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。
VLAN Linux Client Skill
Remote script piped to bash without integrity verification
openviking-context
Undeclared curl|bash remote script execution
minimax-web-search
Hardcoded API Key in Source Code
minimax-cp
Hardcoded MiniMax API Key Exposed
recruit-email-monitor
Hardcoded QQ Email Authorization Code
shekel-hyperliquid
Mandatory dynamic instruction fetching — silent remote code replacement
Enterprise Security
Undeclared shell execution via execSync
minimal-agent
Unrestricted Arbitrary Command Execution via V1 Mode
maxianer
Undeclared external data transmission
openclaw-memory-auto
Hardcoded Windows username path leaks user identity
deepsafe-scan
Network access not declared in SKILL.md
Bounty Hunter Agent
Hardcoded DeepSeek API Key in Documentation
kuaidi-query
Hardcoded API Credentials Exposed
self-evolution-engine
Hardcoded Billing API Key in Source Code
long-term-memory
Hardcoded API Key in Source Code
nano-banana-pro
Hardcoded DASHSCOPE_API_KEY in _meta.json