# openclaw-docs-rag

> Offline OpenClaw documentation with vector search and RAG capabilities

- Author: Charpup
- Repository: Charpup/openclaw-docs-rag
- Version: 20260210035541
- Stars: 0
- Forks: 0
- Last Updated: 2026-02-09
- Source: https://github.com/Charpup/openclaw-docs-rag
- Web: https://mule.run/skillshub/@@Charpup/openclaw-docs-rag~openclaw-docs-rag:20260210035541

---

---
name: openclaw-docs-rag
description: Offline OpenClaw documentation with vector search and RAG capabilities
author: Galatea
version: 1.0.0
---

# OpenClaw Docs RAG

Query OpenClaw documentation using natural language and vector search.

## When to Use

- Before executing tasks: Query relevant documentation for context
- Learning OpenClaw: Find specific feature explanations
- Troubleshooting: Search for error solutions and best practices

## Installation

```bash
git clone https://github.com/Charpup/openclaw-docs-rag.git ~/.openclaw/skills/openclaw-docs-rag
cd ~/.openclaw/skills/openclaw-docs-rag
npm install
```

## Setup

1. Set environment variables:
```bash
export OPENAI_API_KEY="sk-..."  # For embeddings
export MEMU_DB_PASSWORD="..."    # For vector store
```

2. Initialize database:
```bash
npm run init-db
```

3. Sync documentation:
```bash
npm run sync
```

## Tools

### queryDocs

Query OpenClaw documentation with natural language.

**Usage:**
```bash
./query-docs.sh "how to configure cron jobs"
```

**Parameters:**
- `query` (string, required): Search query
- `top_k` (int, optional): Number of results (default: 5)

**Returns:**
- Relevant documentation chunks
- Source URLs
- Similarity scores

### syncDocs

Sync documentation from docs.openclaw.ai.

**Usage:**
```bash
./sync-docs.sh
```

**Parameters:**
- `force` (boolean, optional): Force full re-sync (default: false)

## API Usage

```javascript
const { queryDocs, syncDocs } = require('./index.js');

// Query
const results = await queryDocs("cron jobs configuration", { topK: 5 });

// Sync
await syncDocs({ force: true });
```

## Architecture

- **Vector Store**: MemU PostgreSQL + pgvector
- **Embeddings**: text-embedding-3-small (1536d)
- **Chunk Size**: ~1000 tokens with overlap
- **Query Latency**: <500ms (p95)

## License

MIT