# Seven-brand benchmark completion

This package prepares fresh independent visual scoring for AG1, HexClad, Gymshark, Jones Road Beauty, Grüns, True Classic, and Allbirds. The other 13 completed brand assessments are locked to the accepted recovered-nine report by `baseline-lock.json` and must pass through unchanged.

The fixed rubric is `SCORING-BRIEF.md`. Root first records accepted ordered frames and device qualifications in a copy of `evidence-selection.template.json`, saved as `evidence-selection.json`. Run `python3 build_evidence_manifest.py`; the builder hashes the selected pixels and derives full/mobile score eligibility only from the accepted device qualifications. It contains no brand-specific completion rule.

Each judge independently viewed every selected image and wrote seven JSON files under its own `scores/<judge>/` lane using `scoring-schema.json`. The two lanes remained isolated from one another and from prior scores for the seven brands. `python3 validate_package.py --require-scores` validates the final package. Root accepted all 14 fresh assessments and authorized publication at `/CHL/PDP/seven-completion-20260913-r1/`.