Scan before you install.
Paste any GitHub repo or npm package. Get instant security analysis for prompt injection, jailbreak payloads, and supply-chain risks.
No signup. No AI calls. Pure static analysis in under 3 seconds.
How the Skill Scanner works
Paste a repo or package
Enter any GitHub URL or npm package name. The scanner fetches target files like SKILL.md, package.json, and shell scripts.
Static analysis runs instantly
12+ security signatures check every line for prompt injection, jailbreak attempts, dangerous shell ops, and obfuscated payloads.
Get a risk score and findings
See a 0-100 risk score, severity breakdown, and the exact lines that triggered each rule. Share your results with your team.
Frequently asked questions
What does the Skill Scanner check for?
It scans for 12+ categories of risk including prompt injection, jailbreak attempts, role hijacking, remote code execution, obfuscated payloads, suspicious install hooks, data exfiltration patterns, and hardcoded external URLs. All checks run as static regex analysis — no AI APIs are called.
Is the Skill Scanner free to use?
Yes, completely free with no signup required. You can scan unlimited repos and npm packages.
What files does it scan?
It targets high-risk files: SKILL.md, README.md, package.json, shell scripts (.sh), and any JavaScript/TypeScript files in /scripts, /bin, or /hooks directories.
Can I use this to audit MCP skills and AI agent plugins?
Absolutely — it was built specifically for that. Before installing any MCP skill, Cursor plugin, or AI agent tool, run it through the scanner to catch prompt injection and hidden payloads.
How is the risk score calculated?
Each finding has a severity (CRITICAL, HIGH, MEDIUM, LOW). CRITICAL findings score 25 points, HIGH scores 15, MEDIUM scores 8, and LOW scores 3. The total is capped at 100. Any CRITICAL finding sets the risk level to CRITICAL regardless of score.