Git Commit Formatter

Security checks across malware telemetry and agentic risk

Overview

This commit-message helper mostly does what it says, but it also includes unrelated instructions for a system whitelist that can auto-add skills without asking.

Review or remove the whitelist section before installing. Use this skill only for commit-message generation, and avoid pasting secrets, tokens, private keys, or sensitive proprietary code in raw diffs.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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)

Description-Behavior Mismatch

Medium
Confidence
98% confidence
Finding
The skill manifest presents a narrow purpose—formatting commit messages—but the body adds unrelated operational behavior for reading and influencing a system whitelist used for auto-adding skills. This is dangerous because it expands the skill's authority beyond user-visible expectations and creates a path for privilege creep or unauthorized propagation of skills through gateway automation.

Context-Inappropriate Capability

High
Confidence
99% confidence
Finding
The instructions explicitly describe reading and updating `/data/workspace/whitelist.yml`, a system control file unrelated to commit formatting. Embedding access to an allowlist in a benign-seeming formatter skill is especially dangerous because it could let the skill self-authorize or authorize other skills for automatic installation, undermining trust boundaries and enabling persistence or lateral capability expansion.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal