Chromadb Fastapi MCP

🚀 ChromaDB FastAPI MCP Project

This project provides a FastAPI - based service integrated with ChromaDB, offering a range of APIs for collection and document management. It also supports connection to the MCP server for enhanced functionality.

🚀 Quick Start

📦 Installation

1. Clone the repository.
2. Install dependencies:

   uv sync
   

3. Copy .env.example to .env and configure the settings:

   cp .env.example .env
   

4. Edit .env with your preferred settings.

⚙️ Configuration

Configure the following environment variables in your .env file:

• CHROMA_CLIENT_TYPE: Choose either ephemeral or persistent.
• CHROMA_DATA_DIR: The data directory when using the persistent client.
• OPENAI_API_KEY: Your OpenAI API key for embeddings.

💻 Usage

Run the server:

cd /path/to/chromadb-fastapi-mcp
uv run python -m app.main

Or directly use uvicorn:

uv run uvicorn app.main:app --reload

The server will start at http://localhost:8000, and the API documentation can be viewed at http://localhost:8000/docs.

📚 Documentation

📄 API Endpoints

Collections

• GET /api/collections/ - List all collections.
• POST /api/collections/ - Create a new collection.
• GET /api/collections/{collection_name}/peek - View an overview of the contents in a collection.
• GET /api/collections/{collection_name}/info - Get collection information.
• GET /api/collections/{collection_name}/count - Get the number of documents in a collection.
• PUT /api/collections/{collection_name} - Modify a collection.
• DELETE /api/collections/{collection_name} - Delete a collection.

Documents

• POST /api/documents/add - Add documents to a collection.
• POST /api/documents/query - Query documents from a collection.
• POST /api/documents/get - Get documents from a collection.
• PUT /api/documents/update - Update documents in a collection.
• DELETE /api/documents/delete - Delete documents from a collection.

🌐 MCP Server

The MCP server repository: https://github.com/tadata-org/fastapi_mcp

Connect to the MCP Server using SSE

After running your FastAPI application:

1. In Cursor -> Settings -> MCP, use your MCP server endpoint URL (e.g., http://localhost:8000/mcp) as sse.
2. Cursor will automatically discover all available tools and resources.

Connect to the MCP Server using mcp - proxy stdio

Configure on Claude Desktop:

1. Install mcp - proxy:

   pip install mcp-proxy
   

2. Run the following command in the terminal to create a configuration file:

   python -m mcp_proxy.start_server
   

3. Modify the ~/.config/mcp_proxy/config.json file and add your model endpoint information.
4. Use the Claude application to connect to the MCP server.