Git命令行

• Outcome: Review
• Latest audit: Aug 10, 2026 · 10:51 AM UTC
• Version: 1.0.0

Overview

This Git helper is mostly coherent, but it under-scopes high-impact Git operations that can delete local work or rewrite remote history.

Review this skill before installing. Use it only for explicit Git tasks, and require manual confirmation before branch deletion, reset, rebase, push, tag push, or any operation touching shared branches. Do not provide API keys unless the publisher clearly documents why they are needed and what repository data is sent externally.

SkillSpector

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Tool Misuse: Tool Parameter Abuse, Chaining Abuse, Unsafe Defaults
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (9)

Description-Behavior Mismatch

Medium

• Confidence: 90% confidence
• Finding: The skill is presented as a Git CLI helper, but the documentation expands its scope to unrelated capabilities such as static analysis, vulnerability detection, CI/CD integration, and external service integration. This scope inflation can cause an agent or user to over-trust the skill with tasks and permissions beyond its declared purpose, increasing the chance of unsafe tool use or unintended data exposure.

Context-Inappropriate Capability

Medium

• Confidence: 92% confidence
• Finding: The skill declares API key configuration and external API/LLM integration even though its stated purpose is local Git CLI assistance. This introduces unnecessary credential-handling expectations and may prompt users or agents to expose secrets or route repository data to external services without a justified need.

Intent-Code Divergence

Medium

• Confidence: 88% confidence
• Finding: The document claims that only whitelisted commands should be executed, but it does not define or enforce any whitelist while simultaneously advertising broad Git command execution, including destructive operations. This mismatch creates a false sense of safety and increases the risk that an agent will execute unsafe commands under the skill's authority.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The activation description is overly broad and matches general development, debugging, deployment, and programming-assistance scenarios rather than a narrowly scoped Git helper use case. Over-broad triggering can cause the skill to be invoked in contexts where it has excessive access to files or command execution, increasing misuse risk.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill documents destructive branch deletion operations, including force deletion of unmerged branches, without prominent warnings about irreversible data loss or recommended safety checks. In an agent-executed context, even standard Git commands become riskier because they may be run automatically or with incomplete user understanding.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The documentation recommends `git reset --hard HEAD~1` without a strong safety warning, despite the command irreversibly discarding local changes and rewriting working state. In an agent skill with execution capability, this is especially dangerous because a user may interpret it as routine remediation and lose code or other uncommitted work immediately.

Tool Parameter Abuse

High

• Category: Tool Misuse

Content

git fetch origin
git fetch --all --prune  # 获取所有远程并清理已删除的分支
# 推送
git push origin feature/branch
git push -u origin feature/branch  # 设置上游跟踪
git push --force-with-lease origin main  # 安全强制推送（推荐）
# 注意：避免使用 git push --force，它会覆盖远程历史

• Confidence: 95% confidence
• Finding: The skill includes remote mutation commands such as push and `push --force-with-lease` to main, which can rewrite shared history and alter remote repositories. In a tool-executing agent context, this is dangerous because parameterized branch names and remotes may be influenced by user input, leading to unauthorized publication, history tampering, or disruption of team workflows.

Tool Parameter Abuse

High

• Category: Tool Misuse

Content

| 完成合并 | git commit |

### Q5: 如何在Git CLI中回退到上一个提交？
A: 使用 `git reset --hard HEAD~1` 命令可以回退到上一个提交。

| 命令 | 操作 |
|:-----|:-----|

• Confidence: 98% confidence
• Finding: `git reset --hard` is a high-risk state-destroying command that can discard working tree and index changes without recovery in common workflows. In an agent skill capable of execution, exposing this as a normal operation substantially raises the risk of accidental or induced destructive action.

Tool Parameter Abuse

High

• Category: Tool Misuse

Content

| 命令 | 操作 |
|:-----|:-----|
| git reset --hard HEAD~1 | 回退到上一个提交，丢失当前提交后的所有更改 |
## 主要特点
- **自动化执行**: 用Git CLI检查/暂存/提交/分支/同步代码变更。Helper for using the Git CLI to i
- **文件处理**: 支持多种文件格式的读取、解析和写入操作

• Confidence: 98% confidence
• Finding: The tabular example normalizes `git reset --hard HEAD~1` as a simple way to go back one commit while only briefly noting data loss. This framing understates the operational risk, and in an agent-driven environment it can facilitate accidental deletion of local work or misuse in the wrong repository context.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.