# FLR002-LAB1 — Architecture Implementation Divergence Record v1

Use the eight local synthetic fixtures in `fixtures/cases.json`.

For each case:
1. record direct observations before inference;
2. state the active architecture rule and intended owner;
3. keep source/import structure separate from runtime path;
4. verify whether an allowed exception has valid authority and scope;
5. classify a material gap only when evidence supports it;
6. distinguish containment from repair;
7. define a recurrence sensor for closure; and
8. preserve bounded unknowns as OPEN instead of guessing.

Implement `build_record(case)` in `starter.py`. The learner package contains no hidden expected classification. Everything is synthetic and offline.