ClawHub
Back to skill

Security audit

Git命令行助手免费版

Security checks across malware telemetry and agentic risk

Overview

This Git helper is not deceptive, but it asks agents to run repository-changing commands under broad triggers without strong confirmation controls.

Install only if you want an agent to help run Git commands. Before using it, require the agent to show status and diffs first, confirm exact files, branch, remote, and commit message before any commit or push, and avoid global credential storage unless you understand how Git will store those credentials.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (4)

Intent-Code Divergence

Medium
Confidence
89% confidence
Finding
The document claims a safety-first approach while also presenting destructive commands such as force deletion and force-push as available operations. This mismatch can cause users or agents to over-trust the skill and execute risky repository-modifying commands without sufficient caution, increasing the chance of data loss or remote history corruption.

Intent-Code Divergence

High
Confidence
98% confidence
Finding
The skill states that it avoids concatenating user input into command-line arguments, yet many examples directly interpolate branch names, file paths, commit messages, and URLs into shell commands. In a tool-enabled agent with exec access, this creates a command-injection risk if user-controlled values are not strictly validated and safely passed.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The trigger conditions are broad enough to match many generic development requests, which increases the likelihood that the skill will be invoked in contexts where users did not specifically intend Git command execution. Because the skill has exec capability and includes mutating Git operations, accidental activation can lead to unintended repository changes.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill provides examples for staging, committing, branch changes, pulling, and pushing without a mandatory confirmation step or prominent warning before repository or remote modifications. In an agent setting, this can cause irreversible or hard-to-recover changes, especially when operating on the wrong branch or remote.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.