mapbox/mcp-server

What it solves

This project provides a Model Context Protocol (MCP) server that allows AI agents and applications to access Mapbox's geospatial intelligence platform. It enables AI to understand locations, perform geocoding, search for points of interest, calculate routes, and visualize geographic data directly within a chat interface.

How it works

The server acts as a bridge between an AI client (like Claude Desktop, VS Code, or Cursor) and Mapbox APIs. It implements the MCP standard to expose a set of tools and resources to the LLM. These include:

• Mapbox API Tools: Tools for geocoding, routing, matrix calculations, and POI search.
• Offline Geospatial Tools: Local computations using Turf.js for distance, area, bearing, and spatial analysis without needing API calls.
• Visualization Tools: A specialized render_map_tool that creates interactive Mapbox GL JS maps as HTML panels in supported clients, as well as a tool for static map images.
• MCP Resources: Read-only access to static reference data, such as the Mapbox category list.

Who it’s for

Developers building AI travel assistants, logistics optimizers, location-based recommenders, or any AI agent that needs to reason about the physical world and "where" things are.

Highlights

• Interactive Map Rendering: Renders live, interactive maps directly in the chat via the MCP Apps protocol.
• Comprehensive Spatial Intelligence: Combines global geocoding, multi-modal routing, and route optimization (Traveling Salesman Problem).
• Offline Capabilities: Includes a suite of tools for distance, area, and polygon calculations that work without internet connectivity.
• Broad Integration: Ready-made setup guides for Claude Desktop, VS Code, Cursor, and the Goose AI agent framework.