AI Vulnerability Scanner for LLM Applications
Connect your LLM, chatbot or AI application and scan it with ready-to-run adversarial tests. Get evidence-backed findings and a detailed report — no test scripts to build.
The problem
AI applications fail in ways traditional scanners never check: a crafted prompt can override instructions, extract the system prompt or leak sensitive data. FilterPrompt sends adversarial prompts to your connected endpoint and evaluates every response.
What the scanner tests
- Direct and indirect prompt injection
- Jailbreak and policy bypass attempts
- System prompt extraction
- Sensitive data leakage
- Unsafe or harmful responses
- OWASP LLM Top 10 mapped findings
How the scan works
- Create a free account and connect OpenAI, Anthropic, Gemini, Azure OpenAI or a custom API endpoint with your own key.
- Pick a test suite and review the credit estimate.
- Run the scan and watch results arrive.
- Review each failing prompt, model response, severity and remediation suggestion in the report.
Test only applications you own or are authorized to assess. Results show observed behavior for the selected tests; a passing scan does not guarantee security.
Is this an online scanner?
Yes. FilterPrompt runs in the browser — no CLI or local install. You connect your endpoint and run scans from the web dashboard.
What does it cost?
Your first static scan is free with the welcome credit. A DIY scan credit costs $36, with no subscription.
Is FilterPrompt a firewall?
No. FilterPrompt is a vulnerability scanner: it tests your application and reports findings. It does not sit in front of production traffic.
