chroma-core/chroma-mcp

Overview

A Python-based MCP server that exposes Chroma vector database operations through the Model Context Protocol. It allows AI agents to store, query, and manage embeddings via a standardized interface.

Best for

Best for
Developers building MCP-based AI agents that need persistent vector storage and retrieval

Use cases

• Connect AI agents to a Chroma vector store for semantic search
• Persist and retrieve embeddings from agentic workflows
• Manage collections and documents through MCP tool calls

How to use

Tools exposed

• chroma_list_collections
• chroma_create_collection
• chroma_peek_collection
• chroma_get_collection_info
• chroma_get_collection_count
• chroma_modify_collection
• chroma_delete_collection
• chroma_add_documents
• chroma_query_documents
• chroma_get_documents
• chroma_update_documents
• chroma_delete_documents

Tested with

Claude Desktop

Example client config

{\n    "chroma": {\n        "command": "uvx",\n        "args": [\n            "chroma-mcp"\n        ]\n    }\n}

Notes

A Python-based MCP server that exposes Chroma vector database operations through the Model Context Protocol. It allows AI agents to store, query, and manage embeddings via a standardized interface.

554 stars on GitHub. Last updated 2025-09-17. Licensed Apache-2.0.

Use cases

• Connect AI agents to a Chroma vector store for semantic search
• Persist and retrieve embeddings from agentic workflows
• Manage collections and documents through MCP tool calls

Pros

• Leverages the popular Chroma vector database for production-grade storage
• Simple MCP integration for any MCP-compatible AI agent
• Open source with active community (554 stars)

Cons

• Requires a running Chroma instance or configuration
• Limited to Chroma’s feature set (no multi-vector or hybrid search)
• Python-only, may not fit non-Python agent stacks

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.