securityFree Instant Scan
Is your AI agent
vulnerable?
Paste your endpoint URL. We fire 3 adversarial attacks — a system prompt extraction, a remote code execution injection, and an architecture recon probe — and show you exactly what happens.
The JSON field your endpoint reads the user message from. Common values: message, input, query, prompt
If your endpoint requires authentication. e.g. Authorization / Bearer token
lockWe don't store your endpoint responses. Results are shown once and discarded when you leave this page.
No sign-up required · Your endpoint receives 3 HTTP POST requests