Set Up Bruin MCP with Claude Code

Overview

Goal - Install the Bruin MCP in Claude Code so the AI agent can use the Bruin CLI, query your data, run assets, and understand your pipelines.

Audience - Data engineers who want to accelerate their workflow by giving Claude Code full access to Bruin's capabilities.

Prerequisites

• Bruin CLI installed
• Claude Code installed
• A Bruin project with at least one pipeline (initialize with bruin init duckdb if needed)

Steps

1) Add the Bruin MCP to Claude Code

Run this command in your terminal:

claude mcp add bruin -- bruin mcp

This registers the Bruin MCP server so Claude Code can access all Bruin CLI functionality.

2) Open Claude Code in your project

Navigate to your Bruin project directory and start Claude Code. It will see your pipelines and understand the project context.

3) Ask about your pipelines

Try asking a question about your project:

"In the weather pipeline, how many assets are there?"

The agent uses the Bruin MCP to inspect your pipeline and returns a list of assets.

4) Run assets

Ask the agent to execute assets:

"Do a full refresh of the staging assets."

The agent knows how to use the bruin run command with the correct flags and will execute the run for you.

5) Query your data

Ask analytical questions and the agent will figure out which bruin query command to use:

"Which day had the coldest temperature?"

The agent generates the appropriate SQL, runs it via bruin query, and returns the results.

What the Bruin MCP enables

Once installed, your AI agent can:

• Query data across all your configured databases
• Run and manage assets using the full bruin run command
• Inspect pipelines to understand structure and dependencies
• Build new pipelines by generating asset definitions
• Compare tables across environments using data-diff

Helpful links

• Bruin MCP documentation
• Bruin CLI installation
• Video walkthrough