healthsync

Your Apple Health data, locally stored and queryable by AI agents. Parse exports into SQLite. Query from the CLI, or let your coding agent ask the questions.

Get Started GitHub

Heart Rate Steps Sleep Blood Oxygen VO2 Max Workouts

zsh — 80×24

$ curl -fsSL https://healthsync.sidv.dev/install | bash
✓ healthsync v0.3.0 installed to /usr/local/bin/healthsync

$ healthsync parse export.zip
  ⣾ 540,110 records · 1,011 workouts · 18,432/s · 30s
  Records:  540,110
  Workouts: 1,011
  Duration: 30.621s

$ healthsync query heart-rate --limit 3
┌──────────────────────────┬───────┬──────────────────────────────┐
│ SOURCE_NAME              │ VALUE │ START_DATE                   │
├──────────────────────────┼───────┼──────────────────────────────┤
│ Siddhartha's Apple Watch │ 72    │ 2026-02-12 19:11:31 +0530   │
│ Siddhartha's Apple Watch │ 68    │ 2026-02-12 19:06:22 +0530   │
│ Siddhartha's Apple Watch │ 75    │ 2026-02-12 18:58:14 +0530   │
└──────────────────────────┴───────┴──────────────────────────────┘

$ healthsync skills install # teach your AI agent to query health data
✓ Installed to ~/.claude/skills/healthsync/
  Files: healthsync.md, .healthsync-version

Built for AI agents

Run healthsync skills install to give your AI coding agent the schema, CLI reference, and SQL examples it needs to answer questions about your health data directly in conversation.

"What was my average resting heart rate last month?"
"How many hours did I sleep on average this week?"
"Show me my VO2 Max trend over the past year."
"Which workout type burns the most calories for me?"

The skill installs to ~/.claude/skills/healthsync/ and is picked up automatically by Claude Code and compatible agents on next session start.

What it does

Streaming parser Handles 950MB+ exports at constant ~10MB memory. 500k records in ~30 seconds.
Local SQLite Stored at ~/.healthsync/healthsync.db, WAL mode, auto-deduplication.
40+ metrics Cardiac, activity, body, sleep, mobility, running, blood pressure, workouts, and more.
Flexible output Table, JSON, or CSV. Date range filters with --from / --to.
HTTP server Async upload via POST /api/upload — 202 Accepted with progress polling. Works over Tailscale.
Zero dependencies Single binary, no CGO, pure Go SQLite via modernc.org/sqlite.

Install

macOS / Linux

curl -fsSL https://healthsync.sidv.dev/install | bash

go install github.com/BRO3886/healthsync@latest

Manual

tar -xzf healthsync-darwin-arm64.tar.gz
sudo mv healthsync /usr/local/bin/

Windows: download healthsync-windows-amd64.zip or healthsync-windows-arm64.zip from GitHub Releases and add the binary to your PATH.

Quick start

Export from iPhone
Health app → Profile → Export All Health Data → save the .zip
Parse
```
healthsync parse export.zip -v
```

Query

healthsync query heart-rate --limit 10
healthsync query steps --from 2024-01-01 --format json

Install agent skill
```
healthsync skills install
```
Your AI agent can now query your health data directly in conversation.

How it works


Parser	Streaming XML with DTD stripping via `io.Pipe`. Handles 500k+ records in ~30s.
Storage	Pure Go SQLite. WAL mode, batch inserts (1000/tx), `UNIQUE` constraints for dedup.
Server	Chi router. `POST /api/upload` returns `202`, parses async, poll `GET /api/upload/status`.
CLI	Cobra. `parse`, `query`, `server`, `skills` subcommands. `--db`, `--format`, `--from/--to` flags.
Skills	`go:embed`. Writes schema + query docs to `~/.claude/skills/` for agent pickup.