Free tool
Is your AI agent injectable?
Paste a prompt or an agent configuration. Get a security score, findings and fixes in seconds.
Runs entirely in your browser — nothing you paste is sent or stored.
Security report
Your score and findings appear here after you scan.
A one-off scan isn't runtime protection.
This scan shows risks at a single point in time. Vigil continuously checks every prompt, retrieved document and tool call — and can block attacks before they reach your model or tools.
No credit card required · Full Pro access for 14 days · Then free Starter forever