Airtable MCP Server – Bases and Records
Airtable's official MCP server lets AI tools query, analyze, create, and update Airtable data while respecting the user's existing permissions. Use it when an AI agent needs conversational access to bases, records, interfaces, workspace context, and approved record or base-creation workflows.
Overview
Airtable's official MCP server connects AI assistants directly to Airtable
bases. It allows users to ask questions about their data, create and update
records, analyze information, and work with interface data through natural
conversation in MCP-compatible tools.
What the MCP server enables
Airtable MCP exposes Airtable data and actions according to the authenticated
user's permissions. Depending on access level, an AI agent can:
- Search and analyze records from accessible bases.
- Create new records in tables the user can edit.
- Update existing records where the user has permission.
- Create new bases in workspaces where the user has creator-level workspace
permissions. - Access data from Airtable Interfaces, including interface pages, page records,
and record details. - Help users retrieve, summarize, and compare operational data from CRM, project
tracking, feedback, content planning, inventory, and other Airtable bases. - Use official Airtable skills such as
airtable-searchand
airtable-filtersin supported AI platforms to improve retrieval and query
quality.
When to use it
Use Airtable MCP when an AI workflow needs governed access to structured
business data stored in Airtable. Practical examples include finding customers
who have not been contacted recently, summarizing at-risk projects, adding new
tasks to a project tracker, updating a due date, creating a campaign tracker
base, or reading records surfaced through an Airtable Interface.
Connection and authentication
The official MCP server URL is https://mcp.airtable.com/mcp. OAuth is the
recommended connection method for supported clients and official connectors.
Airtable also documents personal access token configuration for manual client
setup by passing Authorization: Bearer ${AIRTABLE_PAT}.
For ChatGPT custom connectors, Airtable supports OAuth setup with dynamic client
registration or a user-defined OAuth client. Claude Code and Codex can connect
directly to the MCP URL and authorize through the browser. Enterprise users may
need an admin to allowlist the MCP integration when third-party integration
restrictions are enabled.
Key considerations
Airtable MCP mirrors the same permission level the user already has in
Airtable. Owners, creators, and editors can read and update permitted records;
workspace owners or creators can create new bases; commenters and read-only
users can read accessible data but cannot create or update records. MCP calls
use Airtable's public API and count against existing API rate limits. Creating
records is limited to 10 records per request, records created through MCP count
against base limits, and tool behavior may change over time. Because the AI
tool can read and edit data on the user's behalf, connect only trusted clients
and require review before bulk updates, base creation, or sensitive data
changes.
Supported Transports
streamable_http
URL: https://mcp.airtable.com/mcp
streamable_http
URL: https://mcp.airtable.com/mcp
Frequently Asked Questions
- When should an AI agent use the Airtable MCP server?
- Use it when an agent needs to search, analyze, create, or update Airtable records, work with interface-visible data, or create new bases in workspaces where the user has sufficient permissions.
- What does the Airtable MCP server add to an AI agent's capabilities?
- It gives the agent live access to permitted Airtable bases and interfaces, allowing it to answer questions about current records, create and update data, and summarize structured information instead of relying on copied exports or static model knowledge.
- What can an AI agent access or manage through Airtable MCP?
- Depending on the user's Airtable permissions, the agent can read accessible records, update editable records, create records, create new bases in eligible workspaces, and query records exposed through Airtable Interfaces. Read-only and commenter users can read data but cannot create or update it.
- How is authentication configured for the Airtable MCP server?
- OAuth is the recommended setup for official connectors and compatible MCP clients. Manual configurations can use a personal access token sent with the Authorization Bearer header. Enterprise organizations may require an admin to allowlist the MCP integration before users can connect.
- Which transport should be used for the Airtable MCP server?
- Use Streamable HTTP at https://mcp.airtable.com/mcp. Prefer OAuth when the client supports it, and use the personal-access-token header only for manual setups where OAuth is not suitable.