Three Steps to AI-Ready Data

MXCP follows a simple, declarative workflow that gets you from raw data to LLM consumption faster than any alternative.

Declare

Set up your project structure with configuration files

your-project/
├── mxcp-site.yml    # Project configuration
├── tools/           # Tool definitions
├── resources/       # Data sources
├── prompts/         # LLM templates
└── models/          # (Optional) dbt transformations & caches

...

~/.mxcp/config       # Secure credentials and secret declaration

Evaluate

Use CLI commands to validate, test, and manage your data tools

mxcp serve           # Start local MCP server
mxcp list            # List all endpoints
mxcp validate        # Check types, SQL, and references
mxcp test            # Run endpoint tests
mxcp lint            # Metadata helps LLMs
mxcp query           # Execute SQL queries
mxcp init            # Initialize new project
mxcp dbt-config      # Configure dbt integration
mxcp dbt             # Run dbt commands
mxcp drift-snapshot  # Create drift detection baseline
mxcp drift-check     # Check for schema and endpoint drift
mxcp log             # Query audit logs

Serve

Start the server and connect with Claude Desktop or any MCP client

$ mxcp serve
✓ Server started on localhost:3000
✓ MCP protocol active
12 tools loaded, 5 resources available

Claude Desktop → Executing tool: customer_insights
✓ Query executed successfully
✓ Returned 247 customer records