ClawHub
Back to skill

Security audit

med-critical-disease-review

Security checks across malware telemetry and agentic risk

Overview

The skill is purpose-aligned, but it sends sensitive medical records to an API and writes outputs locally despite privacy text claiming de-identification and no local persistence.

Install only if you are authorized to process these medical records with the configured assessment service. Treat inputs, API responses, text summaries, and prepared payload files as sensitive health data; avoid --save-prepared unless needed, set output paths to a protected location, and verify the API destination before use.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding
The skill documentation describes capabilities to read files, write outputs, and make network requests to an internal medical assessment API, but no explicit permissions are declared. This creates a transparency and policy-enforcement gap: operators may approve or run the skill without understanding that it can exfiltrate sensitive medical data over the network and persist results locally.

Intent-Code Divergence

Medium
Confidence
98% confidence
Finding
The document promises that user input and intermediate results are not written to local persistent storage and are destroyed after the call, but later states that raw API responses, natural-language conclusions, and prepared medicalRecord data may be saved to local paths. In a medical-claims context, this contradiction is dangerous because it can lead to retention of highly sensitive health information and create compliance, privacy, and unauthorized-access risks.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The script sends highly sensitive medical record data to a network service via call_major_disease_assess using a user-supplied API key and base URL, but provides no explicit warning, consent gate, minimization, or destination restriction. In a medical insurance review context, transmitting hospitalization records and diagnoses off-host can create serious privacy, compliance, and data-handling risk, especially if the base URL is changed to an unintended or less trusted endpoint.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
When --save-prepared is used, the normalized medicalRecord payload is written to disk in plaintext JSON, which may include diagnoses, admission identifiers, hospital information, and full medical text. Persisting this data locally without a strong warning, secure file-permission handling, or retention controls increases the chance of unauthorized disclosure through shared machines, backups, logs, or later exfiltration.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.