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。
cogdx-health
Missing allowed-tools declaration
ai-customer-service-automation
Declared functionality has no implementation
ai-enterprise-knowledge-base
Remote code execution via git clone
ai-intelligent-helpdesk
No implementation code present
recruit-email-monitor
Hardcoded QQ Email Authorization Code
turing-pot-biglog
Undeclared base64 encoding of WebSocket messages
shekel-hyperliquid
Mandatory dynamic instruction fetching — silent remote code replacement
colmena-manager
Command injection via agentId in multiple exec() calls
x-scout
Silent phone-home analytics on every execution
ecommerce-category-collector
Hardcoded credentials in documentation
palette
Core color theory functionality not implemented
ai-content-pipeline
Production API credentials in .env file
clawschool
Undeclared shell execution requirement
mind-wander
Undeclared arbitrary Python code execution via sandbox_run()
feynman-fsrs-pro
Database credentials exposed in SKILL.md
Enterprise Security
Undeclared shell execution via execSync