Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 70% confidence
- Finding
- Without declared permissions the skill's intent is opaque and cannot be validated.
Patient Consent Simplifier
Security checks across malware telemetry and agentic risk
Overview
The skill is local and does not show data theft, but it overpromises compliant medical consent handling that its simple text-rewrite script cannot actually provide.
Review carefully before installing for any clinical, research, legal, or IRB workflow. Use only de-identified or properly authorized documents, choose output paths carefully, and treat results as a readability draft requiring qualified human review, not as a compliant consent document.
SkillSpector
By NVIDIA
Vulnerability Patterns
- 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
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)
Tp4
High
- Category
- MCP Tool Poisoning
- Confidence
- 94% confidence
- Finding
- This is a mismatch because the description makes a strong claim about preserving regulatory compliance and required legal content, but the code only does naive word substitution and sentence splitting. It has no compliance checks, no legal-element validation, no document-structure awareness, and no safeguards to ensure required consent language is retained. While the core idea of simplifying consent text is present, the declared purpose materially overstates what the code actually accomplishes.
VirusTotal
66/66 vendors flagged this skill as clean.