# LEARNING_PROTOCOL — GCORF-05 v0.1

## Learning event admission

1. Identify changed layer
2. Record before/after state
3. Record update source
4. Test behavioral effect
5. Test persistence
6. Test reusability
7. Test transfer
8. Run attribution / ablation if multiple layers changed
9. Record evidence
10. Classify:
   - BESA
   - learning
   - false_learning
   - unknown

## Learning layers

- parameter
- context
- memory
- retrieval
- tool
- operator
- protocol
- bottom_space

## Bottom-space update audit

Record:
- reachability delta
- representation delta
- evidence delta
- tool delta
- operator-access delta
- protocol delta
- mutual-model delta
- distortion risk