Secure Code Review

Security checks across malware telemetry and agentic risk

Overview

This is a markdown-only secure code review guide that asks the agent to read and search a repository, with one confirmed documentation bug in an example but no hidden install, persistence, credential use, or data export behavior.

Reasonable to install for repository security review if you are comfortable letting the agent read and search the target codebase. Keep shell and write access under normal user control, and treat the early-exit authorization example around unauthorized and encoding errors as flawed until the publisher corrects it.

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: 98% confidence
Finding: The example labeled as the safe refactor still maps the error conditions to the wrong exception types: the unauthorized case raises a wrong-encoding error, and the encoding failure raises an unauthorized error. In a security-review skill, incorrect "safe" examples are dangerous because users may copy them directly, leading to misclassified auth failures, bypass-like logic confusion, and weakened incident detection or auditing.

VirusTotal

58/58 vendors flagged this skill as clean.

View on VirusTotal