Both. Set CHROMA_PATH for a local persistent file (the simplest mode), CHROMA_HOST for a self-hosted server, or pair CHROMA_API_KEY with a Cloud tenant URL for hosted Chroma.

Which embedding function does it use?

Chroma's default embedding function for the collection — usually all-MiniLM-L6-v2 on local or a hosted model in Cloud. Switch per-collection if you need a specific dimensionality.

Does it support metadata filters?

Yes. The query tool accepts a where filter (string equality, IN, NOT IN, comparison) and a where_document filter for inside-text search.

Chroma

Name: Top MCPs — curated MCP directory
Creator: Top MCPs
License: https://creativecommons.org/licenses/by/4.0/

Official

Embedded and hosted vector database for AI agents — open source, zero-ops.

Score 72(?)ChromaApache-2.0 554Verified Jun 2, 2026Top MCPs for AI & Machine Learning →

Install View on GitHub

Quick answer

What it does

Connects to Chroma (in-process, persistent, or Cloud) and exposes collection management, document add/update, and semantic + metadata-filtered queries to AI models.

Best for

Persistent agent memory
Personal knowledge base retrieval
Document-grounded RAG
Local prototyping of a vector pipeline

Not for

Billion-scale workloads
Production multi-tenant SaaS without operator attention

Setup recipe

Pick your client, then follow the three steps.

Install

claude_desktop_config.jsonjson

{
  "mcpServers": {
    "chroma": {
      "command": "uvx",
      "args": [
        "chroma-mcp"
      ],
      "env": {
        "CHROMA_PATH": "${CHROMA_PATH}",
        "CHROMA_HOST": "${CHROMA_HOST}",
        "CHROMA_API_KEY": "${CHROMA_API_KEY}"
      }
    }
  }
}

Paste under mcpServers. Fully quit and reopen Claude after editing.

CLI or .mcp.jsonshell

# export CHROMA_PATH=./chroma_db
# export CHROMA_HOST=optional-for-server-mode
# export CHROMA_API_KEY=optional-for-cloud
claude mcp add chroma -- uvx chroma-mcp

Run from your repo. Commit .mcp.json to share with your team.

.cursor/mcp.jsonjson

{
  "mcpServers": {
    "chroma": {
      "command": "uvx",
      "args": [
        "chroma-mcp"
      ],
      "env": {
        "CHROMA_PATH": "${CHROMA_PATH}",
        "CHROMA_HOST": "${CHROMA_HOST}",
        "CHROMA_API_KEY": "${CHROMA_API_KEY}"
      }
    }
  }
}

Global path: ~/.cursor/mcp.json. Reload window after editing.

.vscode/mcp.jsonjsonc

{
  "servers": {
    "chroma": {
      "command": "uvx",
      "args": [
        "chroma-mcp"
      ],
      "env": {
        "CHROMA_PATH": "${CHROMA_PATH}",
        "CHROMA_HOST": "${CHROMA_HOST}",
        "CHROMA_API_KEY": "${CHROMA_API_KEY}"
      }
    }
  }
}

VS Code uses the "servers" key (not "mcpServers").

~/.codeium/windsurf/mcp_config.jsonjson

{
  "mcpServers": {
    "chroma": {
      "command": "uvx",
      "args": [
        "chroma-mcp"
      ],
      "env": {
        "CHROMA_PATH": "${CHROMA_PATH}",
        "CHROMA_HOST": "${CHROMA_HOST}",
        "CHROMA_API_KEY": "${CHROMA_API_KEY}"
      }
    }
  }
}

Open via Cascade → hammer icon → Configure.

cline_mcp_settings.jsonjson

{
  "mcpServers": {
    "chroma": {
      "command": "uvx",
      "args": [
        "chroma-mcp"
      ],
      "env": {
        "CHROMA_PATH": "${CHROMA_PATH}",
        "CHROMA_HOST": "${CHROMA_HOST}",
        "CHROMA_API_KEY": "${CHROMA_API_KEY}"
      }
    }
  }
}

Open via the Cline sidebar → MCP Servers → Edit.

~/.continue/config.jsonjson

{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "stdio",
          "command": "uvx",
          "args": [
            "chroma-mcp"
          ],
          "env": {
            "CHROMA_PATH": "${CHROMA_PATH}",
            "CHROMA_HOST": "${CHROMA_HOST}",
            "CHROMA_API_KEY": "${CHROMA_API_KEY}"
          }
        }
      }
    ]
  }
}

Continue uses modelContextProtocolServers with a transport block.

~/.codex/config.tomlshell

# ~/.codex/config.toml
[mcp_servers.chroma]
command = "uvx"
args = [
  "chroma-mcp",
]
env = { CHROMA_PATH = "${CHROMA_PATH}", CHROMA_HOST = "${CHROMA_HOST}", CHROMA_API_KEY = "${CHROMA_API_KEY}" }

Codex uses TOML. Each server is a [mcp_servers.<name>] subtable.

