Strava MCP

Strava MCP bridges Claude to your Strava account via OAuth, letting you pull runs, rides, and training data straight into a conversation. You authenticate once, then ask Claude to analyse your last week's mileage, compare power outputs across rides, or spot patterns in your training load. It's a proper OAuth flow, not a hacky API key setup, which means it respects Strava's rate limits and user permissions. I'd reach for this when I want conversational access to my training history without opening the Strava app or exporting CSVs. Ask Claude to summarise your hardest runs this month, and it fetches the data, calculates averages, and spots trends. It's particularly good for runners and cyclists who want a training log assistant that can answer natural-language questions like "How many kilometres did I ride in January?" or "Show me my longest run this year." The OAuth setup is more involved than dropping an API key into an env file, but it's the right way to handle user data. Quirks: you need a Strava API application, which means creating one in the Strava developer portal even if you're just using it for yourself. That's a five-minute job, but it's not instant. The server doesn't cache responses, so repeated queries hit Strava's API each time. If you're asking Claude to reanalyse the same week multiple times in one session, you might bump into rate limits. The docs assume you know your way around MCP config files, so complete beginners might need to cross-reference the Claude Desktop setup guide. Who shouldn't bother: if you only check Strava once a week or you're happy with the app's own stats, this is overkill. It's for people who want to interrogate their data conversationally or build custom coaching workflows. If you don't run or ride regularly, or you're not curious about training load trends, skip it. But if you're logging multiple activities a week and you want Claude to act as a training analyst, this is the cleanest way to connect the two.