This release makes the repository's purpose clearer and establishes an evidence-safe model comparison protocol.
- README hero: Same model. Same prompt. Different workflow.
- screenshots labelled Flutter Mobile and Flutter Tablet/Desktop
- explicit clarification that the showcase is one adaptive Flutter codebase, not a separate web implementation
- shared benchmark prompt, rubric, baseline and skill-guided manifests, and results page
- Claude, Codex, and Grok results remain pending until controlled paired runs exist
- repository validation blocks incomplete, unsupported, or one-sided benchmark publication
Benchmark protocol: https://github.com/musabekisakov-imj/flutter-design-engineer/tree/main/benchmarks/connected-command-center