Scan your AI artifacts for security risks — free, no account required.
Paste a GitHub repo URL. Jiffy reads every Claude skill, MCP config, .cursorrules, agents.md, and prompt template, then flags credential exfil, backdoors, prompt injection, and risky tool permissions.
Try:
We never store your repo contents past 72 hours for anonymous scans. Sign up to save scans and unlock full findings, file evidence, and SARIF export.
- Finds credential exfil, backdoors, prompt injection, over-privileged MCPs
- Reads Claude skills, MCP configs, Cursor/Windsurf rules, agents.md, Copilot instructions
- Free for public repos. Under 30 seconds.