🛡️ OWASP LLM01 Defense • Red Team Jailbreak Audit • Delimiter Hardening
Prompt Injection & Vulnerability Scanner
Test your production system prompts and agent instructions against known attack vectors: DAN overrides, system prompt extraction, delimiter evasion, and indirect payload smuggling.
📝 System Prompt or User Input to Audit
0 chars • ~0 tokens
OWASP LLM01 Hardening Checklist
- ✓ XML Boundary Encapsulation: Wrap untrusted user input within distinct tags like
<user_input>and forbid instructions within that tag from overriding system rules. - ✓ Negative Constraint Assertion: Explicitly forbid revealing internal system prompts, hidden variables, or executing unauthenticated shell commands.
- ✓ Dual-LLM Guardrail Verification: Pass user input through an orthogonal classifier model before sending it to tools or DB execution.
Security Vulnerability Assessment
Automated AST heuristic & adversarial rule match
100 / 100 Hardened
Suggested Defensive Patch