Use a local MCP server to review health trends

Amandeep Khurana linked several health trackers to Claude Desktop so it could inspect half a year of labs, sleep, CGM, and supplement data after a back injury.

Done with Claudeยท Advancedยท Nov 2026

What they did
Khurana asked Claude Code to create a TypeScript MCP server for his Wellavy tracker, and it was done in roughly two hours. He registered that server in Claude Desktop, then added separate MCP servers for Oura, Whoop, and Libre. After the injury, he posed a single question and let Claude gather the bloodwork, sleep, glucose, and supplement history itself.
How it went
Claude took about 3 to 5 minutes to highlight low ferritin, disrupted deep sleep, and other shifts, then ordered the likely risks and suggested actions. His doctors said the readout was on target and generally lined up with their own plan.
Tips
A remote OAuth setup failed quietly in Claude Desktop, so he switched to API keys in the URL even though that sends data through Railway. He also reduced the tool count from more than 20 to 14 before running the analysis.

Try it yourself with Claude

I want you to review my [blood tests, sleep data, CGM readings, supplement log] from the last [time span] after [injury or health event]. Find patterns, compare any values that changed near [date], and explain the findings in clear language. Add a table of the main biomarkers and end with questions I should raise with my doctor. Do not diagnose me or suggest treatment. Before any irreversible action such as sending, submitting, deleting, or paying, stop and ask me to confirm.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on akhurana.substack.com โ†—

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.