Chroma - MeshAgent

Use the following steps to quickly deploy the Chroma MCP server and toolkit inside a Meshagent Room for cloud-based testing. When the room closes, all services and chatbots are automatically cleaned up. Environment Variables Needed:

CHROMA_API_KEY: your-api-key (replace with your own key)

1. Install the Meshagent CLI

pip install "meshagent[all]"

Follow the Meshagent CLI Getting Started Guide to sign up and log in.

3. Start the Chroma MCP Server as a Meshagent Service in a Room

meshagent service test \
  --room=test \
  --role=agent \
  --image=meshagent/mcp-chroma:latest \
  --env MESHAGENT_PORT=8001 \
  --env CHROMA_API_KEY=your-api-key \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-chroma-service-test

This command will start a temporary Meshagent Room (test) with the MCP server agent available. The room (and the agent) are auto-removed when idle.

4. Join the Room with a Chatbot Linked to the Chroma Toolkit

meshagent chatbot join \
  --room=test \
  --agent-name=mcp-chroma \
  --name=mcp-chroma \
  --toolkit=mcp-chroma

This chatbot can invoke the Chroma MCP tools in the same room.
Supports multiple toolkits/agents in the same room.
The command output provides a link to the live room.

5. Interact in the Room

Open the provided room link in your browser. Send a message to the agent to interact with the MCP tools live.

Project Level Deployment: Persistent Project-Wide Services

For production use, you may want MCP tools and a chatbot always available whenever a Meshagent project room is created. This approach creates persistent services that auto-join new project rooms, removing the need to start agents manually or run anything locally.

1. Create the Persistent Chroma MCP Agent Service

meshagent service create \
  --role=agent \
  --image=meshagent/mcp-chroma:latest \
  --env MESHAGENT_PORT=8001 \
  --env CHROMA_API_KEY=your-api-key \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-chroma-service

2. Create the Persistent Chatbot Service

meshagent service create \
  --image="meshagent/cli:latest" \
  --port="num=9001 path=/agent liveness=/ type=meshagent.callable participant_name=mcp-chroma-chatbot" \
  --env="MESHAGENT_PORT=9001" \
  --name="mcp-chroma-chatbot-service" \
  --command="meshagent chatbot service --agent-name=mcp-chroma-chatbot --toolkit=mcp-chroma"

With these, both the MCP API tools and the chatbot automatically join every new project room with no further setup.

Tools Available

Tools Summary

Tool Name	Description
`chroma_add_documents`	Add documents to a Chroma collection.
`chroma_create_collection`	Create a new Chroma collection with configurable HNSW parameters.
`chroma_delete_collection`	Delete a Chroma collection.
`chroma_delete_documents`	Delete documents from a Chroma collection.
`chroma_get_collection_count`	Get the number of documents in a Chroma collection.
`chroma_get_collection_info`	Get information about a Chroma collection.
`chroma_get_documents`	Get documents from a Chroma collection with optional filtering.
`chroma_list_collections`	List all collection names in the Chroma database with pagination support.
`chroma_modify_collection`	Modify a Chroma collection’s name or metadata.
`chroma_peek_collection`	Peek at documents in a Chroma collection.
`chroma_query_documents`	Query documents from a Chroma collection with advanced filtering.
`chroma_update_documents`	Update documents in a Chroma collection.

Tools Details

`chroma_add_documents`

Add documents to a Chroma collection.

Parameter	Type	Description
collection_name	string	Name of the collection to add documents to
documents	array	List of text documents to add
ids	array	List of IDs for the documents (required)
metadatas	string (optional)	Optional list of metadata dictionaries for each document

`chroma_create_collection`

Create a new Chroma collection with configurable HNSW parameters.

Parameter	Type	Description
collection_name	string	Name of the collection to create
embedding_function_name	string (optional)	Name of the embedding function to use (options: ‘default’, ‘cohere’, ‘openai’, etc.)
metadata	string (optional)	Optional metadata dict to add to the collection

`chroma_delete_collection`

Delete a Chroma collection.

Parameter	Type	Description
collection_name	string	Name of the collection to delete

`chroma_delete_documents`

Delete documents from a Chroma collection.

Parameter	Type	Description
collection_name	string	Name of the collection to delete documents from
ids	array	List of document IDs to delete

`chroma_get_collection_count`

Get the number of documents in a Chroma collection.

Parameter	Type	Description
collection_name	string	Name of the collection to count

`chroma_get_collection_info`

Get information about a Chroma collection.

Parameter	Type	Description
collection_name	string	Name of the collection to get info about

`chroma_get_documents`

Get documents from a Chroma collection with optional filtering.

Parameter	Type	Description
collection_name	string	Name of the collection to get documents from
ids	string (optional)	Optional list of document IDs to retrieve
include	array (optional)	List of what to include in response
limit	string (optional)	Optional maximum number of documents to return
offset	string (optional)	Optional number of documents to skip
where	string (optional)	Optional metadata filters
where_document	string (optional)	Optional document content filters

`chroma_list_collections`

List all collection names in the Chroma database with pagination support.

Parameter	Type	Description
limit	string (optional)	Maximum number of collections to return
offset	string (optional)	Number of collections to skip

`chroma_modify_collection`

Modify a Chroma collection’s name or metadata.

Parameter	Type	Description
collection_name	string	Name of the collection to modify
new_metadata	string (optional)	New metadata for the collection
new_name	string (optional)	New name for the collection

`chroma_peek_collection`

Peek at documents in a Chroma collection.

Parameter	Type	Description
collection_name	string	Name of the collection to peek
limit	integer (optional)	Number of documents to peek at

`chroma_query_documents`

Query documents from a Chroma collection with advanced filtering.

Parameter	Type	Description
collection_name	string	Name of the collection to query
query_texts	array	List of query texts to search for
include	array (optional)	What to include in the response
n_results	integer (optional)	Number of results per query
where	string (optional)	Metadata filters
where_document	string (optional)	Document content filters

`chroma_update_documents`

Update documents in a Chroma collection.

Parameter	Type	Description
collection_name	string	Name of the collection to update
ids	array	List of document IDs to update (required)
documents	string (optional)	List of new text documents
embeddings	string (optional)	List of new embeddings
metadatas	string (optional)	List of new metadata dictionaries

Meshagent Resources

Meshagent Website: https://www.meshagent.com/
Read The Docs: https://docs.meshagent.com
MCP Tool Image on Docker Hub: https://hub.docker.com/r/meshagent/mcp-chroma

Overview

Tools

​1. Install the Meshagent CLI

​2. Sign Up & Authenticate

​3. Start the Chroma MCP Server as a Meshagent Service in a Room

​4. Join the Room with a Chatbot Linked to the Chroma Toolkit

​5. Interact in the Room

​Project Level Deployment: Persistent Project-Wide Services

​1. Create the Persistent Chroma MCP Agent Service

​2. Create the Persistent Chatbot Service

​Tools Available

​Tools Summary

​Tools Details

​chroma_add_documents

​chroma_create_collection

​chroma_delete_collection

​chroma_delete_documents

​chroma_get_collection_count

​chroma_get_collection_info

​chroma_get_documents

​chroma_list_collections

​chroma_modify_collection

​chroma_peek_collection

​chroma_query_documents

​chroma_update_documents

​Meshagent Resources