Health Guardian
ReviewAudited by ClawScan on May 10, 2026.
Overview
Health Guardian is mostly aligned with local health monitoring, but it handles very sensitive Apple Health data with scheduled automation and has unclear data-scope/privacy boundaries.
Review this skill carefully before installing. It does not show hidden network exfiltration in the supplied code, but it processes highly sensitive health data, recommends ongoing scheduled monitoring, and has unclear privacy boundaries if alerts are sent through Telegram or another external channel. Use it only with informed consent, verify the exact data source and stored files, and disable automation or external alerts unless you explicitly want them.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The skill may collect and persist highly personal health readings from a local iCloud health export location whenever it runs.
The importer reads Apple Health export data from a fixed local iCloud path and stores it in the skill data directory. Because this is sensitive health data and the source path is hard-coded, users have less clear control over exactly what data is imported.
HEALTH_EXPORT_PATH = os.path.expanduser("~/Library/Mobile Documents/iCloud~com~ifunography~HealthExport/Documents") ... SKILL_DATA_PATH = Path(__file__).parent.parent / "data"Only use this with explicit consent from the person whose health data is involved; verify the import path and stored data, and prefer a configurable, narrowly scoped source and retention policy.
A user could believe all health data stays local while anomaly details may be sent through an external messaging service if the agent follows the alert instructions.
The documentation presents a Telegram alert channel and instructs the agent to alert someone, but also claims that nothing leaves the machine. That is a misleading privacy boundary unless external alert contents are explicitly limited or disabled.
"alert_channel": "telegram" ... "If health data shows anomalies, alert human via preferred channel." ... "Privacy: Nothing leaves your machine. No cloud. No telemetry."
Clarify whether alerts leave the machine, what data they include, and require explicit user approval before sending health information to Telegram or any external channel.
Once enabled, the skill can repeatedly import and analyze health data without a manual prompt each time.
The skill recommends hourly scheduled execution and heartbeat integration. This is disclosed and central to proactive monitoring, but it means the agent may keep processing health data over time.
"schedule": { "kind": "cron", "expr": "0 * * * *" }, "payload": { "kind": "systemEvent", "text": "Run health import and check for anomalies" }Enable the cron/heartbeat workflow only if continuous monitoring is intended, and review or disable the schedule when it is no longer needed.
Users may try to run a missing script or fetch unreviewed code to fill the gap.
SKILL.md references a helper script that is not included in the supplied file manifest or source contents, indicating incomplete packaging or stale documentation.
### `scripts/summary.py` Generates human-readable health summary. ```bash python3 scripts/summary.py --period week ```
Do not download replacement helper scripts from elsewhere unless reviewed; the publisher should include the referenced file or remove the instruction.