PROMPT INJECTION SCANNER
Is your AI prompt hackable?
Find out in 10 seconds.
Paste your system prompt. We fire 15 real injection attacks: prompt extraction, instruction override, data leak probes, output manipulation, encoding evasion. Judge each response with AI, and hand you a grade A–F.
Privacy-first & Automation-ready: Manual scans run fully in your browser. Or use our REST API to scan prompts in GitHub Actions and CI/CD pipelines during deploy. View API docs ↗
Get actionable fixes: For every vulnerability found, we suggest three levels of remediation — quick prompt fixes, input sanitization code, and architectural patterns to harden your system.
OPENROUTER API KEY
Calls run from your browser using your key. Safe: Keys passed in secure headers and never logged. Get a free key →
YOUR SYSTEM PROMPT
The full instructions your app sends to the model as the system message.