Back to skill

Security audit

Git辅助

Security checks for vulnerabilities and agentic risk

Overview

This Git helper is not clearly malicious, but its requested command/write authority is under-scoped and its documentation mixes Git operations with unrelated operations and code-review behavior.

Review this skill before installing. It appears to be a Markdown-only Git assistant, not malware, but it asks for command and write capability and its documentation is broad and inconsistent. Use it only with a clearly limited repository scope, and require explicit confirmation before push, pull, branch changes, checkout/reset-like operations, or any command that can modify local or remote state.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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
Findings (5)

Vague Triggers

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill advertises a broad invocation scope including system monitoring, log analysis, alerting, and deployment management despite being presented as a Git helper with exec/write capabilities. This scope inflation can cause an agent to invoke the skill in higher-risk operational contexts than intended, increasing the chance of unsafe command execution or repository-modifying actions under irrelevant prompts.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The description explicitly states '支持中文交互' as part of the skill behavior, but does not offer a user choice of language or explain a justified locale restriction. Under the policy, forcing a specific language without opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill supports impactful Git actions such as pull, push, branch manipulation, write, and exec, but the documentation lacks a prominent warning that these operations can change repository state or affect remotes. In an agent setting, missing user-facing risk disclosure increases the chance of accidental destructive or unauthorized changes, especially if the skill is invoked autonomously.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The surrounding documentation repeatedly presents the skill as a Git helper for operations like status, pull, push, branch, and log, but the output example returns an 'overall_grade', 'total_score', and review items such as '代码风格' and '安全合规'. This is not a minor omission; it actively describes a different kind of code-analysis skill than the Git-operation skill named elsewhere in the file.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The FAQ answer for using Git Helper states that the system analyzes code and produces a review report, while other parts of the document describe the skill as supporting Git commands like status, pull, push, branch, and log. This is a direct intent/documentation contradiction rather than merely incomplete documentation.

Content

No source excerpt is available for this finding.