Meilisearch

You can run a Meilisearch MCP Server to control Meilisearch through a standardized programmatic interface. This server exposes tools for managing indexes, documents, searches, settings, tasks, and system status, enabling AI assistants to interact with Meilisearch in a structured way.

How to use

Connect your MCP client to the Meilisearch MCP Server using one of the available configurations. You will run the MCP server locally or remotely and then issue operations such as creating indexes, adding documents, performing searches, and adjusting index settings. Start by choosing a local stdio setup (running the MCP server as a Node process) or a manual Claude Desktop or Cursor integration as described in the configuration sections.

How to install

# Prerequisites
- Node.js and npm installed
- Access to a Meilisearch instance (host and API key)

# Manual installation flow
git clone https://github.com/devlimelabs/meilisearch-ts-mcp.git
cd meilisearch-ts-mcp
npm install

# Create .env from example and edit as needed
cp .env.example .env
# Edit .env to configure your Meilisearch connection

# Build and start in development or production as needed
npm run build
npm start

Additional configuration and usage notes

Configure client integrations and environment to connect to your Meilisearch MCP Server. You can use a local stdio configuration to run the MCP server with Node and point it at your Meilisearch instance. The most common runtime is Node executing the built MCP entrypoint, with environment variables providing the Meilisearch host and API key.

Claude Desktop integration

Integrate the MCP Server with Claude Desktop to manage Meilisearch directly from Claude. Use the setup flow that builds the project and configures Claude, or manually add an MCP server entry in Claude’s configuration to point at the local MCP process.

Cursor integration

Connect Cursor to the local MCP server to issue Meilisearch commands from the code editor. Start the MCP server from the project, then add a connection in Cursor with the command and environment details shown in the configuration examples.

Configuration examples

{
  "mcpServers": {
    "meilisearch": {
      "command": "node",
      "args": ["/path/to/meilisearch-ts-mcp/dist/index.js"],
      "env": {
        "MEILISEARCH_HOST": "http://localhost:7700",
        "MEILISEARCH_API_KEY": "your-api-key"
      }
    }
  }
}

Security and best practices

Keep your Meilisearch API key secure. Use environment-based configuration for all connections and restrict MCP server access to trusted clients. Regularly rotate API keys and monitor MCP server logs for anomalous activity.

Troubleshooting tips

If the MCP server fails to start, verify Node.js version compatibility, ensure dependencies are installed, and confirm that the Meilisearch host and API key are reachable from the MCP server environment. Check logs for any connection errors to Meilisearch and adjust the host URL or API key accordingly.

Available tools

create-index

Create a new Meilisearch index with given settings and mappings.

get-index

Retrieve information about a specific index.

list-indexes

List all indexes in the Meilisearch instance.

update-index

Update settings or characteristics of an existing index.

delete-index

Delete an index from the Meilisearch instance.

add-documents

Add documents to a specified index.

get-document

Fetch a document by its ID from an index.

get-documents

Retrieve multiple documents by IDs from an index.

update-documents

Update one or more documents in an index.

delete-document

Delete a single document by ID from an index.

delete-documents

Delete multiple documents from an index.

delete-all-documents

Remove all documents from a specific index.

search

Execute a search against an index with various parameters.

multi-search

Run multiple search queries in a single request.

get-settings

Fetch current index settings.

update-settings

Update index settings.

reset-settings

Reset index settings to default values.

enable-vector-search

Enable vector search features for an index.

get-experimental-features

Query the status of experimental features.

update-embedders

Configure embedding models for vector search.

get-embedders

Retrieve current embedder configuration.

reset-embedders

Reset embedders to defaults.

vector-search

Perform a vector-based search against indexed items.