Getting Started with the Embedded MCP Server in Baserow

Baserow now ships with a powerful, embedded MCP server, giving you instant access to AI-powered automations without hosting anything yourself. As long as your Baserow instance is running, the Model Context Protocol (MCP) server is live and accessible for any remote MCP client that supports the standardized interface. This allows interacting with your Baserow data using natural language—perfect for automating tasks, building internal tools, or even assisting with development workflows.

In this post, we’ll walk through how to configure Baserow’s MCP endpoint, connect it to Claude, and build a simple CRM use case powered by AI. We’ll also explore how this setup works seamlessly with role-based access control, ensuring your data stays secure, even with external automation tools.

Why Using MCP Server

The MCP server offers a secure and flexible bridge between large language models (LLMs) and your Baserow data. With zero hosting or extra setup, Baserow’s embedded server enables AI assistants to read and write data using a universal context protocol MCP connection. Tools like Claude Desktop, Cursor, and automation platforms like n8n can all leverage this to access real-time data and issue commands in a human-readable format.

The MCP server supports all standard CRUD operations through natural language prompts.

Setting Up the MCP Endpoint in Baserow

To start using the MCP server:

1. Open your Baserow workspace.
2. Click your workspace name in the top navigation.
3. Go to My Settings.
4. Click on the MCP Server tab.
5. Select Create New Endpoint.
6. Give the endpoint a descriptive name (e.g., “Claude CRM Assistant”).
7. Choose the workspace it will operate on.
8. Click More Details and choose Claude as your preferred LLM.
9. Your unique MCP URL will be generated—treat this as securely as a password.

Configuring Claude Desktop with the MCP URL

To use your endpoint in Claude Desktop:

1. Open Claude settings (⌘+,).
2. Go to the Develop tab and click Edit Config.
3. Insert this configuration into claude_desktop_config.json:

Now you can issue natural language commands through Claude that interact directly with your Baserow workspace.

To list your tables, just type:

“List all tables in my workspace.”

Claude will respond with your existing tables and structure.

Building a Simple CRM System with MCP in Baserow

Let’s bring it all together with a real-world CRM use case. This CRM will include two tables: Customers and Tasks.

Step 1: Create a New Database

In Claude, send:

“Create a new database called CRM.”

Step 2: Add Tables and Fields

Next:

“In CRM, create a table called Customers with fields: Name (text), Email (text), and Status (single select with New, Active, Inactive).”

Then:

“Create a Tasks table with fields: Title (text), Due Date (date), Assigned To (link to Customers), and Status (single select with To Do, In Progress, Done).”

Within moments, your CRM system is structured and ready.

Managing Tasks Using Natural Language

The MCP tool inside Claude lets you operate your CRM entirely through natural prompts.

Creating a New Task

“Add a task called ‘Follow up with Alice’ due next Friday, assigned to Alice Johnson, and set status to To Do.”

Finding a Task

“Find all tasks with status To Do.”

Updating a Task

“Change the status of ‘Follow up with Alice’ to In Progress.”

Deleting a Task

“Delete the task called ‘Follow up with Alice’.”

These actions all occur instantly—powered by the context protocol MCP server, fully integrated within Baserow.

The power of the MCP server doesn’t stop with CRM or task management. Because it uses a standardized interface, you can plug it into any workflow that benefits from real-time data access, especially when paired with AI-powered assistants.

Collaborative Team Dashboards

Imagine your marketing team maintains a campaign tracker in Baserow. With the MCP endpoint connected to Claude, any team member could ask:

“Show me all campaigns with a budget over $10,000 that are scheduled for this quarter.”

Or:

“Update the status of the ‘Spring Launch’ campaign to Completed.”

Non-technical users can interact with structured data using natural language, making collaboration smoother and more inclusive.

AI-Powered Meeting Assistants

Hook the MCP server into a meeting bot using n8n. After each call, the bot could:

1. Summarize the conversation.
2. Extract action items.
3. Send prompts to the MCP server such as:

“Create a new task titled ‘Prepare Q3 forecast’ assigned to Sam with a due date of next Monday.”

This dramatically reduces manual note-taking and ensures follow-ups are never missed.

Developer Use with Cursor and AI Pair Programming

Using Cursor, developers can integrate their code editor directly with Baserow through the MCP tool.

Prompt:

“List all feature requests with status ‘Waiting for dev’.”

Or while testing:

“Add a bug report titled ‘Login fails on Safari’, assign it to QA.”

You can even fetch schema data inline:

“What are the fields in the Bugs table?”

By enabling AI to interact directly with the database, Cursor + MCP server turns into a powerful assistant that helps manage technical and project-related tasks right from the command line or editor.

With all these capabilities, Baserow’s MCP server offers a unique bridge between AI assistants, tools and data sources, and your day-to-day workflows. Whether you’re building a product, managing a team, or analyzing customer feedback, the model context protocol MCP enables smarter, faster, and more secure access to your data.

Role-Based Access Control (RBAC)

All actions performed through the MCP server respect Baserow’s role-based access control.

If the user connected to the MCP endpoint doesn’t have permission to edit a table, then write actions like “Add a task” or “Delete record” will fail gracefully. This ensures that your AI automations stay within the boundaries of your organization’s security model.

You can also create read-only endpoints that only permit data queries, not modifications—ideal for reporting or analytics use cases.

Beyond Claude: Using MCP with Other Tools

The embedded MCP server isn’t limited to Claude. You can use it with any compatible MCP clients or platforms that support the Model Context Protocol.

Cursor for Developers

Cursor, a powerful AI coding environment, can integrate with your MCP server to help developers interact with Baserow during app development.

For example, you might:

• Query project specs stored in Baserow.
• Update task statuses via code commits.
• Pull up test cases by typing:

“List all tests related to login validation.”

n8n with AI Nodes

With n8n, you can combine AI-powered logic with your open source Baserow backend.

• Use the AI node to parse emails or messages.
• Generate action prompts.
• Send those prompts through the MCP server to add or update records in Baserow.
• Visualize or send alerts via Slack, Google Maps, or email based on task data.

Example Workflow:

1. New customer signs up via form.
2. n8n reads the input.
3. AI node generates prompt:

“Add a customer named John Doe with email john@example.com and status New.”

1. Prompt is sent to the MCP server.
2. Customer is created in Baserow automatically.

This is automation made simple—by combining tools and data sources through a standardized interface.

Final Thoughts

The MCP server inside Baserow is more than just a technical feature—it’s a game changer for building intelligent, responsive systems using natural language. Whether you’re managing a CRM, tracking engineering tasks, or enabling AI workflows with Claude Desktop, Cursor, or n8n, Baserow’s context protocol MCP servers are ready to power your ideas.

With support for secure access, seamless LLM integration, and full CRUD capabilities, Baserow becomes an ideal foundation for building modern, AI-powered internal tools.

Try it today and enable Baserow AI to work directly with your data.

Try Baserow today