RooCode-RAG-Lookup

Provides RAG-based semantic search over documents and code, returning relevant chunks with metadata.

python

0

GitHub Stars

python

Language

6 months ago

First Indexed

3 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "mazchoo-roocode-rag-lookup": {
      "command": "python",
      "args": [
        "mcp_tool.py"
      ]
    }
  }
}

RooCode-RAG-Lookup provides a Python-based MCP server that powers retrieval-augmented generation (RAG) lookups over documents and code repositories. It uses vector embeddings and semantic search to fetch relevant text chunks, helping you answer complex questions with context from your data sources.

How to use

You run the MCP server locally and connect it to your RooCode client. The server exposes a stdio-based interface so you can start it as a local process and pipe JSON-RPC messages to it. Use the Rag Lookup tool to perform semantic search across your indexed documents and repositories, and then feed the results to your LLM prompts to improve accuracy and grounding.

How to install

Prerequisites: you need Python installed on your system. You also use pip to install dependencies.

# Install Python dependencies
pip install -r requirements.txt

Configuration and startup

Configure your MCP client to use the local stdio MCP server by pointing it to the executable that runs the server. The server is started by running the Python script that implements the MCP server, and it is used via a stdio-based transport.

# Start the MCP server locally
python mcp_tool.py

Usage with MCP client

Once the server is running, you can invoke the rag_lookup tool through the MCP interface from your RooCode client. Use the client’s MCP configuration to connect to the local stdio server and invoke rag_lookup to perform semantic search across your indexed documents and repositories.

If you need to verify connectivity, you can use the simple test tool provided by the project to ensure the MCP channel is responsive.

Notes on configuration and usage

The system indexes documents by extracting text from PDFs, chunks the text, and stores embeddings in a vector store. Queries are embedded and matched against stored vectors, returning top-K relevant chunks with metadata.

Troubleshooting and tips

If you encounter issues starting the server, ensure you are in the project directory and that Python dependencies are installed. Check that the mcp_tool.py script is reachable and executable.

Security considerations

Limit access to the local MCP server as appropriate for your environment. Use secure paths for documents and repositories, and manage access to the RAG results to protect sensitive information.

Examples and quick-start flow

Extract and index documents from PDFs into the system, then query the RAG system to retrieve relevant chunks for a given topic.

Available tools

rag_lookup

Perform semantic search across documents and repos to retrieve relevant text chunks with similarity scores and metadata.

say_hello

Simple test tool that returns a greeting with timestamp to verify MCP connectivity.