LYGO Skill Gate

这是专为OpenClaw/ClawHub包设计的本地预安装风险扫描工具。核心检测子进程/网络调用、硬编码密钥、权限声明不符等风险，全程不联网、不执行扫描代码，基于纯标准库运行。典型场景包括ClawHub技能审计、安全安装检查、恶意软件分类等。支持输入技能路径或slug，以退出码标识风险等级，可导出报告，帮助用户或Agent判断技能安全性。

deepseekoracle

热度 15.85

最近更新 2026-08-07 12:23:03

分类：安全依赖：代码执行 / 读写文件应用场景：audit / 报告撰写

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 lygo-skill-gate

Metadata

name: lygo-skill-gate
description: LYGO Skill Gate — local pre-install skill risk scanner for OpenClaw/ClawHub packages. Scan any skill folder before you install or trust it: subprocess/shell, network/HTTP, secrets in source, eval/exec, webhook/exfil hints, permission-claim mismatches. Use when auditing ClawHub skills, reviewing SKILL.md safety, safe install checks, malware triage, or SkillSpector-style local gates. Pure stdlib. No network, no subprocess, no auto-install. Install clawhub:@deepseekoracle/lygo-skill-gate.
version: 1.0.0
license: MIT-0
metadata: {"openclaw":{"emoji":"🛂","homepage":"https://github.com/DeepSeekOracle/lygo-protocol-stack/tree/main/docs/skills/lygo-skill-gate","requires":{"anyBins":["python","python3"]}},"lygo":true,"security":true,"audit":true,"clawhub_safe_install":true,"signature":"Delta9Phi963-SKILL-GATE-v1.0.0","publisher":"deepseekoracle","permissions":{"network":false,"shell":false,"subprocess":false,"filesystem":{"read":"skill directory path you pass to scan","write":"skill state/ only with --i-consent"},"publish":false,"auto_install":false}}

LYGO Skill Gate v1.0.0

Scan before you install.

After ClawHavoc-style incidents, every human and every agent needs a local gate that reads a skill package and scores risk — without calling home, without running the skill.

Signature: Delta9Phi963-SKILL-GATE-v1.0.0
ClawHub: @deepseekoracle/lygo-skill-gate

Problem this solves

Who	Need
Humans	“Is this ClawHub skill safe to install?”
Agents	“Before I load a skill tree, does code match its no-network / no-subprocess claims?”
Operators	Reproducible risk band + findings list for audit logs

Searches: skill security, audit skill, safe install, clawhub malware, skillspector local, scan SKILL.md.

Install

npx clawhub@latest install deepseekoracle/lygo-skill-gate

Commands

cd path/to/lygo-skill-gate
python scripts/self_check.py

# Scan a skill by path
python scripts/skill_gate.py scan "I:/E Drive/.grok/skills/lygo-context-guard"

# Or by slug (searches common skill roots)
python scripts/skill_gate.py scan lygo-ops-detector

# Self-demo (scan this package)
python scripts/skill_gate.py self-demo

# Write report under state/ (consent)
python scripts/skill_gate.py scan ./some-skill --write last_scan.json --i-consent

Exit	Meaning
0	clear / low
5	elevated or claim mismatch
10	high / critical — do not install casually
2	bad path / not a skill

What it checks

subprocess / os.system / Popen / shell=True
HTTP clients, sockets, webhooks
eval / exec, pickle, unsafe yaml
Hardcoded key-like strings
Destructive deletes, git push / auto-publish hints
Claim mismatch: frontmatter says network: false but code uses urlopen

Does not: download skills, install skills, execute scanned code, or phone home.

Pair with

Skill	Role
`lygo-context-guard`	Token budget + secret redact for prompts
`lygo-ops-detector`	Discourse signal heuristics
`lygo-kickstart-wizard`	Onboarding map

License

MIT-0.
Δ9Φ963 — verify before trust · local gate · human remains the publisher.