CoachboxMCP
English

Connect your own AI to the Coachbox data of you and your athletes

Coachbox already has an AI coach built in: it knows your athletes, has the theory behind periodization down and drafts training for you. Prefer to work with your own assistant or your own agents? Connect those to the same data over MCP. The connection is read-only, you approve it once on a Coachbox page, and you can revoke it at any time.

https://mcp.coachbox.app/mcp

Read-only · OAuth 2.1 · For coaches

Claude
Is Emma on track for her half marathon on 2 June?
I'll look at her recent training first.
coachbox - get_workouts(athlete: "Emma", last: "7d")
coachbox - get_milestones(athlete: "Emma")
coachbox - get_annual_plan(athlete: "Emma")
Yes, she is on track. Tuesday's intervals hit all five 4-minute blocks at 312 W, she has completed 86% of the Build block, and the plan peaks two weeks before race day. I would keep Thursday easy.

[you]Ask in plain language
Type a question the way you would ask a colleague.

[toolcall]get_workouts()
The MCP reads Emma's training week over the read-only connection.

[toolcall]get_milestones()
It looks up the race date in her milestones.

[toolcall]get_annual_plan()
It checks where she sits in the season plan.

[LLM]Claude answers
The answer is grounded in her real numbers, not a guess.

What you can ask

"How did the interval sessions of my triathletes go last week?"

The assistant lists their workouts and reads the analysis of each session.

"Prepare my monthly review call with Tom."

It combines workouts, metrics, notes and race results into one summary.

"Does the coming training block still match the season plan?"

It checks the annual plan and milestones against the planned workouts.

We already have one. Yours is welcome too.

Built in

The Coachbox coach is already there

Every team on an active plan already has the AI sparring partner in the app. It knows your athletes, sees their history and helps you draft schedules. Nothing to connect, nothing to set up.

Open Coachbox
This page

Your own AI or agents on top

Already working with Claude, ChatGPT or your own agents? You do not have to trade them in. Connect them to the same training data over MCP, read-only, in the assistant you already have open.

Connect your assistant

Connecting requires a coach account on a team with an active plan or a running trial.

  1. Open claude.ai → Settings → Connectors.
  2. Choose Add custom connector, name it Coachbox and enter https://mcp.coachbox.app/mcp.
  3. Click Connect. You land on the Coachbox approval page: sign in and pick the team you want to share.
  4. Approve read access. Claude can now answer questions about your training data.
Official docs ↗

What the assistant can see

The assistant reads the data below for the one team you approve, nothing else. It cannot create, change or delete anything: this endpoint has no write access, by design.

Sensitive integration data such as passwords and connected-platform tokens is never exposed. Every response is filtered server-side before it leaves Coachbox.

How access is protected

You approve, not the AI

Access starts on a Coachbox approval page where you sign in yourself. The assistant never sees your password.

Read-only, one team

The connection carries a single permission: read the training data of the team you selected. There is no write permission on this endpoint.

Short-lived tokens

Access tokens expire after one hour and rotate automatically, following the OAuth 2.1 standard with PKCE.

Revoke at any time

Disconnect under Settings → Connected apps in Coachbox. Every request is checked server-side, so revoking takes effect immediately.

New to Coachbox?

Try the full platform free for 14 days, MCP included. No credit card required.

Start free trialBook a demo