Leave blank to run in simulation mode (demonstrates the interface without a live API).
Deploy the Laminar API to Railway and paste the URL above for real detections.
How this works: Laminar calibrates on 80 benign flows — no attack samples, no labels, no model training.
The PCF engine builds a structural fingerprint of normal traffic. When attack traffic arrives, the correlation structure
deviates from baseline and both Signal A (PSV cosine distance) and Signal B (attack token fraction) fire independently.
Patent-pending algorithm. Patent U.S. Provisional No. 63/978,633.
2.8M real flows benchmarked. Full pipeline audit available on request.