Your own AI and the platform data: what MCP is and how to connect it

Chatting with an AI about your training runs into one problem: it cannot see your data. You end up retelling it in your own words — and the details that the answer depends on get lost in the retelling.

Training Endurance solves this differently. The platform runs an MCP server, so your AI client reaches the data directly: not your summary of it, but the workouts, measurements, meal plans and zones themselves.

What MCP is

Model Context Protocol is an open protocol for connecting an AI client to an external service. The model receives tools rather than text: "show the workouts for this week", "give me the effective running zones", "calculate today’s meal plan". Which tool to call, the model decides from your question.

The difference from pasting numbers into a chat is simple: the model works with current values, sees how they relate, and can ask for anything it is missing. You do not have to guess in advance which figures will matter.

What the AI gets in Training Endurance

The server exposes more than a hundred and fifty tools. The essentials:

• Calendar and workouts — planned and actual, workout structure, segments, markers, extended workout data;
• Metabolic profile and zones — effective heart-rate, speed, power and lactate zones, and the history of how they changed;
• Nutrition — meal plans, products and ingredients, daily norms, menu calculation and application;
• Health and wellness — measurements, the daily log, hydration, electrolyte losses;
• Athletes and groups — for coach accounts: the athlete list, tags, bulk operations;
• Plans and library — ready-made training plans, applying and undoing them;
• Synchronization — activities from connected services and devices.

What to ask

Phrase it the way you would to a person. A few prompts that work right away.

For an athlete:

• "Break down yesterday’s long run: how much time did I spend above LT2, and did I overcook the first third?"
• "I race in three weeks. Look at the past month and tell me what my preparation is missing."
• "Work out how many carbs and how much fluid I need for a four-hour trail in the heat, given my profile."

For a coach:

• "Show the athletes whose weekly load grew by more than 30% over the past two weeks."
• "Compare planned against actual for Ivan this month: where does he consistently push the recovery sessions?"
• "Build next week as a recovery microcycle for the group that has just raced."

For a nutritionist:

• "Check this athlete’s meal plan against their load this week: where are they short on carbohydrates?"
• "Build a three-day carbohydrate loading plan before the race, based on their metabolic profile."
• "Summarise the group: who has had a chronic iron and magnesium deficit over the past month."

The answer is built on your real numbers, not on population averages.

Who can use it

Access to the MCP server is open to:

• athletes with an active PRO subscription;
• coaches with a paid coach account.

Other accounts can authenticate, but the server will refuse the request — the subscription can be upgraded in the app.

How to connect

The server URL is the same for every client:

https://mcp.trainingendurance.com/

Authorization goes over OAuth: on the first connection the Training Endurance sign-in page opens, your password stays with us, and the client receives only an access token.

Claude Desktop

1. Open Claude Desktop.
2. Click + in the chat area, then Connectors → Manage Connectors.
3. Scroll down and click Add custom connector.
4. Paste the server URL.
5. Click Add and complete the sign-in.

Claude Code

One command in the terminal:

claude mcp add trainingendurance --transport http https://mcp.trainingendurance.com/

ChatGPT

1. Settings → Apps → Advanced settings, enable Developer mode.
2. Settings → Connectors → Create.
3. Paste the server URL and complete the sign-in.

Perplexity

1. Settings → Connectors, then + Custom connector in the top-right corner.
2. Choose Remote, name it TrainingEndurance.
3. Server URL, transport Streamable HTTP, authentication OAuth.

Qwen

1. In the desktop app, click MCP in the chat window.
2. MCP Settings → Add MCP, choose Streamable HTTP.
3. Paste the server URL and complete the sign-in.

Any other client

Any MCP-capable client will do. It needs three things: Streamable HTTP transport, the server URL, and OAuth 2.1 — the client discovers the parameters itself via .well-known.

About your data

The client sees exactly what you see in the app: your workouts, your measurements, your athletes if you hold a coach account. Access is granted by a token and revoked in the same client where you added it — remove the connector and the access is gone.

Where to start

Start simple: connect a client and ask it to break down your latest workout. From there it becomes clear which questions you actually need — planning the week, reviewing a race, keeping nutrition in check, or a weekly summary across the group.