Labsco
MCP SERVER

AI Endurance

by ai-endurance

Talk to your AI Endurance training plan — move or create structured workouts, read power curves and pace splits, check HRV recovery, and get ML race-time predictions.

Health, Fitness & Wearables
Summary
It changes your plan, not just reports on it.

Most fitness integrations are read-only dashboards; this one moves workouts, creates intervals and pushes them out to Garmin, TrainingPeaks and Zwift, which means a conversation can rewrite the week. Zone updates take units in every value — pace as `mm:ss /km` or `mm:ss /mi`, power as `XXX W` — so a bare number will not do. The durability and computed-analytics tools are the differentiator: normalized power, intensity factor, time-in-zone and fade tables come back computed rather than as raw streams you have to process.

What it is

The hosted MCP server for the AI Endurance training platform, covering 27 tools across your plan, activities, recovery and zones. It is a two-way surface: you can read what happened and change what is scheduled, including creating structured interval workouts for cycling, running and swimming.

What you get
  • Planned workouts for a date range, defaulting to the next 14 days, with a lightweight summary mode or full machine-readable step structure — `getPlannedWorkouts`
  • Workouts moved, skipped, re-advised or re-intensified, synced through to Garmin, TrainingPeaks and Zwift — `changeWorkoutDate`, `skipWorkout`, `changeWorkoutAdvice`, `changeWorkoutIntensity`
  • Structured workouts created for ride, run, swim and strength — `createRideRunWorkout`, `createRideRunWorkoutAdvanced`, `createRideRunWorkoutByIntensity`, `createSwimWorkout`, `createStrengthOtherWorkout`
  • Activity detail by sport, with a deeper variant for each — `getCyclingActivity`, `getRunningActivity`, `getSwimmingActivity`, `getOtherActivity`, and the `…Detail` versions; raw streams analysed server-side by `analyzeActivityStream`
  • Profile, zones and availability — `getUser`, `setZones` for Run or Ride with Endurance, Tempo, Threshold and VO2Max upper bounds, `getAvailability`
  • Recovery, predictions, nutrition, plan progress and race goals — `getRecoveryModel`, `getPrediction`, `getNutritionModel`, `getPlanProgress`, `getRaceGoalEvent`
  • Indoor, virtual and erg detection corrected, and bad-sensor activities excluded from analysis — `setActivityFlags`
Requirements

An AI Endurance account, authorized over OAuth. In Claude.ai add a custom connector with the remote MCP server URL `https://aiendurance.com/mcp`. In ChatGPT, connect it from the plugin directory — ChatGPT also renders interactive widgets for most tools. Any MCP 2025-06-18 compliant client can connect over streamable HTTP with OAuth against the authorize and token endpoints, requesting the read and write scopes; an SSE configuration is documented as legacy.