Introduction to the MCP Server
Learn what the Braking Lab MCP Server is and how it connects AI clients to your racing data.
The Braking Lab MCP Server is an intelligence layer that gives any AI assistant full access to your telemetry data, race calendar, track notes, coaching reports, and hardware setups. Instead of separate tools, it exposes a single code-mode sandbox where AI writes TypeScript to query your data.
How It Works
You ← AI Assistant (Claude, ChatGPT, etc.) ← MCP Protocol → MCP Server → Your Data- You chat with an AI assistant (Claude Desktop, ChatGPT, Cursor, etc.)
- The AI writes TypeScript code using the
brakinglab.*namespace - The MCP Server transpiles and executes the code in a secure sandbox
- Each function queries your data through Supabase with ownership verification
- Results flow back to the AI for interpretation
Everything happens in real time. The AI has the same access to your data that you have in the web app — nothing more, nothing less.
What Makes It Different
| Traditional MCP | Braking Lab Code-Mode |
|---|---|
| Dozens of separate tools | One execute_code tool |
| Fixed parameters | Full TypeScript with type hints |
| Simple queries | Compose multiple functions in one call |
| Limited flexibility | AI writes custom analysis logic |
The code-mode approach means the AI can do things like “compare my last 5 sessions at Spa, find my weakest corner, and generate a coaching report” — all in a single execution.
What’s Available
The server exposes 55 functions organized into 11 domain groups:
| Group | What It Does |
|---|---|
| Sessions | Browse and explore telemetry sessions |
| Stints | Lap-by-lap data, corner metrics, consistency |
| Corners | Entry/apex/exit speeds, lateral G, steering corrections |
| Braking | MRP detection, trail braking, brake pressure profiles |
| Coaching | Save and retrieve structured coaching reports |
| Races | Calendar management, session linking, race prep |
| Notepads | Track corner notes, pins, videos |
| Analysis | Lap comparison, progress tracking, weak corner identification |
| Strategy | Auto-generate race strategy from practice data |
| iRacing | Series search, schedule lookup, race creation |
| Vault | Datapacks, hardware setups |
Requirements
To use the MCP Server, you need:
- A Braking Lab account with telemetry data synced via Capture
- An MCP-compatible AI client (Claude Desktop, ChatGPT, Cursor, Windsurf, or any MCP client)
- An active internet connection
Getting Started
Ready to connect? Follow the Setup Guide to connect your AI client to the MCP Server in under 5 minutes.