coredipper/operon-membrane
0
🛡️ Operon Prompt Injection Detector
Two-layer screening that analyzes prompts for injection patterns -- like a cell membrane and innate immune system working together to screen pathogens.
What to Try
- Select a preset from the Example dropdown (e.g. "Role hijack", "Nested code block", or "Data exfiltration") and click Analyze to see how both defense layers respond.
- Type your own prompt into the input box -- try mixing normal questions with injection patterns like "ignore previous instructions" to see threat levels escalate.
- Compare the Membrane result (SAFE/SUSPICIOUS/DANGEROUS/CRITICAL) with the InnateImmunity result (inflammation levels NONE through ACUTE) to see how two independent layers provide defense in depth.
How It Works
The Membrane uses adaptive pattern matching to classify threats and maintain an audit trail, while InnateImmunity applies TLR-style pattern recognition with inflammation escalation. Both layers analyze independently, and the combined verdict provides a defense-in-depth screening signal for prompt injection, jailbreaks, and data exfiltration patterns.