Strava MCP Server

Overview

The Tomekkorbak Strava MCP Server bridges the gap between conversational AI assistants and the rich activity data available on Strava. By exposing a set of well‑defined tools over the Model Context Protocol, it allows models such as Claude to retrieve, filter, and interrogate a user’s athletic history without the model having to manage OAuth flows or API rate limits. This eliminates boilerplate code for developers who want to build fitness‑centric chatbots, dashboards, or analytics pipelines that respond naturally to user queries.

At its core, the server implements four activity‑query tools. The tool fetches a configurable number of the most recent workouts, while limits the result set to a sliding window of days. For more precise temporal queries, accepts ISO‑formatted start and end dates, and returns the full details of a single activity. All responses are normalised into a consistent schema that includes distance, speed, elevation, and caloric burn metrics, ensuring that downstream models can parse the data without custom adapters.

Developers benefit from this consistency and the server’s built‑in error handling. Invalid date strings, authentication failures, or network hiccups trigger clear, human‑readable messages that the model can surface to users. The server also manages token refresh automatically, so developers need only provide their Strava client credentials and a refresh token once. This lightweight integration means that an AI assistant can answer questions like “Show me my longest run in the past month” or “What was my most recent cycling activity?” by simply invoking a single tool, without any additional code.

Typical use cases include fitness coaching bots that generate personalized summaries, activity‑tracking apps that surface insights via natural language, or data science pipelines where models annotate and interpret Strava logs. Because the MCP server exposes a small, focused API surface, it can be embedded in existing Claude Desktop or Web configurations with minimal effort. The server’s design prioritises clarity and reliability, making it a dependable component for any project that wants to combine conversational AI with real‑world athletic data.