Discover the toolingbehind modern intelligence
The operating system for AI infrastructure discovery. Search a live map of MCP servers, AI agents, LLM tools, automation systems, and developer infrastructure.
skill-scanner
Security Scanner for Agent Skills
2.1kstars
259forks
Python
agentagent-skillssecurity
Quality signals
ActiveNo testsNo CIThin docsNo licensePopular
AI summary
Auto-generated overview of what this skill does and when to reach for it.
Scans AI agent skills for security threats like prompt injection and malicious code using pattern matching, LLM analysis, and dataflow checks.
Use this when
- Debugging a skill that might be vulnerable to prompt injection
- Checking a new skill before deploying to production
- Integrating security checks into a CI/CD pipeline
- Reviewing third-party skills for potential risks
- Auditing skills for compliance with security policies
Don't use when
- Validating that a skill is 100% secure
- Scanning non-agent code or non-skill files
- Relying on this as the sole security measure
Capabilities
- Detects prompt injection and data exfiltration patterns
- Uses YARA and YAML rules for static analysis
- Runs LLM-based semantic analysis for threat detection
- Performs behavioral dataflow analysis to track skill behavior
- Outputs results in SARIF for GitHub Code Scanning
- Supports OpenAI Codex and Cursor Agent Skills formats
InstallSKILLInstall as a Claude Code skill
Universal installANY TOOL
$ npx skillcity@latest cisco-ai-defense/skill-scannerAuto-detects the skill or MCP server in this repo and installs it for Claude Code, Codex, Cursor or Cline.
Requires Node.js 18+ (ships with npx).
Or install manually
# Install into ~/.claude/skills/ mkdir -p ~/.claude/skills/skill-scanner curl -fsSL https://raw.githubusercontent.com/cisco-ai-defense/skill-scanner/main/SKILL.md -o ~/.claude/skills/skill-scanner/SKILL.md
Readme
Loading README…