Risk Sample Stream

Which skills recently failed
or triggered trust review

This is not a popularity board. It shows recently reviewed skills that the system believes should be blocked or at least manually reviewed. The point is not how popular they are, but why they should not be installed blindly.

349 Risky samples surfaced
4 New in 7 days
0 Platform misses surfaced
All Code Exec Credential Theft Data Exfil Priv Esc Supply Chain Doc Deception Prompt Injection Obfuscation
65 /100
Trust
Review

seedance-creator

Remote script execution via curl|bash

Manual upload Apr 3, 2026
Open Report ↗
30 /100
Trust
High Risk

gitlab

Hardcoded GitLab API Token

Manual upload Apr 3, 2026
Open Report ↗
42 /100
Trust
High Risk

income-lab

Hardcoded API Key Exposed in Source Code

Manual upload Apr 3, 2026
Open Report ↗
60 /100
Trust
Review

varg-ai

远程脚本管道执行提示

Manual upload Apr 3, 2026
Open Report ↗
65 /100
Trust
Review

metacomp_visionx_kyt

npx远程代码执行风险

Manual upload Apr 3, 2026
Open Report ↗
45 /100
Trust
High Risk

AI Agent Skills Workspace

InStreet API Key 硬编码泄露

Manual upload Apr 3, 2026
Open Report ↗
55 /100
Trust
Review

linux-cron-panel

强制下载并执行第三方仓库代码

Manual upload Apr 3, 2026
Open Report ↗
65 /100
Trust
Review

mingquan-mcp

影子功能:未声明的 claw_report 遥测

Manual upload Apr 3, 2026
Open Report ↗
65 /100
Trust
Review

silicaclaw-owner-push

SKILL.md 安全边界声明不完整

Manual upload Apr 3, 2026
Open Report ↗
55 /100
Trust
Review

yuketang

未声明的远程数据上报

Manual upload Apr 3, 2026
Open Report ↗
55 /100
Trust
Review

OpenClaw Skills Collection

硬编码真实API密钥

Manual upload Apr 3, 2026
Open Report ↗
65 /100
Trust
Review

openclaw-free-search

未声明的 shell 执行能力

ClawHub Apr 3, 2026
Open Report ↗
0 /100
Trust
Block

math-calculator

隐藏的反向Shell Payload

GitHub Apr 2, 2026
Open Report ↗
← Previous
19 / 19