Type a prompt. Watch all 14 layers evaluate it in sequence. See the final decision.
Try:
Kor'aelin
Avali
Runethic
Cassisivadan
Umbroth
Draumric
ALLOW
Safe operation. All layers pass.
What's happening: Each layer processes the input in sequence. The tongue bar shows which semantic
domains activated (bright = high activation). Layer scores accumulate into a final risk score.
The governance gate makes the binary decision: pass or block.
This runs in your browser and evaluates any prompt deterministically through a simplified
14-layer reference path (hyperbolic distance, harmonic wall scaling, and spectral proxies). It does not call a cloud model.