What is MCP?

MCP, the Model Context Protocol, is an open standard for letting an AI assistant read data from somewhere else. It is the plumbing behind “connect your account” in tools like Claude and ChatGPT.

The problem it solves

An AI assistant only knows what is in the conversation. Ask it about your training and it can talk about running in general, but it has never seen your runs. You can paste data in, which is tedious and immediately goes stale.

MCP gives the assistant a defined way to ask an outside service for that data itself. Before it existed, every assistant needed a bespoke integration with every service. MCP is the shared standard that replaced that, which is why the same Stride connection works in more than one assistant.

What actually happens when you approve

Approving a connection does not hand over your password, and it does not copy your data to the assistant in bulk. It issues a token, a revocable key scoped to one assistant.

After that, nothing moves until you ask a question. When you ask about last week’s mileage, the assistant calls Stride, Stride checks the token, and it returns just that. Data you never ask about is never sent.

The Stride connector is read-only. It can view your Strava activities, athlete profile, and training history, plus your WHOOP recovery and sleep data if you have connected WHOOP. It cannot create, change, or delete anything, and it cannot read your saved API keys.

Staying in control


For what Stride can do once connected, see Stride MCP. For exactly what data is handled and who processes it, see the connector privacy policy.

What is MCP? | Stride