc philosophy applied to Hermes Agent
---
## Contents
- [What is ClawDoc?](#what-is-clawdoc)
- [Which install mode?](#which-install-mode)
- [Who is this for?](#who-is-this-for)
- [Install](#install)
- [Use it](#use-it)
- [Reference docs](#reference-docs)
- [Skill tree](#skill-tree)
- [Ready-to-use examples](#ready-to-use-examples)
- [Community](#community)
- [For contributors](#for-contributors)
- [Operating principles](#operating-principles)
---
## What is ClawDoc?
ClawDoc is a specialized OpenClaw agent that knows the entire OpenClaw system inside and out. It handles configuration auditing, plugin integration, troubleshooting, memory setup, multi-agent design, and channel configuration — with precision, not guesswork.
> [!NOTE]
> ClawDoc is a community project. It is **not affiliated with or endorsed by** OpenClaw.
> [!TIP]
> Every answer is grounded in the actual OpenClaw schema and docs. Before/after diffs, exact config patches, and real command references — always.
---
## Which install mode?
ClawDoc supports two adoption paths. Pick the one that fits your setup:
| | Mode 1 — Persistent Agent | Mode 2 — Skills Only |
|---|---|---|
| **What it does** | Creates a dedicated ClawDoc agent with its own workspace | Adds ClawDoc skills to your existing agent |
| **Best for** | Serious ongoing maintenance, system doctor use | Quick OpenClaw help in an existing agent |
| **New agent created?** | ✅ Yes | ❌ No |
| **Separate identity?** | ✅ Yes | ❌ Your agent keeps its identity |
| **Guide** | [AGENT_INSTALL.md](./AGENT_INSTALL.md) | [SKILLS_INSTALL.md](./SKILLS_INSTALL.md) |
Not sure which to pick? Start with **Mode 2 — Skills Only** for the lightest path. Switch to Mode 1 if you want ClawDoc to have its own persistent workspace and identity.
---
## Who is this for?
ClawDoc is for OpenClaw operators who need reliable help with:
- setting up channels and providers
- debugging config and runtime issues
- adding memory, tools, plugins, and agents
- generating safe config patches instead of guessing
## Scope
**ClawDoc is for OpenClaw only.** It will help with — and only with — OpenClaw configuration, troubleshooting, plugin integration, memory setup, and agent design.
**What ClawDoc will not do:**
- Help with non-OpenClaw frameworks (LangChain, AutoGen, CrewAI, etc.) — it will redirect you back to OpenClaw
- Build custom plugins from scratch unless you explicitly ask
- Execute destructive commands (`rm -rf`, mass git push, gateway reset) without your explicit confirmation
- Hold credentials or secrets in conversation state
If you ask about something outside OpenClaw, ClawDoc will say so rather than guess.
## How ClawDoc routes requests
```mermaid
flowchart TD
Q[Operator question] --> R{openclaw-master
routes by area}
R -->|config| SC[openclaw-config]
R -->|memory| SM[openclaw-memory]
R -->|channels| SCh[openclaw-channels]
R -->|agents| SA[openclaw-agents]
R -->|plugins| SP[openclaw-plugins]
R -->|troubleshoot| ST[openclaw-troubleshooting]
R -->|CLI| SCli[openclaw-cli]
R -->|providers| SPr[openclaw-providers]
R -->|install| SIns[openclaw-install]
SC --> Ref[references/
ground truth]
SM --> Ref
SCh --> Ref
SA --> Ref
SP --> Ref
ST --> Ref
SCli --> Ref
SPr --> Ref
SIns --> Ref
Ref --> A[Cited answer
+ before/after diff]
```
Every routing decision is grounded in `references/` — never from memory alone.
---
## Install
**AI-agent install (recommended):**
```
"Install ClawDoc from https://github.com/Hashi-Ai-Dev/openclaw-clawdoc"
```
Your agent reads the repo, picks up all 24 skills, and is ready to help. That's it.
**Manual install — choose your mode:**
- [AGENT_INSTALL.md](./AGENT_INSTALL.md) — Mode 1: persistent dedicated ClawDoc agent
- [SKILLS_INSTALL.md](./SKILLS_INSTALL.md) — Mode 2: add ClawDoc skills to an existing agent
**Need help getting started?** → [QUICKSTART.md](./QUICKSTART.md) (10 min)
---
## Use it
```
@your-agent [any OpenClaw config question]
```
ClawDoc routes to the right skill, reads the reference docs, and gives you a precise, grounded answer.
