Fetch raw JSON for ad creatives, apps, rankings, market data, and download or revenue estimates.
4.3k installs296 stars
Data & analysis
Skill Security Audit
Audit installed skills for risky behavior and produce evidence, ratings, verdicts, and remediation text.
What it does
Audit installed agent skills from their SKILL.md, scripts, references, and related files. Review command execution, network and file access, data exfiltration, dependencies, prompt-boundary bypasses, and broad triggers. Produce evidence-based findings, severity ratings, verdicts, remediation text, and a portfolio summary with priorities.
When to use it
- Pre-installation review of downloaded skills
- Portfolio-wide audit of installed skills
- Investigation of broad or unsafe triggers
- Prioritization of security remediations
The skill document
Skill Security Audit 🔐
对已安装的 Skills 进行安全审计，识别风险行为并提供修复建议。
审计类别
| # | 类别 | 说明 |
|---|---|---|
| 1 | 命令执行 | 检查不安全的 shell/python/node 执行 |
| 2 | 网络访问 | 检查未经授权的网络请求 |
| 3 | 文件访问 | 检查过度文件系统访问 |
| 4 | 数据泄露 | 检查未授权数据外传 |
| 5 | 依赖风险 | 检查不安全依赖 |
| 6 | 提示词越权 | 检查绕过安全边界 |
| 7 | 触发条件 | 检查描述是否过宽 |
严重程度
| 等级 | 说明 |
|---|---|
| Critical | 明显允许危险操作 |
| High | 重大滥用风险 |
| Medium | 潜在滥用风险 |
| Low | 小问题 |
| Info | 设计选择 |
使用方法
1. 准备审计材料
用户提供：
- 已安装 Skills 列表
- 每个 Skill 的文件内容（SKILL.md、scripts、references 等）
2. 执行审计
按照审查矩阵检查每个 Skill：
- 盘点文件
- 分类能力
- 检查风险
- 收集证据
- 评级
- 修复
3. 输出报告
每个 Skill 的报告结构：
# [skill name]
## verdict
- overall rating: [block/review/acceptable]
- top risks: [风险列表]
## findings
- category:
- severity:
- evidence:
- impact:
- remediation:
## replacement text
修复建议文本
审查矩阵
1. 任意命令执行
- ⚠️ 高风险：允许任意 bash/sh/powershell/python/node 执行
- ✅ 修复：限制为固定命令列表
2. 外部网络访问
- ⚠️ 高风险：访问任意 URL
- ✅ 修复：限制为白名单域名
3. 本地文件访问
- ⚠️ 高风险：读取整个目录/主目录
- ✅ 修复：限制为用户提供的文件
4. 数据泄露
- ⚠️ 高风险：未经确认发送/上传数据
- ✅ 修复：需要明确用户确认
5. 依赖风险
- ⚠️ 高风险：未固定版本的可疑依赖
- ✅ 修复：固定版本，使用标准库
6. 提示词越权
- ⚠️ 高风险：忽略系统/策略约束
- ✅ 修复：重述系统规则优先
7. 触发条件过宽
- ⚠️ 高风险：描述过于宽泛
- ✅ 修复：精确触发场景
输出要求
最终报告包含：
- 每个 Skill 的详细报告
- 总体摘要
- 最高风险项
- 修复优先级
示例
输入
Skills: [peekaboo, admapix, humanizer]
文件内容: SKILL.md, scripts/, ...
输出
# Portfolio Summary
- audited skills: 3
- block: 0
- review before use: 1
- acceptable: 2
- most common risk patterns: [列表]
- immediate remediation priorities: [优先级]
Questions people ask
- What materials are required for an audit?
- Provide the installed skills list and each skill’s files, including SKILL.md, scripts, references, and other relevant contents.
- How are risks classified and rated?
- Findings are grouped into seven categories: command execution, network access, file access, data exfiltration, dependency risk, prompt privilege escalation, and trigger scope. Severity uses Critical, High, Medium, Low, or Info.
- What does the audit report include?
- Each skill receives an overall block, review, or acceptable verdict, plus risks, evidence, impact, remediation, and replacement text. The final report also includes a portfolio summary, highest-risk items, and remediation priorities.
Related skills
Create or revise Draw.io, Mermaid, and Excalidraw diagrams from natural-language requirements.
1.0k installs47 stars
Search, read, query, and—with explicit approval—modify content in connected Notion workspaces.
339 installs8 stars
Read and manage Google Merchant Center resources through v1 APIs with managed OAuth and approval-gated writes.
231 installs5 stars
Convert Chinese finance questions into checked Tushare research workflows, summaries, and data files.
559 installs33 stars
Access supported Square resources through managed OAuth for read and approved administrative operations.
229 installs3 stars