Skill Security Verifier
Run 5-layer security verification on MCP skill files. Checks schema integrity, code injection patterns, dangerous commands, network leaks, and DAG integrity.
When to Use
- User wants to verify if a skill file is safe
- User wants to check a skill before installing it
- User asks about skill security or validation
Usage
jadegate verify ./my_skill.json # Verify a local skill file
jadegate verify slack # Verify a skill by name from registry
5 Security Layers
- Schema validation - Structural integrity against jade-schema-v1
- Code injection scan - 22 dangerous patterns (eval, exec, subprocess, etc.)
- Dangerous command detection - 25 patterns covering destructive ops, permission escalation, etc.
- Network/data leak analysis - Whitelist-based URL and domain filtering
- DAG integrity - Circular dependency and reachability analysis
Output
Pass/fail for each layer with detailed findings.