Skip to content
MCP server

Quantified Self MCP

Local MCP server that lets AI assistants analyse your own health data with caveats

Runs locallyStatus and history
Install
Platforms
CLI, self-hosted, macOS
Runs
Locally
Version
0.8.1
Listing updated
today

Similar MCP servers

About Quantified Self MCP

Quantified Self MCP is an MCP server for health and science. It runs locally on your machine and is started by your AI client.

Quantified Self MCP keeps a SQLite database called health.db on your own machine and answers questions about it through an MCP client. Its distinguishing idea is a claim-evidence step: when an assistant reports a trend, the result carries how many days were logged, the largest gap and how fresh the data is, then labels the claim as insufficient, suggestive, detectable but not meaningful, or supported.

You can try it with 60 days of synthetic data using a demo flag, or import exports from Apple Health, WHOOP, Oura, Android Health Connect or CSV files. A HEALTH_PRIVATE_FIELDS setting hides chosen metrics such as weight or mood from every AI query while keeping them stored.

The README describes 20 tools covering raw measurements, workouts, longitudinal statistics and data health, plus a doctor command that checks the install. It installs with pip or runs through uvx over stdio.

Install

Run in a terminal

claude mcp add quantified-self-mcp -- uvx quantified-self-mcp

Add to claude_desktop_config.json

{
  "mcpServers": {
    "quantified-self-mcp": {
      "command": "uvx",
      "args": [
        "quantified-self-mcp"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "quantified-self-mcp": {
      "command": "uvx",
      "args": [
        "quantified-self-mcp"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "quantified-self-mcp": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "quantified-self-mcp"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "quantified-self-mcp": {
      "command": "uvx",
      "args": [
        "quantified-self-mcp"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "quantified-self-mcp": {
      "command": "uvx",
      "args": [
        "quantified-self-mcp"
      ]
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

Tools this server provides (3)

Taken from the project documentation; the running server may differ.

calculate_metric_trend
Computes a metric trend and returns a claim tier with caveats.
explain_metric_change
Explains a change in a metric with supporting evidence.
detect_metric_anomalies
Flags unusual readings with evidence and caveats.

What it does

Evidence-tiered claims
Every analysis returns coverage, sample profile and a claim tier with required caveats.
Local SQLite storage
Health history stays in a database on your machine.
Field-level privacy
Chosen metrics can be blocked from all AI queries and analytics.
Wearable imports
Reads Apple Health, WHOOP, Oura, Health Connect and CSV exports.
Provenance tracking
Results are stamped with the source device or app they came from.
Snapshots and doctor check
WAL-safe backup and restore, and a diagnostic command for setup.

Good for

  • Ask how sleep and resting heart rate changed over 30 days
  • Find out why HRV dropped last week and how reliable the answer is
  • Look for links between workout minutes and sleep quality
  • Keep health analysis offline with a local model

Quantified Self MCP FAQ

Where is my data stored?

In a local SQLite file named health.db on your own machine.

How do I try it without my own data?

Run the init command with the demo flag to seed 60 days of synthetic data.

Which wearable exports can it import?

Apple Health XML, WHOOP and Oura JSON, Android Health Connect zips, and CSV.

Can I stop an AI from seeing certain metrics?

Yes. Set HEALTH_PRIVATE_FIELDS, for example to weight_kg,mood.

How do I add Quantified Self MCP to Claude Code?

Run this command in your terminal: claude mcp add quantified-self-mcp -- uvx quantified-self-mcp

What are the best alternatives to Quantified Self MCP?

The closest alternatives to Quantified Self MCP are Agent-Native Analytics, Google Ads + Meta Ads + SEO/GEO MCP, dashboard-icons, Tooluniverse and Oya Browser.

Where this comes from

  • Description: written from the project's README.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers