Database capabilities for AI models via the Model Context Protocol
Visit WebsiteFreeVisit Website
TL;DR - Chroma MCP
- Open-source MCP server that gives AI models database capabilities for storing and retrieving data using vector search, full text search, and metadata filtering.
- Supports multiple client types including ephemeral, persistent, HTTP, and cloud, with flexible embedding functions from providers like OpenAI and Cohere.
- Integrates with MCP-compatible clients like Claude Desktop, enabling LLMs to maintain memory and context across sessions.
Pricing: Free forever
Best for: Individuals & startups
What is Chroma MCP?
Chroma MCP is an open-source server implementation of the Model Context Protocol (MCP) that provides database capabilities powered by Chroma, the open-source embedding database. It enables AI models, particularly large language models (LLMs), to create collections over generated data and user inputs, and retrieve that data using vector search, full text search, metadata filtering, and more. The server is designed for self-hosting, giving developers full control over their data and infrastructure.
The server supports multiple client types: ephemeral (in-memory) for testing and development, persistent for file-based storage, HTTP client for self-hosted Chroma instances, and a cloud client for Chroma Cloud integration. It offers comprehensive collection management, document operations, and a suite of tools for querying and filtering data. Embedding functions from providers like Cohere, OpenAI, Jina, VoyageAI, and Roboflow are supported, with the selected embedding function persisting per collection for consistent retrieval.
Chroma MCP is ideal for developers building LLM applications that require memory, context, and retrieval-augmented generation (RAG). It integrates seamlessly with MCP-compatible clients like Claude Desktop, allowing AI assistants to store and retrieve information dynamically. The project is actively maintained by Chroma, the company behind the popular open-source embedding database.
Pros & Cons
Pros
- Open-source and free to use with no licensing costs
- Flexible deployment options from in-memory testing to cloud production
- Seamless integration with MCP ecosystem for AI model context management
Cons
- Requires self-hosting and configuration, which may be complex for non-technical users
- Limited to Chroma's ecosystem; not compatible with other vector databases out of the box
Preview
Key Features
Flexible client types: ephemeral (in-memory), persistent (file-based), HTTP (self-hosted Chroma), and cloud (Chroma Cloud)Collection management: create, modify, delete collections with pagination and HNSW parameter configurationDocument operations: add, update, delete, and query documents with metadata and custom IDsAdvanced search: semantic vector search, full text search, and metadata filteringEmbedding function persistence: selected embedding function is stored per collection for consistent retrievalSupported embedding functions: default, Cohere, OpenAI, Jina, VoyageAI, and RoboflowIntegration with MCP-compatible clients like Claude Desktop via simple configurationSelf-hosted for full data control and privacy
Pricing Plans
Pricing checked Jul 31, 2026
Open Source
Free
- Full source code access
- Community support
- Self-hosted
Reviews
Be the first to review Chroma MCP
Your take helps the next buyer. Verified LinkedIn reviewers get a badge.
Write a reviewBest Chroma MCP Alternatives
Top alternatives based on features, pricing, and user needs.
BrandfolderPaid
Manage, organize, and distribute all your brand assets
BynderPaid
Organize, manage, and distribute creative content globally
SmartlingPaid
End-to-end localization & translation services for global growth powered by AI.
MonkeyLearnFreemium
Uncover high-impact insights from unstructured text data with AI-powered omnichannel text analytics.
Intelligence NodePaid
AI-powered competitive intelligence for enterprise retail to optimize digital shelf performance and drive revenue.
EvisortPaid
AI-powered contract management software for end-to-end contract lifecycle management.
Explore More
Chroma MCP FAQ
How does Chroma MCP persist embedding functions across sessions?
Chroma MCP persists the selected embedding function per collection using Chroma's collection configuration feature, introduced in version 1.0.0. Once a collection is created with a specific embedding function, future queries and inserts automatically use the same function without needing to specify it again. Collections created with Chroma versions 0.6.3 or earlier do not support this persistence.
Can I use Chroma MCP with Claude Desktop?
Yes, Chroma MCP integrates with Claude Desktop by adding a configuration entry to the claude_desktop_config.json file. You can configure it for ephemeral, persistent, or cloud clients by specifying the appropriate command and arguments, such as 'uvx chroma-mcp' for ephemeral or adding '--client-type persistent' for file-based storage.
What embedding functions are supported by Chroma MCP?
Chroma MCP supports several embedding functions: default, Cohere, OpenAI, Jina, VoyageAI, and Roboflow. Each function requires the corresponding API key set as an environment variable in the correct format. The embedding function is stored per collection and reused automatically for all subsequent operations.
How do I connect Chroma MCP to Chroma Cloud?
To connect to Chroma Cloud, configure the client with '--client-type cloud' and provide your tenant ID, database name, and API key. The server automatically connects to api.trychroma.com using SSL. For security, you can specify a custom path for your environment configuration file using the '--dotenv-path' argument instead of passing API keys directly in the command.
What types of search does Chroma MCP support?
Chroma MCP supports semantic vector search, full text search, and metadata filtering. You can query documents using semantic similarity, filter by metadata fields and document content, and combine these methods for advanced retrieval. The server also provides tools for retrieving documents by IDs or custom filters with pagination.
Can I use Chroma MCP for production workloads?
Yes, Chroma MCP is designed for production use with support for persistent file-based storage and HTTP client connections to self-hosted Chroma instances. For cloud-scale production, you can use the Chroma Cloud client, which automatically connects to api.trychroma.com with SSL. The server also supports HNSW parameter configuration for optimized vector search performance.
How does Chroma MCP handle document updates and deletions?
Chroma MCP provides dedicated tools for updating and deleting documents. The 'chroma_update_documents' tool allows you to modify existing documents' content, metadata, or embeddings. The 'chroma_delete_documents' tool removes specific documents from a collection. Both operations work with custom document IDs for precise targeting.
What is the difference between ephemeral and persistent client types in Chroma MCP?
The ephemeral client stores data in-memory and is ideal for testing and development, as data is lost when the server stops. The persistent client stores data on disk using a specified data directory, allowing data to survive server restarts and be reused across sessions. The persistent client is configured by adding '--client-type persistent' and '--data-dir' arguments.
Source: github.com