Learn
How to connect Strava to Claude with KBrain MCP
A practical guide to using KBrain as a remote MCP connector between your Strava activity data and Claude for AI powered training analysis.
Analyze your Strava training data with KBrain
Subscribe and add StravaClaude supports remote MCP connectors. KBrain uses that capability to bring your authorized Strava data directly into Claude, so you can ask training questions without exporting a single file.
How the connection works
KBrain acts as the structured data layer between Strava and Claude. You authorize Strava inside KBrain. KBrain exposes your activity data through an MCP endpoint. Claude connects to that endpoint as a remote server and can call it during any conversation.
KBrain handles the Strava authorization, data normalization, derived metrics, and MCP tool surface. Claude handles the reasoning and conversation. You handle the questions.
What you need to get started
- A KBrain subscription with Strava added as a private data source
- Your KBrain MCP endpoint URL
- Claude access with connector settings enabled
- A completed OAuth authorization when prompted
What you can ask Claude once connected
- Which runs showed aerobic decoupling this month?
- How did my heart rate zone distribution change over the past six weeks?
- Where on my usual route does heart rate consistently spike?
- What does my training stress balance look like right now?
- Compare my performance on this route across the last three months
Keep it read only
The Strava connector should stay read only. Training analysis needs retrieval and calculation, not activity modification. KBrain is designed around this boundary: access to your data for analysis, with no write actions back to Strava.
One brain, multiple assistants
Once your Strava brain is set up in KBrain, the same MCP endpoint can support Claude, ChatGPT, and other MCP compatible clients. You configure the data source once. It travels with you across assistants.
Analyze your Strava training data with KBrain
Subscribe to KBrain, add Strava as a private data source, and connect the same training brain to Claude, ChatGPT, or any MCP compatible assistant.
Subscribe and add StravaFrequently asked questions
Why use MCP instead of uploading Strava exports?
MCP lets Claude call a live, structured tool layer instead of relying on one-off pasted files or stale exports. The context is always current and structured for analysis.
Does Claude host the Strava integration?
No. Claude connects to KBrain as a remote MCP server, while KBrain handles the authorized Strava data layer.
Should the connector be read only?
Yes. Training analysis needs retrieval and calculation, not activity modification. KBrain keeps the Strava connection read only by design.
Can I use the same brain with other assistants?
Yes. The same KBrain MCP endpoint works with any MCP compatible client, including ChatGPT and other assistants that support remote server connections.