Grist

An MCP server implementation that provides tools for interacting with the Grist API, enabling integration between Grist spreadsheets and language models.

Features

• Access Grist data directly from language models
• List organizations, workspaces, documents, tables, and columns
• Query, add, update, and delete records in Grist tables
• Filter and sort data with rich query capabilities
• Secure API key-based authentication

Tools

list_organizations

List all Grist organizations that the user has access to.

Returns: A list of organizations with their IDs, names, and domains.

list_workspaces

List all workspaces in a Grist organization.

Inputs:

• org_id (int or string): The ID of the organization to list workspaces for

Returns: A list of workspaces with their IDs and names.

list_documents

List all documents in a Grist workspace.

Inputs:

• workspace_id (int): The ID of the workspace to list documents for

Returns: A list of documents with their IDs and names.

list_tables

List all tables in a Grist document.

Inputs:

• doc_id (string): The ID of the document to list tables for

Returns: A list of tables with their IDs.

list_columns

List all columns in a Grist table.

Inputs:

• doc_id (string): The ID of the document containing the table
• table_id (string): The ID of the table to list columns for

Returns: A list of columns with their IDs and field data.

list_records

List records in a Grist table with optional filtering, sorting, and limiting.

Inputs:

• doc_id (string): The ID of the document containing the table
• table_id (string): The ID of the table to list records from
• filter_json (string, optional): JSON string for filtering records (e.g., '{"column_name": ["value1", "value2"]}')
• sort (string, optional): Comma-separated list of columns to sort by (prefix with '-' for descending order)
• limit (int, optional): Maximum number of records to return

Returns: A list of records with their IDs and field data.

add_grist_records

Add records to a Grist table.

Inputs:

• doc_id (string): The ID of the document containing the table
• table_id (string): The ID of the table to add records to
• records (array): List of records to add as dictionaries of field values

Returns: A list of IDs for the newly created records.

update_grist_records

Update records in a Grist table.

Inputs:

• doc_id (string): The ID of the document containing the table
• table_id (string): The ID of the table to update records in
• records (array): List of records to update (each must include "id" and field values)

Returns: A list of IDs for the updated records.

delete_grist_records

Delete records from a Grist table.

Inputs:

• doc_id (string): The ID of the document containing the table
• table_id (string): The ID of the table to delete records from
• record_ids (array): List of record IDs to delete

Returns: A dictionary containing the operation status and a message.

Usage

The Grist MCP Server is designed for:

• Analyzing and summarizing Grist data
• Creating, updating, and deleting records programmatically
• Building reports and visualizations based on Grist data
• Answering questions about data stored in Grist tables
• Connecting Grist with language models for natural language queries

Requirements

• Python 3.8+
• A valid Grist API key
• The following Python packages: fastmcp, httpx, pydantic, python-dotenv

Setup

Environment Variables

Create a .env file with the following:

GRIST_API_KEY=your_api_key_here
GRIST_API_HOST=https://docs.getgrist.com/api

Replace your_api_key_here with your actual Grist API key, which you can find in your Grist account settings.

Configuration

Usage with Claude Desktop

Add this to your claude_desktop_config.json:

Direct Python

{
  "mcpServers": {
    "grist-mcp": {
      "command": "python",
      "args": [
        "-m", "grist_mcp_server"
      ]
    }
  }
}

Docker

{
  "mcpServers": {
    "grist-mcp": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "-e", "GRIST_API_KEY=your_api_key_here",
        "-e", "GRIST_API_HOST=https://docs.getgrist.com/api",
        "mcp/grist-mcp-server"
      ]
    }
  }
}

Building

Docker:

docker build -t mcp/grist-mcp-server .

Installation

pip install mcp-server-grist

Or manually:

git clone https://github.com/yourusername/mcp-server-grist.git
cd mcp-server-grist
pip install -r requirements.txt

License

This MCP server is licensed under the MIT License.