ClawHub
Back to skill

Security audit

健康数据助手免费版

Security checks for vulnerabilities and agentic risk

Overview

This appears to be a real Apple Health/Transition assistant, but it gives broad agent authority around sensitive health data without enough scoping or safeguards.

Install only if you are comfortable sending Apple Health-derived questions and results to Transition's API. Before use, limit it to explicit Apple Health tasks, avoid generic data-analysis routing, store the API key in a secret manager or locked-down environment variable, and do not cache health results locally unless you understand where that file is stored and protected.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Description-Behavior Mismatch

Medium
Confidence
86% confidence
Finding
The skill is presented as a health-data query assistant, but the documented interfaces also authorize create/modify/reset/import/export/save style operations without clearly constraining what data or state may be changed. That scope expansion increases the chance an agent will perform unintended state-changing actions or write sensitive health-related output to local files, which is risky given the broad tool permissions (`read`, `write`, `exec`).

Description-Behavior Mismatch

Medium
Confidence
78% confidence
Finding
The documentation says PMC/performance metrics are unavailable in the free edition, but later instructs users to call the PMC endpoint directly for fatigue analysis. This inconsistency can mislead an agent into invoking unsupported or more sensitive analytics endpoints than the user expects, undermining trust and safe scoping of health-data access.

Vague Triggers

High
Confidence
83% confidence
Finding
The trigger condition is far broader than the stated Apple Health query scope, telling the agent to use the skill for generic data analysis, report generation, statistical insight, and visualization. In an agent environment with `exec`, `read`, and `write`, this kind of overbroad routing can cause the skill to be selected for unrelated tasks and potentially process or transmit sensitive data outside the user’s intended health-query context.

Missing User Warnings

High
Confidence
95% confidence
Finding
The skill handles highly sensitive personal health information through an external service, but it does not prominently warn users that their Apple Health-derived data will be transmitted to `api.transition.fun`. Missing disclosure and consent around third-party transmission of health data creates significant privacy and compliance risk, especially because users may assume analysis is local or limited to the agent.

Credential Access

High
Category
Privilege Escalation
Content
export TRANSITION_API_KEY="sk_your_key_here"
# ...
# 方式二: 配置文件
cat > ~/.apple-health.env << 'EOF'
TRANSITION_API_KEY=sk_your_key_here
EOF
# ...
Confidence
81% confidence
Finding
The documentation instructs users to place the API key in a plaintext file under the home directory and source it later. While common in ad hoc setups, this creates avoidable credential exposure risk through weak file permissions, shell history, backups, or accidental inclusion in other tooling, especially in an agent ecosystem with file read/write capabilities.

Static analysis

No suspicious patterns detected.