feat: Placeholder Registry Framework + Part A Nutrition Metrics
Part A Implementation (Nutrition Basis Metrics):
- Registry-based metadata system (flexible, not hardcoded)
- 4 placeholders registered: kcal_avg, protein_avg, carb_avg, fat_avg
- Evidence-based tagging (code-derived, draft-derived, unresolved, to_verify)
- Single source of truth for all consumers (Prompt, GUI, Export, Validation)
Technical:
- backend/placeholder_registry.py: Core registry framework
- backend/placeholder_registrations/nutrition_part_a.py: Part A registrations
- backend/placeholder_registry_export.py: Export integration
- backend/routers/prompts.py: /placeholders/export-values-extended integration
Metadata completeness:
- 22 metadata fields per placeholder
- Evidence tracking for all fields
- Architecture alignment (Layer 1/2a/2b)
NO LOGIC CHANGE:
- Data Layer unchanged (nutrition_metrics.py)
- Resolver unchanged (placeholder_resolver.py)
- Values identical (only metadata/export enhanced)
Breaking Change Risk: NONE
Deploy Risk: VERY LOW (only export enhancement)
Plan: .claude/task/rework_0b_placeholder/NUTRITION_PART_A_CHANGE_PLAN.md
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Lars
645967a2ab