ClawHub
Back to skill

Security audit

Personal Health Tool Free

Security checks across malware telemetry and agentic risk

Overview

This health-management skill is mostly coherent, but it asks for broad command execution and persistent access to sensitive health data without enough scoping or privacy safeguards.

Install only if you are comfortable with an agent storing health records in plaintext local files under ~/.health and having command-execution authority. Keep sensitive reports out of shared machines, review files before adding real medical data, and prefer a version that narrows triggers and limits exec/write behavior.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
83% confidence
Finding
The skill repeatedly claims the free edition is local-only and does not upload user health data, yet it also exposes exec capability and even documents network troubleshooting via ping. That mismatch can mislead users into trusting the skill with sensitive medical information while the runtime still has the ability to invoke network-capable commands or scripts.

Context-Inappropriate Capability

High
Confidence
95% confidence
Finding
Broad exec access is unnecessary for a personal health record and report-interpretation skill whose documented functions are mostly local data parsing, storage, and summarization. If the agent follows malicious or malformed user input through this skill, exec can be abused for arbitrary command execution, data theft, persistence, or system modification.

Vague Triggers

Medium
Confidence
79% confidence
Finding
The trigger conditions are so broad that the skill may activate on generic analysis, reporting, or visualization requests outside the health domain. In a skill with write and exec permissions, overbroad routing increases the chance that unrelated prompts are handled with unnecessarily powerful tools, causing accidental file writes or command execution.

Missing User Warnings

Medium
Confidence
84% confidence
Finding
The quick-start flow immediately creates directories and writes persistent health data files without a nearby warning, consent checkpoint, or backup/privacy notice. Because the data is sensitive personal health information, silent local writes can expose users to unintended retention, disclosure on shared machines, and accidental modification of records.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.