Connect table to your AI client with MCP — table

What MCP does

MCP, or Model Context Protocol, is a standard way for AI clients to connect to tools and data from other apps. table's MCP server lets compatible AI clients request access to your table account and use approved table tools.

The MCP server is hosted separately from the main app and uses its own authorization flow. This keeps the MCP boundary explicit.

Set up table MCP

Choose your AI client and connect it to table's remote MCP server. Each client uses the same production server URL, but the setup step differs.

Check your table account

Pick your AI tool

Choose the client you want to connect. We will show the setup command or settings path for that tool.

ClaudeChatGPT / CodexGeminiCursorOther

Register table MCP for Claude

Connect table from Claude's official connector directory, or use Claude Code if you prefer terminal setup.

Connect table from Claude Directory

table is listed as an official Claude connector, so you can open the connector page and connect your account directly from Claude.

Open connector

Setup command

claude mcp add --transport http --scope user table https://mcp.usetable.ai/mcp

Authorize table

Start authorization from your AI client. table will open a browser sign-in and approval screen so you can confirm the client before it can use tools.

Start a new Claude Code session after adding the server.
Open the MCP menu and authenticate table when Claude asks for access.

MCP status menu

claude /mcp

Test the connection

Ask your client to use table with explicit contact details. Keep your first prompt simple so you can confirm the authorization and tool call worked.

Prompt to try

Create a contact in table from this email.

Approving and revoking access

When an MCP client requests access, table shows an authorization screen that identifies the client and asks you to approve or cancel.

Approving a client lets that client use the table MCP tools available to your account. If a client does not look familiar, cancel the authorization flow.

You can revoke access for clients you no longer use or do not recognize.

When to use MCP

Use MCP if you want to:

Save relationship context from an AI client into table
Create or update contacts from explicit contact data
Let an assistant help maintain your CRM without giving it broad database access
Build workflows that start outside table but end in your table account

Do not use MCP for sensitive workflows unless you trust the client you are connecting.

Data access

The current MCP surface is intentionally narrow.

The connector can create or update one contact from explicit contact data provided by the MCP client. It may write contact fields such as name, email, phone, organization, job title, location, birthday, and status.

It does not provide raw database access.
It does not let MCP clients browse all table data.
It uses the authenticated table user from OAuth, not a user ID supplied by the MCP client.
The current MCP surface writes contact data. If read tools are added later, this section should be updated.

What to ask after setup

Once the first test works, use MCP for small, explicit contact updates like:

Save a person from a pasted email signature.
Add a founder, investor, or candidate from details you provide in chat.
Update a known contact with a new company, role, location, phone number, or status.
Turn meeting notes into a contact update when the person and details are explicit.

Security model

MCP clients must authenticate.
table uses the authenticated OAuth user, not a user ID supplied by the MCP client.
table validates tool input before writing contact data.
MCP tools do not expose raw database rows.
Errors are kept stable and bounded.
table avoids logging personal data such as names, emails, phone numbers, organizations, prompts, tokens, and provider payloads.

Privacy and support

Support: support@usetable.ai.

Revoke access for clients you no longer use or do not recognize.

Troubleshooting

I do not see table in Claude Connectors: open the table connector page from Claude Directory, then connect from there.
Authorization opens but does not complete: disconnect and reconnect table, then start the authorization flow from your AI client again.
The client name or redirect URL looks unfamiliar: cancel authorization and contact support.
Tool call fails: check that your table account is active and try reconnecting.
Invalid authorization request: start the authorization flow from your MCP client.
Unknown client: cancel the request if the client name or redirect URL is not expected.
Approved client cannot call tools: disconnect and reconnect the MCP server in the client.
Unrecognized client: cancel authorization and do not approve the request.