clawsec-clawhub-checker by prompt-security/clawsec
npx skills add https://github.com/prompt-security/clawsec --skill clawsec-clawhub-checker增强 ClawSec 套件的受保护技能安装器,增加 ClawHub 信誉检查功能。在允许安装技能前,通过检查 VirusTotal 代码洞察分数和其他信誉信号,增加第二层安全防护。
clawhub install - 拦截技能安装请求此技能必须在 clawsec-suite 之后 安装:
# 首先安装套件
npx clawhub@latest install clawsec-suite
# 然后安装检查器
npx clawhub@latest install clawsec-clawhub-checker
# 运行设置脚本以集成到 clawsec-suite
node ~/.openclaw/skills/clawsec-clawhub-checker/scripts/setup_reputation_hook.mjs
# 重启 OpenClaw 网关使更改生效
openclaw gateway restart
设置完成后,检查器会在 clawsec-suite/scripts 目录下添加 和 ,并更新咨询守护钩子。原始的 不会被替换。
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
enhanced_guarded_install.mjsguarded_skill_install_wrapper.mjsguarded_skill_install.mjs设置完成后,运行包装器(直接替换路径)或直接运行增强脚本:
# 推荐使用直接替换的包装器
node scripts/guarded_skill_install_wrapper.mjs --skill some-skill --version 1.0.0
# 或者直接调用增强脚本
node scripts/enhanced_guarded_install.mjs --skill some-skill --version 1.0.0
增强流程:
--confirm-reputation0 - 可以安全安装(无咨询匹配,信誉良好)42 - 找到咨询匹配(现有行为)43 - 信誉警告(新增 - 需要 --confirm-reputation)1 - 错误环境变量:
CLAWHUB_REPUTATION_THRESHOLD - 最低信誉分数(0-100,默认:70)检查器增强但不替换现有安全功能:
# 尝试安装一个技能
node scripts/guarded_skill_install_wrapper.mjs --skill suspicious-skill --version 1.0.0
# 输出可能显示:
# 警告:技能 "suspicious-skill" 信誉分数较低(45/100)
# - 被 VirusTotal 代码洞察标记:加密密钥、外部 API、eval 使用
# - 作者没有发布其他技能
# - 技能发布时间少于 7 天
#
# 要忽略信誉警告进行安装,请运行:
# node scripts/guarded_skill_install_wrapper.mjs --skill suspicious-skill --version 1.0.0 --confirm-reputation
# 带确认安装
node scripts/guarded_skill_install_wrapper.mjs --skill suspicious-skill --version 1.0.0 --confirm-reputation
--confirm-reputation 安装前,始终审查技能代码ClawHub 在技能页面上显示两个安全徽章:
来自 clawsec-suite 页面的示例:
我们的检查器无法访问 OpenClaw 内部检查警告,因为它们未通过 clawhub CLI 或 API 公开。
为了启用完整的信誉检查,ClawHub 应通过以下方式公开内部检查结果:
clawhub inspect --json 端点clawhub install 警告输出中我们的启发式检查(技能年龄、作者信誉、下载量、更新)提供了类似的风险评估,但会错过关于绕过、缺少签名等具体的操作警告。始终在 ClawHub 网站上查看完整的安全评估。
要修改信誉检查逻辑,请编辑:
scripts/enhanced_guarded_install.mjs - 主要的增强安装器scripts/check_clawhub_reputation.mjs - 信誉检查逻辑hooks/clawsec-advisory-guardian/lib/reputation.mjs - 钩子集成GNU AGPL v3.0 或更高版本 - ClawSec 安全套件的一部分
每周安装量
119
代码仓库
GitHub 星标数
839
首次出现
2026年2月17日
安全审计
安装在
gemini-cli113
codex113
openclaw113
opencode113
cursor113
github-copilot112
Enhances the ClawSec suite's guarded skill installer with ClawHub reputation checks. Adds a second layer of security by checking VirusTotal Code Insight scores and other reputation signals before allowing skill installation.
clawhub install - Intercepts skill installation requestsThis skill must be installed after clawsec-suite:
# First install the suite
npx clawhub@latest install clawsec-suite
# Then install the checker
npx clawhub@latest install clawsec-clawhub-checker
# Run the setup script to integrate with clawsec-suite
node ~/.openclaw/skills/clawsec-clawhub-checker/scripts/setup_reputation_hook.mjs
# Restart OpenClaw gateway for changes to take effect
openclaw gateway restart
After setup, the checker adds enhanced_guarded_install.mjs and guarded_skill_install_wrapper.mjs under clawsec-suite/scripts and updates the advisory guardian hook. The original guarded_skill_install.mjs is not replaced.
After setup, run the wrapper (drop-in path) or the enhanced script directly:
# Recommended drop-in wrapper
node scripts/guarded_skill_install_wrapper.mjs --skill some-skill --version 1.0.0
# Or call the enhanced script directly
node scripts/enhanced_guarded_install.mjs --skill some-skill --version 1.0.0
The enhanced flow:
--confirm-reputation0 - Safe to install (no advisories, good reputation)42 - Advisory match found (existing behavior)43 - Reputation warning (new - requires --confirm-reputation)1 - ErrorEnvironment variables:
CLAWHUB_REPUTATION_THRESHOLD - Minimum reputation score (0-100, default: 70)The checker enhances but doesn't replace existing security:
# Try to install a skill
node scripts/guarded_skill_install_wrapper.mjs --skill suspicious-skill --version 1.0.0
# Output might show:
# WARNING: Skill "suspicious-skill" has low reputation score (45/100)
# - Flagged by VirusTotal Code Insight: crypto keys, external APIs, eval usage
# - Author has no other published skills
# - Skill is less than 7 days old
#
# To install despite reputation warning, run:
# node scripts/guarded_skill_install_wrapper.mjs --skill suspicious-skill --version 1.0.0 --confirm-reputation
# Install with confirmation
node scripts/guarded_skill_install_wrapper.mjs --skill suspicious-skill --version 1.0.0 --confirm-reputation
--confirm-reputationClawHub shows two security badges on skill pages:
Example from clawsec-suite page:
Our checker cannot access OpenClaw internal check warnings as they're not exposed via clawhub CLI or API.
To enable complete reputation checking, ClawHub should expose internal check results via:
clawhub inspect --json endpointclawhub install warning outputOur heuristic checks (skill age, author reputation, downloads, updates) provide similar risk assessment but miss specific operational warnings about bypasses, missing signatures, etc. Always check the ClawHub website for complete security assessment.
To modify the reputation checking logic, edit:
scripts/enhanced_guarded_install.mjs - Main enhanced installerscripts/check_clawhub_reputation.mjs - Reputation checking logichooks/clawsec-advisory-guardian/lib/reputation.mjs - Hook integrationGNU AGPL v3.0 or later - Part of the ClawSec security suite
Weekly Installs
119
Repository
GitHub Stars
839
First Seen
Feb 17, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykWarn
Installed on
gemini-cli113
codex113
openclaw113
opencode113
cursor113
github-copilot112
OpenClaw 安全 Linux 云部署指南:私有优先、SSH隧道、Podman容器化
40,100 周安装