1. Final decision
The final detector verdict after combining the NLP extraction layer with the SMT2 policy engine.
2. NLP extraction
This is what the NLP and semantic extraction layer spat out from your input before formal policy reasoning kicked in.
No NLP extraction yet.
3. SMT2 policies hit
These are the formal SMT policies that fired, plus the frame/directive/predicate evidence used by those policies.
No SMT policy evidence yet.
4. Full SMT2 program
The exact SMT-LIB2 emitted for this input, including declarations, assertions, policy booleans, and the final injection decision logic.
No SMT2 program yet.
NLP facts (directive / predicate compatibility layer)
Compatibility facts derived from the semantic frame layer. Useful for debugging transitions from legacy logic to frame-first logic.
No directive/predicate facts yet.
Raw detector hits
Low-level grammatical and semantic detector outputs before they are folded into frames and policies.
No grammatical or semantic detections yet.
Processing details
Extra context about preprocessing, structured chat handling, concealment signals, and rule application.
No processing details yet.