You Can Now Manage Your Training From Claude, Cursor, and Claude Desktop
by The Next Race
Your training data has been sitting behind a UI for a while now. Races, plans, workouts, zones — useful, but you have to click around to get to them.
That changes today. TheNextRace now exposes a proper Model Context Protocol (MCP) server, which means you can connect it to Claude Desktop, Claude Code, Cursor, or claude.ai directly and have a conversation with your training data.
What that actually looks like
Open Claude Desktop and ask:
What races do I have planned this season?
Claude reaches into your TheNextRace account, pulls back your races, and answers — race names, dates, priority levels, locations. No screen-switching, no copy-paste.
A few more that work today:
Add the Berlin Marathon to my plan as my A-race on September 27.
Save this as a library workout: 6×3min at 5K pace with 3min jog recovery, run, called "VO2 6×3."
What are my running zones? My FTP?
Show me the workouts I've saved for swimming.
Update my FTP to 285 watts and my run threshold pace to 4:05 per km.
Delete my "Spring base block" plan.
Each one of those is the AI calling a tool against your account, on your behalf, with the same authentication you'd use signed in to the website.
Setting it up
You need an API token first. Open your TheNextRace settings, click the API Tokens tab (the key icon), and generate one. The full token is shown once — copy it somewhere safe.
Then pick your client:
Claude Desktop, Cursor, or Claude Code — the easy path is to install the small companion CLI:
npm install -g @thenextrace/tnr
tnr login
…then add this to your client's MCP config:
{
"mcpServers": {
"tnr": { "command": "tnr", "args": ["mcp"] }
}
}
Restart the client. The tools show up automatically.
Claude.ai (web) — go to Settings → Connectors, add a custom connector, and paste:
https://www.thenextrace.app/mcp
Sign in when prompted, click Allow, and you're done. The web flow handles the rest.
What this is good for
The honest answer: anything that's annoying to do via clicks but obvious to say in words.
Setting up a new race entry mid-conversation while you're discussing your goals. Asking what your zones are without having to navigate to your profile. Reviewing your races against a calendar. Pasting a session your coach gave you and saving it as a library workout in one go.
We've also seen people use it as a kind of training journal — talking through a session, asking the AI to add it as a workout with the right structure, all without leaving the chat.
What it doesn't do (yet)
The AI can read and write your training data. It can't yet generate a training plan for you — the "give me a 16-week marathon block based on my FTP, my available hours, and my race calendar" feature isn't built. That's a real product feature with real periodization logic, and we're working on it.
For now: if you have a coach, tnr makes it easier to keep your TheNextRace account in sync with what they prescribe. If you're self-coached, it's a smoother way to manage your own data.
Why we shipped this
The number of athletes using AI tools day-to-day has grown sharply in the last 18 months. If we'd waited until "AI coaching" was a complete product before integrating, we'd have missed the part of the wave where people just want their training stack to talk to the tools they already use.
So this is the foundation. Plan generation comes next — once it lands, the same conversation surface gets a lot more powerful.
If you try it, we'd love to hear what you used it for. The full setup guide and the source for the CLI are at @thenextrace/tnr on npm.