最近有哪些 skills
不值得默认信任
这不是热度榜。这里展示的是最近被送来判断、且系统认为需要阻止或至少人工复核的 skills。重点不是它们多流行,而是为什么它们不该被直接装进环境。
kuaishou-genius-actual
--insecure 模式禁用 TLS 证书校验
skill-gatekeeper
未声明的外部命令执行
hpr-solver
Undeclared LLM API calls to OpenRouter
evolution-watcher
Documentation mismatch - file modification not declared
问专家技能
Bypass robot detection declared as legitimate use case
colmena-manager
Command injection via agentId in multiple exec() calls
lessac_offline_voice_system
False claim of offline operation
lock-me-in
Undeclared stealth/anti-detection browser scripts
calendar_memo
Undeclared shell command execution
claw-shell
Unrestricted shell execution with weak safety controls
aagent-system
Undeclared External Script Execution
buy-domain-helper
Undeclared shell execution via execSync and spawn
Rune
SSRF proof-of-concept with live metadata service IP
rewrite_question
Network capability declared as NONE but actual traffic exists
Grok Swarm
Undeclared credential access from OpenClaw auth profiles