AI Agent Skills Explorer
Discover and analyze AI agent skills with security risk scoring. Every skill is analyzed for permissions, complexity, and potential risks.
⭐
OpenClaw
Curated Collection
Hand-picked Claude Code skills from the awesome-claude-skills list. Each skill analyzed for permissions and security context.
1271
Total Skills
594
Low Risk
Community CuratedAI Analyzed
Browse OpenClaw
📦
Community Skills
Full Directory
The complete community skills ecosystem with StackFox security analysis. Install counts, popularity, and risk ratings.
28,556
Total Skills
27,950
Low Risk
Install CountsAI Analyzed
Browse Directory
How Our Analysis Works
We use automated analysis to help you understand what each skill does
29,827
Skills analyzed
What We Analyze
Permission Detection
We scan for shell access, file system operations, network requests, credential handling, and database access
Risk Scoring
Based on detected permissions, we calculate a weighted risk score: Low (minimal access), Medium (some system access), or High (significant access)
AI-Generated Notes
An LLM reviews the skill content and provides plain-language explanation of what the skill does and potential concerns
Good to Know
⚠️
Automated, not manual
Our analysis is automated pattern matching + AI review. We don't manually audit each skill.
💡
High risk ≠ bad
Skills that need shell access or file permissions are marked higher risk, but that's often necessary for useful functionality.
✓
Always review before installing
Our analysis helps you understand risks, but you should always review skill code before installing.
Check security ratings before installing any AI agent skill