MCP

Strava's MCP Connector for Claude: What It Does — and What It Can't (Yet)

Joules June 23, 2026 · 7 min read

In June 2026, Strava released an official MCP connector for Claude. Add it and you can ask Claude real questions about your training history — and it answers from your actual rides. It's a genuinely useful upgrade. It's also read-only, which means it stops one step short of coaching. Here's the honest breakdown of what it does, where the line is, and how to cross it.

What MCP is, in one paragraph

MCP — the Model Context Protocol — is a standard way to give an AI assistant live access to a tool or data source. Instead of pasting your data into the chat, you add a connector once and the assistant can query it directly inside a normal conversation. Strava's connector is one of these: it plugs your ride history into Claude. (Worth stating up front: custom MCP connectors work on every Claude plan, including the free tier — free Claude allows one custom connector. On ChatGPT they need a paid plan, Plus/Pro.)

What the Strava MCP connector actually lets you do

Once it's connected, Claude can read your Strava data and reason over it in plain language. That unlocks a real category of questions that used to mean exporting CSVs or clicking through charts:

• Summarize a block. "How much did I ride in the last four weeks, and how does that compare to the four before?"
• Spot patterns. "Are my Saturday rides getting slower at the same heart rate?"
• Pull up specific efforts. "What was my best 20-minute power this month?"
• Sanity-check a feeling. "I feel flat — has my volume actually dropped, or is it in my head?"

This is great. Conversational analysis of your own history is exactly the kind of thing chat assistants should do, and having it read from the source instead of your memory makes the answers trustworthy. If all you want is a smarter lens on what you've already done, the Strava connector is worth adding today.

The read-only line — and why it matters for coaching

The connector reads your past. It does not plan your future. That's not a knock — it's the design. But it's the exact line that separates analysis from coaching, so it's worth being precise about what falls on the other side:

• No season built backward from a goal. Coaching starts with an A-race and a date and works backward. Reading history can't set that target or structure the months toward it.
• No prescription for tomorrow. "What should I ride tomorrow?" isn't a question about the past. It needs a plan, your current freshness, and the intent of the week — none of which live in your ride feed.
• No accountability loop. A good plan ramps on purpose and adjusts when life interrupts. That requires writing the next week, not just reading the last one.

Put simply: a read-only connector can tell you that you did three hard rides last week. It can't decide whether next week should get harder or back off — because deciding that, and committing it to a plan you'll be held to, is the coaching part.

Read your past, plan your future — the two compose

The clean way to think about it: training has a read side and a write side. Strava's MCP connector is the best read side going — your real history, queryable in Claude. The missing write side is a connector that gives the same assistant actual coaching tools: an interview to build your athlete profile, a season planner, weekly structure, and a place to log sessions so the plan progresses and stays honest.

That write side is the category Joules is built for. It's an MCP app that gives Claude (or ChatGPT) a cycling coach's brain: it interviews you, builds your season, plans every week, and grades each week against your real rides — device-free, just duration and perceived effort if that's all you've got. Add it the same way you added Strava: paste the connector URL once.

And they stack. Run both connectors in the same Claude session and you get the full loop: Strava tells the assistant what you did; Joules tells it what to do next, then checks whether you did it. Read side and write side, in one conversation.

That's one line of one day. Want to see the whole thing before you add anything? Build your own sample week right in your browser — pick your goal event and weekly hours and watch the coach assemble it, plus see how it grades a week against your real rides and adjusts the next one. No account, no connector, nothing to install. (Prefer a written walkthrough? Here's one real week, day by day.)

So, can AI coach you now?

Closer than ever — and Strava's connector is a real piece of it. The honest summary: a read-only connector makes the AI you already pay for a sharp analyst of your past. Add the planning-and-accountability half and it becomes something it wasn't last year: a coach that remembers, plans forward, and adjusts week by week. The two aren't competitors. They're two halves of the same loop.

New to the idea? Start with Can AI Be Your Cycling Coach? A Practical Guide for 2026 for the full setup walkthrough.