airtable

This MCP server exposes a set of tools that allow a caller, such as an LLM, to inspect and mutate Airtable bases via the Model Context Protocol. It provides read and write access to Airtable schemas and data, enabling you to explore base structures, list bases and tables, and perform CRUD operations on records and tables. The available tools include list_bases, list_tables, describe_table, get_record, list_records, search_records, create_record, update_records, delete_records, create_table, update_table, and create_field. To use the server, provision an Airtable API key (PAT) with appropriate scopes and provide it to the server via the AIRTABLE_API_KEY environment variable when launching the MCP server through npx. The tools are designed to be called with standard MCP input parameters (baseId, tableId, recordIds, fields, etc.) and will return the corresponding Airtable responses. This setup is suitable for enabling LLM-driven database exploration, dynamic schema understanding, and controlled data manipulation within Airtable bases.

Prerequisites:

• Node.js and npm installed on your machine
• An Airtable account and a Personal Access Token (PAT) with appropriate scopes

Step-by-step installation:

1. Ensure Node.js is installed. Verify with:

node -v
npm -v

2. Obtain an Airtable personal access token (PAT) with at least schema.bases:read and data.records:read scopes, and optionally write scopes if you need write access.
3. Install and run the MCP server via npx, supplying your Airtable API key as an environment variable:

AIRTABLE_API_KEY=pat<your-token-here> npx -y airtable-mcp-server

4. Alternatively, run a hosted/dev version by configuring your environment as shown in the documentation for Claude Desktop, Cursor, or Cline (see README for exact JSON examples).
5. If you need to persist configuration, you can embed the same command and env in your MCP tooling configuration (e.g., Claude Desktop, Cursor, or Cline) as documented.

Tips and considerations:

• Keep your AIRTABLE_API_KEY secure and rotate it if needed.
• The server respects Airtable API rate limits; design calls to batch or stagger requests where possible.
• For write operations, ensure the PAT has the necessary write scopes (schema.bases:write, data.records:write) if you intend to modify data or schema.
• When using list_tables or describe_table, you can specify detail levels to control the amount of metadata returned (e.g., identifiersOnly, full).
• If you encounter authentication errors, double-check that AIRTABLE_API_KEY is correctly set in the environment and that the token remains valid.
• The provided tools cover a broad set of Airtable actions; implement proper error handling in your client to interpret Airtable API responses and MCP errors.