Security audit
Security checks for vulnerabilities and agentic risk
This Git helper is mostly a markdown guide, but it gives broad execution/write authority and includes destructive Git guidance plus unclear API and scope claims.
Review this skill before installing. Use it only for Git tasks, require explicit confirmation before any command that changes files or history, make a backup branch or tag first, and do not provide API keys or external-service credentials unless the publisher clearly documents why they are necessary.
This section includes destructive Git operations such as hard resets, clean, filter-branch, and aggressive garbage collection without prominent warnings about irreversibility or repository-wide impact. In a skill with exec capability, such guidance materially increases the risk of data loss, branch corruption, and accidental destruction of working trees or history.
The documented use of git reset --hard HEAD~1 is a destructive command that discards local changes and rewrites local history. In the context of a skill exposing exec and positioning itself as directly usable, this is dangerous because users may copy, approve, or execute it without understanding that uncommitted work and recent commits can be lost permanently.
git reset --soft HEAD~1
git reset --hard HEAD~1
git commit --amend -m "新信息"
The use of git reset --hard COMMIT_ID during recovery guidance can permanently discard the current working state and move the branch pointer abruptly. In a Git helper skill, this is especially risky because it is presented as a recovery action, which may lead users to trust it as safe even though it can destroy unpushed work and complicate team collaboration.
git checkout COMMIT_ID -- file.txt
# ...
git reflog # 查找提交
git reset --hard COMMIT_ID # 恢复
# ...
git reflog # 找到分支最后的提交
git checkout -b branch_name COMMIT_ID
The description says to use the skill for '代码生成、编程辅助、调试测试、开发部署', which is far broader than a Git workflow helper and does not clearly distinguish when this skill should be invoked instead of many other generic coding skills. This creates an ambiguous trigger scope likely to collide with everyday development requests.
The description states '支持中文交互' and the skill branding is explicitly Chinese-focused, but it does not clearly present this as an opt-in language choice or a justified locale-restricted tool. This can violate language/locale policy when a skill effectively forces a specific language without user selection.
‘专业版增强能力’列出了代码静态分析、依赖漏洞检测、批量代码审查、CI/CD 流水线集成等能力，这些明显超出‘分支管理、冲突解决、提交规范’的 Git 工作流范围。该扩展改变了技能的实际能力边界，属于 manifest 描述与实际文档行为范围不一致。
Manifest 和前文主要能力聚焦于分支管理、冲突解决、提交规范，但输出说明定义的是 overall_grade、代码风格、安全合规、评分和改进建议等通用代码审查结果，而非 Git 工作流操作结果。这表明技能文档宣称的核心用途与实际呈现的行为模式不一致，用户可能预期执行 Git 工作流辅助，却得到代码审查式输出。
Git 分支管理、冲突解决和提交规范通常可在本地仓库和 Git CLI 范围内完成，不需要外部 API 凭证。文档却要求配置 API Key，并宣称可通过标准化接口调用外部服务，这为技能引入了与其 stated purpose 不直接相关的外部集成能力。
The skill presents conservative Git safety principles, yet elsewhere includes history-rewriting and cleanup commands that can permanently delete data. This contradiction makes unsafe operations more likely because users may trust the skill as safety-oriented while receiving destructive guidance without adequate guardrails.
安全章节和错误处理章节多次讨论 HTTPS、SSL 证书、401/429/504 接口错误、DNS 与防火墙等网络调用场景，说明该技能被设计为具备网络访问能力。对于一个以 Git 工作流、分支和提交规范为核心的技能，这类网络服务能力并非显而易见的必要实现细节。
The example '帮我解决 Git 合并冲突' is understandable, but the document does not provide any explicit trigger list, scope limits, or negative examples anywhere nearby. Without clearer activation boundaries, the skill may be invoked for loosely related requests beyond its intended Git workflow domain.
No suspicious patterns detected.