ClawHub
Back to skill

Security audit

unisound-prescription-assist

Security checks across malware telemetry and agentic risk

Overview

This medical assistant has a real privacy concern because it sends full patient-related input to a remote model while promising de-identification that the code does not perform.

Use caution before installing this in any clinical or regulated environment. Do not provide identifiable patient information unless the remote service, privacy basis, consent, retention, and governance controls have been approved; the current script does not enforce the de-identification promised by the documentation.

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 (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding
The skill documentation describes executable behavior that reads local files, optionally writes output files, and makes outbound network requests, yet it declares no permissions. This is a real security issue because operators and policy engines cannot accurately understand or constrain what the skill can access, and the skill processes potentially sensitive medical data that could be sent to a remote API.

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The script sends raw prescription and patient information to a remote HTTPS endpoint, while the skill description only says it uses an 'internal medical model' and does not clearly disclose remote transmission of sensitive medical data. In a healthcare context, this creates a real privacy and data-governance risk because diagnosis, renal function, allergy history, and co-medications are highly sensitive and may be regulated data.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
At the call site, the full prescription_info content is sent to the LLM without any user-facing warning or confirmation, even though the input may contain personal health information. For a prescription-assistance skill, this is especially dangerous because the data can include diagnoses, age, kidney function, allergies, and concurrent medications, making accidental privacy violations more severe.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.