Connect SeaTable with Cursor

In this guide, you will connect Cursor to your SeaTable base. Cursor is an AI-powered code editor whose chat function is excellent for interacting with your data. Once set up, you can ask questions about your SeaTable data in the Cursor chat and have entries edited directly. The setup takes about five minutes.

Prerequisites

• A SeaTable Cloud account with at least one base
• Cursor (available at cursor.com )
• A Cursor subscription that supports MCP (Pro plan or higher)

Step 1: Create an API token in SeaTable

The API token gives Cursor access to a specific base. You decide whether Cursor can only read or also write data. For instructions on creating an API token, see the article Creating an API Token .

Step 2: Add MCP server in Cursor

Cursor manages MCP servers through the settings. Open them via Cursor → Settings → Tools & MCP or use the keyboard shortcut Cmd+Shift+J (Mac) or Ctrl+Shift+J (Windows/Linux).

1. Click + Add new MCP Server.
2. Add the following configuration to the .cursor/mcp.json file that opens:

{
  "mcpServers": {
    "seatable": {
      "type": "streamable-http",
      "url": "https://mcp.seatable.com/mcp",
      "headers": {
        "Authorization": "Bearer INSERT-YOUR-API-TOKEN-HERE"
      }
    }
  }
}

3. Replace INSERT-YOUR-API-TOKEN-HERE with the token from Step 1.

You can connect multiple bases at the same time. Create a separate entry for each base with a unique name, for example seatable-crm and seatable-projects.

Step 3: Verify the connection

After saving, the MCP server should appear as connected in the Cursor settings under Tools & MCP — indicated by a green dot next to the name. If a red dot appears instead, check the URL and token again.

Now open the Cursor chat (Cmd+L / Ctrl+L) and ask a first test question:

“What tables are in my SeaTable base?”

Cursor will then query the table structure via the MCP server and list all tables with their columns. If this works, the connection is established.

Asking your first questions

You can now ask questions about your data in the Cursor chat as if you were talking to a colleague. Here are a few examples to try:

• “How many entries does the Contacts table have?”
• “Show me all entries where the status is ‘Open’.”
• “Summarize the data in the Revenue table by month.”

Your questions should refer to tables and columns that actually exist in your base. If you are unsure, simply ask about the structure first.

You don’t need to type table and column names exactly. Cursor recognizes minor typos and corrects them automatically. Feel free to write “Contacts” instead of “contacts” or “Projects” instead of “projects” — Cursor will find the right table.

Common issues

Next steps

• Asking good questions: How to get the best answers
• Permissions and data privacy for AI agents