Phy Deserialization Audit

Security checks across malware telemetry and agentic risk

Overview

This is a local source-code scanning skill with no evident network, credential, persistence, or destructive behavior, but the embedded scanner currently appears to fail at startup if copied exactly.

Install only if you are comfortable with the agent reading the project directory you ask it to scan. Before relying on results, fix or verify the embedded Python script because the current snippet appears to fail immediately due to the _build_js_patterns definition order.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Intent-Code Divergence

Medium

Confidence: 93% confidence
Finding: The Python snippet initializes PATTERNS using _build_js_patterns() before that helper is defined, which will raise a NameError at import/runtime and break the scanner. In a security tool, a fail-to-run condition is dangerous because users may assume the audit completed successfully and miss real vulnerabilities.

VirusTotal

59/59 vendors flagged this skill as clean.

View on VirusTotal