Chroma MCP Server

What is this?

Chroma MCP Server is a specialized implementation of the Managed Connector Protocol (MCP) designed to simplify semantic search and context retrieval over vector databases. It uses Chroma, a leading open-source vector database, to index and query large collections of embeddings with sub-millisecond latency, making it perfect for AI applications like Retrieval-Augmented Generation (RAG) and AI agents.

By adhering to the MCP 1.x specification with HTTP and gRPC endpoints, token-based authentication, and role-based permissions, Chroma MCP Server lets you seamlessly plug vector search and filtering into any MCP-compatible client or orchestration framework. With features like advanced k-NN indexing, metadata filters, batch operations, and extensibility through plugin hooks, it provides a unified, secure, and scalable solution for modern AI workflows.

Quick Start

Install the server using npm:

npm install @modelcontextprotocol/server-chroma-mcp-server

Then add it to your MCP client configuration:

{
  "mcpServers": {
    "chroma-mcp-server": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-chroma-mcp-server"],
      "env": {
        "API_KEY": "your-api-key-here"
      }
    }
  }
}

Key Features

Feature 1: Standardized Connector Interface—compliance with MCP lets you swap or combine backends without changing client code.

Feature 2: High-Performance Similarity Search—HNSW and IVF indexing for sub-millisecond k-NN queries at scale.

Feature 3: Seamless LLM Integration—out-of-the-box support for popular frameworks like LangChain, LlamaIndex, and Haystack.

Example Usage

Here’s a quick example of performing a semantic search using the MCP client to retrieve top hits from a collection.

// Example code
const result = await client.callTool({
  name: "chroma-mcp-search",
  arguments: {
    collectionId: "col_123abc",
    values: [0.130, 0.450, /* ... */],
    topK: 3,
    filter: { category: { eq: "nda" } }
  }
});

This code sends a query embedding to your collection and returns the top three matching documents with metadata.

Configuration

The server accepts the following environment variables:

API_KEY – your MCP token for authenticating requests to the server.

SETTING_1 (optional) – LOG_LEVEL to control the verbosity of server logs.

Available Tools/Resources

chroma-mcp-cli: Command-line interface for manual operations and health checks.

TypeScript SDK: Auto-generated client library for TypeScript applications.

Who Should Use This?

This server is perfect for:

Use case 1: AI developers building RAG pipelines needing fast, scalable context retrieval.

Use case 2: Teams integrating semantic search into web or chatbot applications.

Use case 3: Enterprises requiring secure, multi-tenant vector storage with role-based access control.

Conclusion

Chroma MCP Server brings high-performance vector search, standardization, and enterprise-grade security to your AI workflows. Try it out today to simplify context injection and build scalable RAG and semantic search applications.

Check out the GitHub repository for more information and to contribute.