ClawHub
Back to skill

Security audit

Dev Git Guard

Security checks across malware telemetry and agentic risk

Overview

This skill is a transparent git workflow helper, but it will inspect worktrees, run validation commands, and create local commits automatically when used.

Install only if you want Codex to enforce this git workflow, including reviewing dirty files, running validation commands, staging changes, and creating detailed Chinese commit messages. Avoid it in repositories where commits require manual approval, English-only metadata, or stricter team governance.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (4)

Natural-Language Policy Violations

High
Confidence
90% confidence
Finding
The skill description hard-codes a requirement for detailed Chinese git commits before and after work, removing user choice over commit language. In enterprise or regulated environments, forcing a specific language can reduce auditability, hinder code review and incident response, and cause policy noncompliance if teams require English or locale-specific standards.

Natural-Language Policy Violations

High
Confidence
96% confidence
Finding
The commit policy explicitly forces every commit message to be written in Chinese, regardless of repository norms or user requirements. This can obstruct peer review, automated governance processes, and forensic analysis by making commit metadata less accessible to maintainers and compliance personnel who do not read Chinese.

Vague Triggers

Medium
Confidence
94% confidence
Finding
The default prompt and implicit invocation policy make this skill eligible to activate across a wide range of repository tasks without tight preconditions. That can cause the agent to impose workflow steps such as inspecting dirty files and creating commits in contexts where the user did not explicitly request git operations, increasing the chance of unintended repository state changes or disclosure of local worktree details.

Natural-Language Policy Violations

Medium
Confidence
91% confidence
Finding
Forcing Chinese commit messages without user opt-in overrides user and project conventions and can lead to unauthorized or policy-noncompliant commit metadata. In regulated, collaborative, or automated environments, this may disrupt review workflows, violate contribution rules, or cause unintended disclosure through generated commit text.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.