MCP Server in Baserow

Baserow’s Model Context Protocol (MCP) Server lets you connect Large Language Models like Claude, Cursor, or Windsurf directly to your workspace. You can create, read, update, and delete data using natural language prompts without writing code or API calls.

What is MCP Server in Baserow?

MCP Server makes it easy to manage Baserow data through AI assistants in less time than manual database operations.

The Model Context Protocol (MCP) Server provides seamless integration between Baserow and Large Language Models. Once configured, you can interact with your Baserow databases using conversational commands through supported LLM clients like Claude Desktop, Cursor, or Windsurf.

Integrate Baserow with LLMs using the Model Context Protocol (MCP) Server

How it works

Each MCP server endpoint generates a unique URL that grants your chosen LLM secure access to your workspace. The LLM can then perform database operations on your behalf by interpreting natural language requests and translating them into Baserow actions.

Use cases:

  • Query databases without writing SQL or formulas
  • Bulk update records through conversational commands
  • Generate reports by asking questions in plain English
  • Automate data entry workflows using AI assistance

Supported operations

The MCP server supports all standard CRUD operations through natural language prompts. We recommend using high-parameter models for optimal performance.

Operation Description Example Prompt
Create Add new records to tables “Add a new task called ‘Review Documentation’ with priority set to High”
Read Query and retrieve existing data “Find all projects due this week and show their status”
Update Modify existing records “Change the status of task #47 to ‘In Progress’”
Delete Remove records from tables “Delete all completed tasks that are older than 30 days”

Setting up the MCP Server

Prerequisites

  • An active Baserow workspace
  • A supported LLM client installed (Claude Desktop, Cursor, or Windsurf)
  • Admin or appropriate workspace permissions

Step-by-step setup

1. Access MCP settings

  • Click your workspace name in the top navigation bar
  • Select My Settings from the dropdown menu
  • Navigate to the MCP Server tab

2. Create an endpoint

  • Click Create Endpoint
  • Enter a descriptive name (e.g., “Marketing Database Assistant”)
  • Select the workspace you want to connect to, then create the endpoint.

3. Choose your LLM client After you create the endpoint, click More Details to view configuration options Select from the available options:

  • Claude
  • Cursor
  • Windsurf

4. Copy your MCP URL Baserow will generate a unique endpoint URL. Treat this URL as a password; it grants full access to modify data in your workspace.

5. Configure your LLM client Follow the specific configuration instructions for your chosen LLM (detailed in sections below)

Configure your LLM client

Claude Desktop

  1. Open Claude Desktop settings (⌘+, on Mac)
  2. Navigate to the Develop tab
  3. Click Edit Config to open claude_desktop_config.json
  4. Add the JSON configuration below
{
  "mcpServers": {
    "Baserow MCP": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "YOUR_MCP_URL_HERE"
      ]
    }
  }
}

Cursor

  1. Open Cursor settings (⇧+⌘+J on Mac)
  2. Navigate to the MCP tab
  3. Click Add MCP Server
  4. Paste the JSON configuration
{
  "mcpServers": {
    "Baserow MCP": {
      "url": "YOUR_MCP_URL_HERE"
    }
  }
}

Windsurf

  1. Open Windsurf Settings → Advanced Settings (or use Command Palette → “Open Windsurf settings page”)
  2. Scroll to the Cascade section
  3. Click to add a new server or view the raw JSON config file at mcp_config.json
  4. Add the JSON configuration
{
  "mcpServers": {
    "Baserow MCP": {
      "serverUrl": "YOUR_MCP_URL_HERE"
    }
  }
}

Replace YOUR_MCP_URL_HERE with the MCP URL copied from your Baserow settings.

Security best practices

⚠️ Critical security note: Your MCP URL contains credentials that grant full access to your workspace data.

Do:

  • Store your MCP URL securely like a password
  • Revoke and regenerate URLs if compromised
  • Use descriptive endpoint names to track usage
  • Create separate endpoints for different use cases

Don’t:

  • Share your MCP URL publicly or in version control
  • Store it in unsecured locations
  • Use the same endpoint across multiple team members
  • Post screenshots containing the full URL

Frequently asked questions

What LLMs are compatible with Baserow’s MCP Server?

Currently, Baserow’s MCP Server works with Claude Desktop, Cursor, and Windsurf. Any client that supports the Model Context Protocol specification can potentially connect to Baserow using the MCP endpoint.

Can I create multiple MCP endpoints for different purposes?

Yes, you can create multiple endpoints within the same workspace. This is useful for organizing access by project, team member, or use case. Each endpoint gets a unique URL with independent access controls.

What happens if my MCP URL is exposed?

If your MCP URL is compromised, immediately delete the endpoint in your Baserow MCP settings and create a new one. The old URL will no longer function, preventing unauthorized access to your workspace.

Do I need programming knowledge to use MCP Server?

No programming required. The MCP Server translates natural language prompts into database operations, making it accessible to non-technical users who can describe what they need in plain English.

What are the limitations of using AI to manage database operations?

While powerful, AI interpretations depend on model quality and prompt clarity. Always verify critical operations, especially bulk updates or deletions. We recommend using high-parameter models and reviewing actions before confirming destructive changes.

Related content

Still need help? If you have questions about MCP Server setup or usage, visit the Baserow community or contact support.