---
## Reference docs
**622 docs** copied from the official OpenClaw source and versioned against the tracked OpenClaw release.
| Area | What's covered |
|------|---------------|
| Config | All gateway config keys, defaults, secrets, retry, failover |
| Memory | builtin / QMD / Honcho setup, embeddings, citations |
| Agents | Multi-agent, bindings, sandbox, tool policies |
| Channels | Discord, Telegram, WhatsApp, Slack, Signal, and 36+ more (41 total) |
| Concepts | Architecture, session, compaction, streaming, queue |
| Providers | 60+ model providers: OpenAI, Anthropic, Gemini, Bedrock, Ollama... |
| CLI | Every openclaw command with examples |
| Troubleshooting | Error codes, diagnostic flows, common fixes |
---
## Skill tree
### Core
| Skill | What it does |
|-------|-------------|
| `openclaw-master` | Top-level routing — maps your question to the right skill |
| `openclaw-config` | Gateway config reference — all keys, all patterns |
| `openclaw-memory` | Memory backends: builtin, QMD, Honcho |
| `openclaw-agents` | Multi-agent, bindings, sandbox, tool policies |
| `openclaw-channels` | Discord, Telegram, WhatsApp, Slack, Signal — and 36+ more (41 total) |
### Operations
| Skill | What it does |
|-------|-------------|
| `openclaw-troubleshooting` | Diagnosis flows, error codes, common fixes |
| `openclaw-automation` | Cron, hooks, tasks, Task Flow |
| `openclaw-logging` | Logging configuration and management |
| `openclaw-gateway` | Gateway HTTP API, telemetry, secrets, sandbox, runbooks |
### Channels & Platforms
| Skill | What it does |
|-------|-------------|
| `openclaw-nodes` | Mobile/desktop node pairing and device management |
| `openclaw-platforms` | Docker, Railway, Fly, Raspberry Pi, DigitalOcean... |
| `openclaw-install` | Install guides for all platforms |
| `openclaw-start` | First-run wizard and onboarding |
| `clawdoc-onboarding` | Guided setup for new ClawDoc users |
### Tools & Providers
| Skill | What it does |
|-------|-------------|
| `openclaw-tools` | exec, browser, cron, sessions, subagents, ACP |
| `openclaw-clawhub` | ClawHub registry: discover, install, publish, verify skills and plugins |
| `openclaw-providers` | Model providers: 60+ individual files: OpenAI, Anthropic, Gemini, Bedrock, Ollama, DeepSeek, Groq, Together, Cohere... |
| `openclaw-cli` | All openclaw CLI commands |
| `openclaw-web` | Web UI, dashboard, TUI, webchat |
| `openclaw-prose` | OpenProse: markdown-first multi-agent workflow language (`/prose`, `.prose` files) |
### Concepts & Help
| Skill | What it does |
|-------|-------------|
| `openclaw-concepts` | Architecture, session, compaction, streaming, SOUL.md, OAuth, presence, OpenClaw SDK |
| `openclaw-plugins` | Plugin slots, SDK, hook system, ClawHub registry |
| `openclaw-help` | FAQ, help commands, usage patterns |
| `openclaw-reference` | Catch-all reference: tokens, transcripts, SecretRef, release |
---
## Ready-to-use examples
Apply any example with:
```bash
openclaw config merge examples/NAME.json && openclaw gateway restart
```
| Scenario | Example |
|----------|---------|
| Just installed — verify it works | `install-verify.json` |
| Discord bot, single server | `discord-single.json` |
| Discord full-featured (threads + exec) | `discord-full.json` |
| Discord + Telegram together | `discord-telegram.json` |
| Voice output (TTS) | `tts-minimax.json` |
| Conversation memory (builtin) | `memory-builtin.json` |
| Semantic search over your files | `memory-qmd.json` |
| Full memory with external search | `memory-honcho.json` |
| Full Honcho memory layer | `honcho-memory.json` |
| Different agents per Discord channel | `multi-agent-discord.json` |
| Locked-down sandboxed agent | `per-agent-sandbox.json` |
| Receive webhooks | `webhook-basic.json` |
| SMS via Twilio | `sms-channel.json` |
| WeChat (Tencent iLink) | `wechat-channel.json` |
| iMessage (native imsg) | `imessage-native.json` |
| Signal messenger | `signal-channel.json` |
| Slack workspace | `slack-channel.json` |
| Matrix homeserver | `matrix-channel.json` |
| Microsoft Teams | `msteams-channel.json` |
| WhatsApp via Baileys | `whatsapp-channel.json` |
| Zalo Bot API | `zalo-channel.json` |
| Skill Workshop (review proposals) | `skill-workshop.json` |
| Codex harness (ACP agent) | `codex-harness.json` |
| Production deploy (Tailscale + sandbox) | `production-deploy.json` |
**Beginner path:** `install-verify.json` → `discord-single.json` → `memory-builtin.json`
---
## Community
- 📖 [OpenClaw Docs](https://docs.openclaw.ai)
- 💬 [Discord](https://discord.com/invite/clawd)
- 🛒 [ClawHub](https://clawhub.ai) — find new skills
- 🐙 [Source](https://github.com/openclaw/openclaw)
---
## For contributors
See [CONTRIBUTING.md](./CONTRIBUTING.md) for conventions, style guide, and how to add new skills or reference docs.
---
## Operating principles
- **Precision over speed** — Quote the schema, cite the docs, show the exact patch.
- **No hand-waving** — If we're not sure, we say so and investigate.
- **Show your work** — Before/after diffs make answers learnable.
- **Community-minded** — Design for clarity and generalizability, not just your own setup.