# Clean classification sample bundle

These files were generated by the Labelyst CLI from the repository's nine-row `manual-test-datasets/01-clean-classification.csv` fixture on this release. The result is a test fixture, not a customer dataset and not a claim about a production corpus.

The run scored 90/100 with warnings because the fixture has only three validation examples. It intentionally demonstrates that a high score does not hide limited class-coverage evidence.

- `result.json`: versioned core result
- `report.sarif`: machine-readable findings
- `report.html`: portable report
- `evidence.zip`: reproducible evidence bundle