1.1 KiB
Unified Optimization Case Library
We will model long-term storage as a unified optimization case library rather than separate archives for article optimization and human-tone copy optimization. Cases are saved automatically during optimization so original inputs, audit summaries, process summaries, and formal outputs are not lost; users can enrich them afterward with titles, ownership metadata, tags, and notes. Re-running optimization from a case creates a new result version rather than overwriting prior output, preserving comparison and learning history. The shared library supports type-based filtering, shared ownership metadata, shared LLM audit summaries, and shared list/detail navigation, while each case type keeps type-specific detail modules. Both case types include a first-version entry point and empty state for publication performance and learning feedback so the case library can also become the performance learning library. Publication records and performance snapshots are shared concepts, but quality scoring uses type-specific rubrics so article quality and human-tone copy quality are not forced into one artificial score.