code-rag: Semantic Code Search MCP Server

Search your codebase using natural language. Built as an MCP server for AI coding assistants.

What It Does

code-rag is a semantic code search engine that lets you query code with natural language:

"how are users fetched from the database"
"feature flag check pattern"
"authentication middleware"

Instead of exact keyword matching, it understands the meaning of your query and finds relevant code.

Features

| Feature | Description | |---------|-------------| | Natural Language Search | Query code semantically | | Related Files | Find files semantically related to a given file | | Context Expansion | Get surrounding code for search results | | Multi-Repo Support | Index multiple repositories | | MCP Protocol | Works with Claude, OpenCode, and other MCP clients | | Local Execution | 100% local - your code never leaves your machine |

MCP Tools

| Tool | Description | |------|-------------| | search_code | Search code with natural language | | find_related | Find files related to a given file | | get_context | Expand context around a search result | | index_stats | View indexing statistics |

Installation

cd code-rag
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Configure repos in config.yaml, run the indexer, then start the server:

./run-indexer.sh
python3 server.py

Requirements

• Python 3.10+
• ChromaDB
• MCP-compatible client

License

MIT