Skip to main content
mcpSource-backedReview first Safety Privacy

Chroma MCP Server

Official Chroma MCP server for connecting Claude to Chroma collections, documents, semantic search, full-text search, metadata filtering, persistent storage, self-hosted Chroma, and Chroma Cloud.

by Chroma·added 2026-06-06·
Claude CodeClaude Desktop
HarnessClaude CodeClaude Desktop
Review first review before installing

Open the source and read safety notes before installing.

Safety notes

  • Chroma MCP can create, modify, and delete collections.
  • Document tools can add, update, query, retrieve, and delete documents, metadata, custom IDs, and embeddings.
  • Persistent, HTTP, and cloud modes can mutate durable retrieval stores rather than temporary test collections.
  • External embedding functions can send document text or image-derived content to third-party embedding providers.
  • Chroma Cloud and self-hosted HTTP modes require careful handling of tenants, databases, API keys, custom auth credentials, SSL settings, and network exposure.
  • Retrieved context can influence Claude output even when stale, irrelevant, over-broad, or not authorized for the task.

Privacy notes

  • Collections can store source documents, chunks, embeddings, metadata, IDs, filters, and query results that reveal private project, customer, or research data.
  • Query text, retrieved documents, metadata filters, and embedding inputs may expose sensitive information to Chroma Cloud, self-hosted operators, embedding providers, logs, or downstream model providers.
  • CHROMA_API_KEY, custom auth credentials, embedding provider API keys, tenant IDs, database names, hostnames, and dotenv files should stay out of prompts, issues, logs, screenshots, and committed files.
  • Persistent data directories and exported collections need the same access control, backup, encryption, and retention review as the source documents they index.

Prerequisites

  • Python 3.10 or newer available through uv for local stdio usage.
  • [object Object]
  • Storage, backup, and retention plan for persistent collections before enabling write tools.
  • Chroma Cloud tenant, database, and API key or self-hosted Chroma host credentials if using remote modes.
  • Embedding provider API keys if using Cohere, OpenAI, Jina, Voyage AI, or Roboflow embedding functions.

Schema details

Install type
cli
Troubleshooting
No
Source repository stats
Scope
Source repo
Collection metadata
Estimated setup
15 minutes
Difficulty
advanced
Tool listing metadata
Full copyable content
{
  "mcpServers": {
    "chroma": {
      "command": "uvx",
      "args": ["chroma-mcp"]
    }
  }
}

About this resource

Content

Chroma MCP Server is the official Chroma Model Context Protocol server for self-hosting Claude access to Chroma data retrieval workflows. It connects MCP clients to Chroma collections and documents through ephemeral, persistent, HTTP, or Chroma Cloud client modes.

The server supports collection management, document ingestion, semantic search, full-text search, metadata filtering, document retrieval, document updates, and document deletion. It can also use several embedding functions, including default, Cohere, OpenAI, Jina, Voyage AI, and Roboflow.

Source Review

These sources were reviewed on 2026-06-06. Prefer the Chroma docs, repository README, PyPI metadata, package metadata, server implementation, and license for current setup commands, supported tools, client modes, embedding providers, and authentication behavior.

Features

  • Official MCP server from chroma-core/chroma-mcp.
  • Local stdio launch with uvx chroma-mcp.
  • Ephemeral in-memory mode for testing and development.
  • Persistent mode for file-backed Chroma storage.
  • HTTP mode for self-hosted Chroma instances.
  • Cloud mode for Chroma Cloud via api.trychroma.com.
  • Collection listing, creation, metadata updates, count retrieval, peeking, and deletion.
  • Document add, query, get, update, and delete tools.
  • Semantic search, full-text search, metadata filtering, and pagination.
  • Optional HNSW configuration when creating collections.
  • Embedding functions for default, Cohere, OpenAI, Jina, Voyage AI, and Roboflow providers.
  • Environment variable and dotenv-file configuration for client modes, credentials, and embedding keys.

Installation

For an ephemeral local client:

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

For durable or remote use, follow the README's persistent, HTTP, and cloud configuration examples. Prefer environment variables or a reviewed dotenv file for credentials instead of putting API keys directly into shared MCP config.

Use Cases

  • Let Claude search a Chroma collection before answering a project question.
  • Create temporary collections for experiments or evaluation runs.
  • Add curated documents to a persistent Chroma retrieval store.
  • Query documents with metadata and document-content filters.
  • Inspect collection counts and sample documents while debugging ingestion.
  • Use Chroma Cloud or a self-hosted Chroma instance as a shared retrieval backend.
  • Compare retrieval results before feeding context into a coding, support, or research workflow.
  • Delete stale collections only after verifying the target database and collection name.

Safety and Privacy

Chroma MCP is a retrieval database control surface, not just a read-only search helper. It can add, update, and delete indexed content, and remote modes can touch shared production or team databases. Require explicit approval for collection creation, document writes, document updates, document deletion, and collection deletion.

Treat retrieved context as untrusted input until it is checked for relevance, freshness, permissions, and prompt-injection risk. Keep API keys and custom auth credentials out of prompts and version control, and review whether embedding providers or Chroma Cloud are approved for the documents being indexed.

Disclosure

Chroma includes an Apache-2.0 open-source project and hosted Chroma Cloud offerings. This listing is not sponsored, paid, or affiliate-driven, and it is scoped to the source-backed MCP server.

Duplicate Check

Existing content includes a general Chroma tools entry in content/tools, but no Chroma MCP Server entry was found in content/mcp. This submission is scoped to chroma-core/chroma-mcp, the official MCP server, and does not change the existing Chroma database listing.

#chroma#vector-database#embeddings#retrieval#rag

Source citations

Signals

Loading live community signals…

More like this, weekly

A short, calm digest of reviewed Claude resources. Unsubscribe any time.