A production-ready, self-hosted RAG infrastructure that provides local long-term memory and hybrid search capabilities for your AI agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install clawrag
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install clawrag using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is ClawRAG Connector?
ClawRAG serves as the central intelligence hub for Openclaw Skills, offering a privacy-focused environment where your data never leaves your machine. It functions as a sophisticated RAG (Retrieval-Augmented Generation) engine that combines semantic vector search with traditional keyword matching (BM25) to provide the most relevant context to your AI agents.
Built to be MCP-native, this skill enables seamless integration between local documents and Large Language Models. By utilizing a Docker-based architecture, it ensures that your vector databases and document ingestion pipelines remain secure and entirely under your control, making it an essential tool for developers handling sensitive or proprietary information.
ClawRAG Connector Use Cases
- Creating a secure, local knowledge base for private project documentation.
- Enhancing AI agent responses with verifiable citations from uploaded PDF, Office, and Markdown files.
- Building project-specific memory collections to help Openclaw Skills maintain context across long development cycles.
- Implementing hybrid search (Semantic + BM25) to improve the accuracy of information retrieval in complex datasets.
How ClawRAG Connector Works
- The user deploys the ClawRAG infrastructure locally using Docker Compose, which initializes the API and ChromaDB.
- Documents are ingested into the system, where they are processed by Docling and stored as vector embeddings.
- The @clawrag/mcp-server acts as a bridge, connecting the Openclaw Skills environment to the local ClawRAG API.
- When a query is made, the system performs a hybrid search, ranking results via Reciprocal Rank Fusion (RRF) to provide the AI with the most precise context and source citations.
ClawRAG Connector Setup
1. Start ClawRAG Service
First, deploy the core infrastructure using Docker:
git clone https://github.com/2dogsandanerd/ClawRag.git
cd ClawRag
cp .env.example .env
docker compose up -d
2. Connect the MCP Skill
Once the health check at http://localhost:8080/health returns OK, add the connector to your environment:
openclaw mcp add --transport stdio clawrag npx -y @clawrag/mcp-server
ClawRAG Connector Data Schema & Taxonomy
ClawRAG organizes information into a structured hierarchy to ensure efficient retrieval within Openclaw Skills:
| Level | Data Type | Description |
|---|---|---|
| Collection | Container | Logical grouping for project-specific data |
| Document | File | Supports PDF, DOCX, TXT, and Markdown via Docling |
| Embedding | Vector | High-dimensional data stored in ChromaDB for semantic search |
| Metadata | JSON | Contains source tracking, citations, and original file paths |
ClawRAG Connector Advanced Features
- Hybrid Search Architecture: Combines Vector similarity with BM25 keyword matching and RRF ranking.
- Privacy-First Design: All vector processing and storage occur on the local machine.
- Smart Ingestion: Automated parsing of complex document formats like Office and PDF using Docling.
- Multi-Collection Support: Ability to segment knowledge across different projects or domains for cleaner context retrieval.
- MCP-Native: Full support for the Model Context Protocol, ensuring high compatibility with modern Openclaw Skills workflows.
SKILL.md
Loading
Related Openclaw Skills
MikroTik RouterOS Management Skill
A comprehensive AI agent skill for managing MikroTik RouterOS devices through API-driven automation and real-time monitoring.
Gaode Map (AMap) Integration
A Python-based skill for integrating Gaode Map (AMap) capabilities like POI search and multi-modal route planning into AI agent workflows.
Music Manager
A universal music downloader and manager that extracts audio from YouTube and Bilibili, converts it to MP3, and organizes it into local libraries.
Runware
A comprehensive media generation skill for creating, upscaling, and animating high-fidelity images and videos via the Runware API.
LINE Client Skill
A comprehensive LINE messaging integration for AI agents utilizing the Chrome extension gateway for JSON-based API access.
Spotify Control for macOS
A macOS-native skill for controlling Spotify playback, volume, and media selection via CLI and AppleScript.
Featured*
