Connecting Stride to Claude (MCP)

Stride runs a Model Context Protocol (MCP) server, so an AI assistant such as Claude can read your training data and write to your calendar directly. Once it is connected you can ask about your own rides, runs and swims in ordinary language, and the answer comes from the same deep analysis that powers Stride's AI insights rather than from a summary you have pasted into a chat.

This is separate from the AI features inside Stride itself. The Training Planner and AI insights work on their own with no setup. The MCP server is for when you want to bring your training into a conversation you are already having somewhere else.

Before you start#

• An active Stride trial or subscription. The connection itself will complete without one, but every request that reads or writes data is refused until your subscription is active.
• A client that supports remote MCP servers. Claude, Claude Code, ChatGPT and most AI coding tools do. Adding a custom connector requires a paid plan on some of them.
• Your Stride login. You sign in to Stride as part of connecting, and approve the request there.

Connect Claude#

The steps are the same in the Claude web app and the Claude desktop app.

1. Open Settings, then Connectors.
2. Choose Add custom connector.
3. Name it Stride and enter the server URL https://api.stride.is/mcp.
4. Choose Connect. A Stride window opens. Sign in if you are not already signed in, then approve the request.
5. Claude returns to the connector list and shows Stride's tools. You are ready to ask it something.

Connect Claude Code#

In a terminal, run claude mcp add --transport http stride https://api.stride.is/mcp. Then start Claude Code, run /mcp and choose to authenticate. Your browser opens for the same Stride sign-in and approval, and the connection is remembered afterwards.

Other clients#

Anything that speaks remote MCP over HTTP works the same way. Point it at https://api.stride.is/mcp and complete the sign-in it opens. Clients register themselves, so the URL is the only thing you ever need to enter.

What it can do#

Read your training#

• List your activities across any date window, completed and planned, so it can look forwards as well as back.
• Open the full analysis of a single session: detected intervals and their compliance, W' balance, aerobic decoupling, normalised power, pacing splits, laps and climbs, time in zones, route weather with the wind split into headwind, tailwind and crosswind, and your own RPE and comments.
• Fetch the workout that was prescribed for a session, or a short-lived link to the original recorded file if it needs the raw second-by-second data.
• Read your metrics at a date or across a range: FTP, critical power and W', weight, threshold and maximum heart rate, threshold pace, and wellness values such as HRV, recovery score and sleep score.
• Pull your mean-max power curve over a period, compared with a previous window and with the durability curves that show what you can produce deep into a ride.
• Check your profile, including which integrations are connected and your subscription status.

Read the day around the training#

There is a day-by-day view that returns one row per date with everything Stride holds against that date rather than against a workout: races, holidays, illness, injury, rest days and notes, that day's wellness numbers, the night's sleep, your cycle position, and the sessions themselves. It is what makes questions about context answerable, rather than only questions about a file.

Search the workout library#

It can search the workout library in plain language, filter by tag, sport, creator and length, and open any workout in full with its intervals. That covers the public catalogue plus your own workouts and those of your teams.

Write to your calendar#

The connection is not read-only. It can schedule a planned activity, update one, delete one, save a new structured workout to your library, and record a metric such as a new FTP or weight.

Coaches and teams#

If you administer a team, the assistant can list your teams and their members, then run any of the tools above against one of those athletes. Ask about an athlete by name and it will look the right person up rather than guessing. Everything in the coach dashboard stays available in the app as usual.

Things to ask#

• "How did yesterday's threshold intervals compare with the same session a month ago?"
• "Look at my last six weeks and tell me honestly whether my endurance is improving or I am just riding more."
• "I felt awful on Tuesday's ride. What else was going on that week?"
• "Where does my power curve sit against three months ago, and how does it hold up after 2,000 kJ?"
• "Find me a 60 minute sweet spot session and put it on Thursday."
• "I tested at 285 watts this morning. Record it."
• "Go through my squad's week and tell me which three riders need a conversation."

Getting better answers#

• Give it a window. "Since the start of June" or "my last four weeks" is far more useful than "my training", and keeps the answer quick.
• Name what you care about. A single activity carries a great deal of detail. Asking about the intervals, or the zones, or the weather gets a sharper answer than asking for everything at once.
• Ask it to check before it writes. Telling it to show you the plan first, then apply it once you agree, works well and mirrors how the in-app planner behaves.
• Wait for your history to import. If you have only just joined, let the initial sync finish, otherwise the assistant is reasoning about a fraction of your training.

Permissions and privacy#

• The connection is bound to your Stride account. It can reach your own data and the athletes on teams you administer. Any other athlete is refused, whatever the assistant is asked.
• You sign in on Stride's own page, so the AI client never sees your password.
• Access is granted in short-lived tokens that the client renews quietly in the background. A connection you leave alone will eventually ask you to sign in again.
• Removing the connector in your AI client ends its access. If you need a connection cut off immediately, email support@stride.is.
• Whatever the assistant reads is then handled under your AI provider's own terms, not Stride's, so treat it as you would anything else you paste into a chat.

More on how your data is stored and used is in Data & Privacy.

Troubleshooting#

• The connector will not add. Check the URL is exactly https://api.stride.is/mcp, with no trailing slash and no www.
• It says it is not authorised. The sign-in has expired. Remove the connector and add it again, or reauthenticate from your client's connector settings.
• The tools are listed but every one fails. This is almost always the subscription check. A lapsed trial or subscription still connects and still lists tools, so that the first request can explain itself rather than the connection dying silently.
• It cannot see an athlete. You can only reach athletes on a team you administer. If you coach them outside Stride, they are not reachable this way.
• An activity is missing. The assistant sees what Stride holds, so if a session has not synced from your device yet, it is not there. Check your integrations first.