Security audit

Medical Ip Director

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only medical marketing skill with disclosed but privacy-sensitive photo and content-matrix workflows, and no executable or hidden behavior.

Install only if you are comfortable using it for medical-aesthetic marketing content. Do not put patient-identifying details, raw consultation notes, or facial photos into saved matrices, and make sure any photo collection uses explicit consent, secure intake, retention limits, deletion procedures, and human legal/medical review before publication.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (8)

Intent-Code Divergence

High

Confidence: 97% confidence
Finding: The skill explicitly promises that user photos and identifiable health data will not be stored, while other sections require maintaining a cross-conversation content matrix and forced archival. In a medical context, this creates a direct policy-to-behavior contradiction that can mislead users into sharing sensitive images under false assumptions about deletion and retention.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill encourages users to transmit photos and consultation details for medical-style assessment, but it does not provide a meaningful warning about the risks of sending sensitive medical imagery over chat systems. In context, face photos and health-related details are highly sensitive personal data, so insufficient risk disclosure can lead to privacy harm, unintended retention, or secondary use.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The workflow directs automatic creation or updating of a Markdown knowledge base but does not clearly warn users that submitted content may be written to persistent files. This is especially risky because the skill also solicits medical and image-based inputs, creating a realistic chance that sensitive source material or derived details could be retained beyond the user's expectation.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The script explicitly asks users to send original front and side facial photos for individualized assessment, which involves highly identifiable biometric/health-adjacent personal data. There is no accompanying notice about consent, secure transmission, retention, access controls, or how the images will be used, creating a real privacy and data-protection risk if users comply.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The script explicitly instructs users to send front and side facial photos for assessment, which are sensitive biometric/personal data, but it provides no privacy notice, consent language, minimization guidance, or secure submission channel. In a medical/cosmetic context, this is more dangerous because the images may reveal identity and health-related information, increasing privacy and compliance risk.

Ssd 3

Medium

Confidence: 92% confidence
Finding: Maintaining a persistent content matrix across conversations encourages retention of prior user-provided material and is inconsistent with the claimed temporary-use privacy model. In this medical marketing context, even derived notes can encode sensitive personal or health-related information, increasing privacy and compliance risk.

Ssd 3

Medium

Confidence: 95% confidence
Finding: A forced archiving step semantically instructs the system to store processed source material into a knowledge base. Because the skill operates on doctor interviews, consultation context, and user-submitted assessment material, this creates a realistic path for sensitive or proprietary content to be persisted without sufficient consent or minimization.

Ssd 3

Medium

Confidence: 95% confidence
Finding: The `/save_matrix` trigger explicitly forces saving to persistent storage, enabling retention of prior user-supplied content and derived material. In context, the skill handles potentially sensitive medical-adjacent inputs, so a direct save command without strong data classification and consent controls increases the chance of inappropriate retention.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal