360 Secure SkillHub
360 安全技能中心

Skillscan

gh-skillscan是OpenClaw技能安全扫描工具，需先启动Python服务，通过POST请求上传SKILL.md内容完成扫描。核心检测凭证窃取、数据泄露、混淆命令、权限越权4类威胁，输出含安全评分（1.0最安全）、威胁列表、SAFE/CAUTION/DANGEROUS结论及技能名称的JSON结果。典型场景为安装前扫描，可配合clawdhub下载后先检测，规避安全风险。

mirni
mirni
热度 42.99
最近更新 2026-05-11 17:42:41
分类：安全依赖：代码执行 / 读写文件 / 第三方 API应用场景：DevOps

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add gh-skillscan
Metadata
name
gh-skillscan
description
Scan an OpenClaw SKILL.md file for security threats before installing it. Posts the raw SKILL.md content and gets back a safety score (0-1), detected threat patterns (credential harvesting, data exfiltration, obfuscated commands, permission overreach), and a SAFE/CAUTION/DANGEROUS verdict.
metadata
{"openclaw":{"emoji":"🔍","requires":{"bins":["python"]},"install":[{"id":"pip","kind":"uv","packages":["fastapi","uvicorn","pydantic"]}]}}

SkillScan

Check if a SKILL.md is safe before you install it.

Start the server

uvicorn skillscan.app:app --port 8001

Scan a SKILL.md file

curl -s -X POST http://localhost:8001/v1/scan-skill \
  -H "Content-Type: application/json" \
  -d "{\"skill_content\": $(cat path/to/SKILL.md | jq -Rs)}" | jq

Returns safety_score (1.0 = safe, 0.0 = dangerous), findings (list of threat names), verdict (SAFE/CAUTION/DANGEROUS), and skill_name.

What it detects

  • credential_harvesting — accessing $API_KEY, $TOKEN, $SECRET, $PASSWORD
  • data_exfiltration — curl/wget sending data to external URLs
  • obfuscated_command — base64 decode piped to bash, eval, exec
  • permission_overreach — accessing /etc/shadow, .ssh/, reverse shells

Example: scan before install

clawdhub inspect some-skill > /tmp/skill.md
VERDICT=$(curl -s -X POST http://localhost:8001/v1/scan-skill \
  -H "Content-Type: application/json" \
  -d "{\"skill_content\": $(cat /tmp/skill.md | jq -Rs)}" | jq -r '.verdict')
echo "Verdict: $VERDICT"