Ofox editorial fixture, 2026-11-17. No model run is claimed.
Requires Python 3. Run python3 -m unittest -v (or python on Windows).
Starter: 4 tests, 2 failures. Reference: 4 tests pass.
Extract into two independent folders before a Desktop/CLI comparison.
Use the same model and equivalent permissions; do not change the tests.
Reference solution is editor-authored and is not imported by the tests.
This tiny fixture is not a coding-performance benchmark.
