M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
A

Airtable MCP Server

Community
Give AI systems read and write access to your Airtable databases
How to install ↓ Compare GitHub source repository ↗
Category
Database #52 of 58
Stars
★ 455 Very popular
Transport
stdio (local process)
Runtime
Node.js · Docker
Credentials
API key / credential required
License
MIT
Last commit
Tools
15
44FMRS · D

The server is feature-rich, covering major Airtable operations, suitable for developers and teams integrating AI assistants. However, attention must be paid to token management and security risks, especially in HTTP mode.

Strongest · Documentation 15/20 Weakest · Security and permissions 3/20

Reliability
6/20
Security and permissions
3/20
Maintenance
7/20
Documentation
15/20
Setup experience
13/20
Why each score
Reliability 6/20
The provided evidence is limited to the manifest and README; no source files, tests, or CI workflows were included. The README enumerates 16 tools with parameter schemas that appear internally consistent, and the manifest declares a standard stdio transport, so a happy path is plausible. However, declared tools cannot be cross-checked against implementation, and there is no evidence of error handling, timeouts, or edge-case tests. Deducted heavily for the absence of verifiable execution evidence.
Security and permissions 3/20
The manifest marks AIRTABLE_API_KEY as secret and the README recommends least-privilege scopes (read-only unless write is needed). This is positive. However, the server exposes delete_records and update_records without any described confirmation or dry-run step, which is an irreversible destructive default — a red-line issue. The optional HTTP transport has no built-in authentication; the README warns about DNS rebinding, but the risk remains. Scored 0-4 range per red-line rule.
Maintenance 7/20
The repository is not archived and has an MIT license; the README documents an npm-based release workflow, implying some release governance. However, no commits, release history, issue response times, or dependency update logs were supplied. The metadata gives only open issue count (2) without dates. Deducted for lack of concrete evidence of sustained maintenance.
Documentation 15/20
The README is well-structured: installation via an interactive config generator, token creation with scopes, per-tool parameter tables, HTTP-transport security warning, contributing guide, and release steps. The manifest also contains short descriptions and icons. Missing are a troubleshooting section, explicit rate/cost limits, and examples of actual tool outputs. Deducted for these gaps; otherwise comprehensive.
Setup experience 13/20
Setup is a few steps: generate config via the install-mcp link, set AIRTABLE_API_KEY. Multiple distribution channels exist (npm, Docker, mcpb) and the README gives an alternative HTTP transport command. However, the setup guide relies on an external generator and uses placeholder versions in the manifest; no real CI/test output was provided to verify the exact commands. Capped at 15 per static calibration due to lack of execution evidence.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkConnects to a database

Best for

  • Teams looking to integrate Airtable data into AI workflows
  • Individual developers who want to use natural language to operate Airtable data
  • Scenarios requiring rapid prototyping or automation

Not for

  • Teams requiring complex permission control or auditing (the server uses a single token)
  • Scenarios requiring real-time collaboration or complex data relationships
  • Users with high security requirements who cannot restrict network exposure

Required permissions

  • Requires an Airtable personal access token
  • Token needs at least schema.bases:read and data.records:read scopes
  • Optional write scopes: schema.bases:write, data.records:write, data.recordComments:read, data.recordComments:write
  • Token must have access to target bases

Risks and side effects

  • Token leakage could allow malicious reads/writes of data
  • HTTP mode has no built-in authentication and may be vulnerable to DNS rebinding attacks
  • Delete or update operations are irreversible; mistakes may cause data loss
  • Shared tokens may exceed intended access scope

Setup

Before you start

Runtime：Node.js · Docker

AIRTABLE_API_KEY requiredsecret Airtable personal access token (looks like pat123.abc123) with schema.bases:read and data.records:read scopes (plus write scopes for write tools); create at https://airtable.com/create/tokens/new.
Other optional settings (2)
MCP_TRANSPORT optional Optional; set to http to run in HTTP transport mode instead of the default stdio.
PORT optional Optional; the port for the HTTP server (3000 in the documented example).
  1. Generate an Airtable personal access token (https://airtable.com/create/tokens/new) with scopes: schema.bases:read and data.records:read (optionally write scopes).
  2. Use the install-mcp tool (https://adamjones.me/install-mcp/?config=...) to generate MCP client configuration, or manually configure.
  3. Set the environment variable AIRTABLE_API_KEY to your token.
  4. For stdio mode, run npx -y airtable-mcp-server.

Check that it works

Check the client's tool list for tools like list_bases, then ask it to run list_bases — if it returns your Airtable bases, the connection works.

Troubleshooting

  1. Confirm that the AIRTABLE_API_KEY environment variable is set correctly
  2. Check that the token's scopes include the required operations
  3. Ensure the token has access to the target base
  4. If using HTTP mode, ensure the network environment is secure
  5. Check server logs for error details

Things to try

Once connected, you can ask your AI assistant things like:

  • List all the Airtable bases I have access to
  • Search a table for records containing a specific keyword
  • Create a new record in this table with these field values
  • Show all comments on a specific record

Tools 15

list_records read-only
List records from a specified Airtable table
search_records read-only
Search for records containing specific text
list_bases read-only
List all accessible Airtable bases
list_tables read-only
List all tables in a specific base
describe_table read-only
Get detailed information about a specific table
get_record read-only
Get a specific record by ID
create_record writes
Create a new record in a table
update_records writes
Update one or more records in a table
Show 7 more tools
delete_records destructive
Delete one or more records from a table
create_table writes
Create a new table in a base
update_table writes
Update a table's name or description
create_field writes
Create a new field in a table
update_field writes
Update a field's name or description
create_comment writes
Create a comment on a record
list_comments read-only
List comments on a record

Use cases

Allow AI assistants to query and update business data in Airtable
Automate data organization and bulk record updates
Use Airtable as a knowledge base for AI to read
Create and manage table schemas through AI

Supported clients

Claude Desktop
Claude Code
Cursor
Cline
VS Code

Listed from the project's documentation, not tested by this site.

Overview

This is a Model Context Protocol (MCP) server that provides read and write access to Airtable databases. It enables LLMs to inspect database schemas, then read and write records. It supports a variety of tools including listing records, searching records, creating tables, updating fields, and adding comments. The server requires an Airtable personal access token with scopes such as schema.bases:read and data.records:read. It uses stdio transport by default, and also supports HTTP mode (with security caveats).

Similar servers

Compare with other MCP servers

Source revision b28244860d90 Data synced 2026-10-19 Read the FMRS scoring method