~/.config/zed/settings.jsonjsonc

{
  "context_servers": {
    "chroma": {
      "command": {
        "path": "uvx",
        "args": [
          "chroma-mcp"
        ]
      },
      "env": {
        "CHROMA_PATH": "${CHROMA_PATH}",
        "CHROMA_HOST": "${CHROMA_HOST}",
        "CHROMA_API_KEY": "${CHROMA_API_KEY}"
      }
    }
  }
}

Zed calls them "context_servers". Settings live-reload on save.

ChatGPT → Apps directorynone

Chroma doesn't ship a hosted HTTPS endpoint today. ChatGPT supports remote MCP servers only — to use this server in ChatGPT you'll need to deploy it to a public HTTPS URL first (e.g. via Cloudflare Workers or Vercel) or wait for an official remote build.

2
Set required secrets
Set CHROMA_PATH, CHROMA_HOST, CHROMA_API_KEY in your shell environment before launching your MCP client.
3
Try a minimum working prompt
Minimum working prompt pending verification. Try any prompt from the MCP’s README once installed.

Tools & permissions

Tool	Description	Args	Side effects
`list_collections`	List collections in the Chroma instance.	—	Read
`create_collection`	Create a new collection.	name: string	Write
`add_documents`	Add documents to a collection.	collection: string, documents: array, metadatas?: array	Write
`query`	Run a semantic query with optional metadata filters.	collection: string, query_text: string, n_results?: number	Read

Security & scope

Access scope: Read + write
Sandbox: Local mode runs entirely in the MCP process — data lives at CHROMA_PATH. Cloud mode authenticates with an API key scoped to one tenant. No outbound network calls in local mode.
Gotchas: Local mode writes to disk on every add — back up the CHROMA_PATH directory if it matters.
Cloud API keys are full-tenant — scope by tenant rather than relying on collection-level isolation.
The default embedding function downloads model weights on first run — pre-warm in CI to avoid first-call latency.

Agent prompt pack

— copy into Claude, Cursor, or ChatGPT.

Paste into Claude, Cursor, or ChatGPT. Edit the [brackets] before sending.

Recommend the best MCP servers for [task: e.g. ai & machine learning work] in [client: Claude].

Constraints:
- Prefer tools that are [official | open-source | read-only] — pick what matters for my use case.
- Exclude MCPs that require [e.g. a paid plan, OAuth-only flows, remote-only transport].
- Return at most 3 picks, ranked.

For each pick include:
1. One-sentence rationale.
2. The ready-to-paste install snippet for my client.
3. Any required secrets I need to create before installing.

Cross-check the top-mcps.com listing: https://top-mcps.com/top-mcps-for-ai-machine-learning

Compare Chroma against a real alternative. Swap the second MCP in [brackets] if you want a different match.

Compare Chroma MCP vs [Pinecone MCP] for the following job: [describe the job, e.g. "let an agent create GitHub issues on bug triage"].

Judge them on:
- Setup time and complexity (what a new user hits first).
- Auth model (none / API key / OAuth 2.1) and credential risk.
- Transport (stdio / Streamable HTTP / SSE) and where the server runs.
- Required secrets and the blast radius if they leak.
- Operational risk in an unattended agent loop.
- Which one is "good enough" for a weekend prototype vs. production.

End with one sentence: which should I pick for my scenario, which is: [my scenario].

References:
- https://top-mcps.com/mcp/chroma
- top-mcps.com listing for Pinecone

Asks the agent to install and verify. Works inside Claude Code, Cursor Agent, Codex CLI.

Install the Chroma MCP server for my [client: Claude] at the default config path for that client.

Use the exact install snippet published at https://top-mcps.com/mcp/chroma (fetch https://top-mcps.com/mcp/chroma.json for the canonical server.json if you can read URLs).

Before finishing:
1. Create the required secrets (CHROMA_PATH, CHROMA_HOST, CHROMA_API_KEY) and put them in the appropriate env block — do not hard-code them.
2. Restart or reload the client so it picks up the new server.
3. Verify the server is connected (green / running state) and at least one tool is listed.
4. If anything fails, read the client's MCP logs and report the exact error — do not silently retry.

Confirm when done and list the tools the server now exposes.

Frequently asked questions

What changed

— 2 updates tracked.

Jun 2, 2026
Refreshed install snippets and fact sheet; verified for 2026.
Feb 8, 2025
Initial directory listing.

More AI & Machine Learning MCPs

Other tools in the same category worth evaluating.

Hugging Face

Search, inspect, and run Hugging Face models and datasets from an agent.

huggingfacemlmodelsinference

3 minLow

Replicate

Run any open-source model on Replicate from inside an AI agent.

aimlreplicateinference

3 minLow

OpenRouter

Route one prompt across 300+ LLMs with a single API key.

aillmopenrouterinference

2 minLow

Compared with Chroma

Side-by-side breakdowns for the choices people most often weigh against this MCP.

ChromavsPinecone ChromavsQdrant ChromavsMem0

Exploring Top MCPs for AI & Machine Learning? See all AI & Machine Learning MCPs →