Skill Shield

SkillShield is the first security-scored directory for AI agent skills and MCP servers. We automatically scan every extension for malicious code, credential theft, prompt injection attacks, and vulnerabilities before listing them.

Our scanner has analyzed 33,530+ skills from ClawHub, SkillsMP, and MCP Market. We discovered that 12% contain critical security risks including reverse shells, data exfiltration, and API key stealers.

Every skill receives a 0-100 trust score and detailed risk breakdown. Browse by category, search by name or author, and filter by security level. Free CLI scanner available for local analysis. Key Features: Pre-scanned database of 33,530+ AI skills Security scoring (0-100) with risk levels Static code analysis and behavior detection Free CLI tool for local auditing Open source with transparent methodology

Built for developers who don't want to get hacked by their own AI agents.