中文 EN

扫描报告

扫描新文件

⚠ 可疑 — 风险评分 45/100
上次扫描：19 小时前 重新扫描
45 /100
Self-Audit
Audit your own tool usage. Discover which calls are necessary vs reflexive.
Skill declares a 'self-audit' CLI script and 'audit/' directory that do not exist; undeclared sensitive config.json with LLM tool permissions is present but not documented.
技能名称Self-Audit
分析耗时35.9s
引擎pi
⚠
谨慎使用
Do not use. The skill has no implementation files despite SKILL.md declaring a main CLI script. The undeclared config.json with system prompts and tool permissions should be removed or documented.

安全发现 3 项

严重性 安全发现 位置
中危
Declared entry point does not exist 文档欺骗
SKILL.md lists 'self-audit' as the 'Main CLI script' but no such file exists in the package. This is a significant doc-to-code mismatch.
- `self-audit` — Main CLI script
→ Either implement the declared script or remove the reference from documentation.
 SKILL.md:66
中危
Undeclared sensitive configuration file 文档欺骗
config.json contains LLM system prompts and tool permission mappings but is not mentioned anywhere in SKILL.md. This file includes allowed-tools definitions (Bash, Read, Write, WebFetch) which map to resource permissions.
{"role":"system","content":"You are a helpful coding assistant...","tools":{"Bash":{...},"Read":{...},"Write":{...}}}
→ Document all configuration files in SKILL.md or remove unnecessary sensitive files.
 config.json:1
低危
Declared storage directory does not exist 文档欺骗
SKILL.md mentions an 'audit/' directory for audit log storage that does not exist.
- `audit/` — Audit log storage (created on first run)
→ Either implement the directory creation or remove from documentation.
 SKILL.md:67
资源类型声明权限推断权限状态证据
文件系统 NONE NONE — No scripts exist to infer capabilities
网络访问 NONE NONE — No network code exists
命令执行 NONE NONE — No shell scripts exist
环境变量 NONE NONE — No env access code exists

目录结构

2 文件 · 2.4 KB · 88 行
Markdown 1f · 75L JSON 1f · 13L
├─ 🔑 config.json ⚠ JSON 13L · 387 B
└─ 📝 SKILL.md Markdown 75L · 2.0 KB

安全亮点

✓ No executable malicious code present in the package
✓ No credential harvesting or exfiltration mechanisms
✓ No network communication code detected
✓ No obfuscated or base64-encoded payloads