synthetic-narrative-plan-eval

To grade a synthetic narrative plan, evaluate paired markdown and YAML files against project design documents, app deploy summaries, and stakeholder fitness bars. This process generates weighted dimension scores and a structured verdict YAML file.

Synthetic narrative evaluation in ACE Phase 7 assesses demo readiness by checking narrative coherence, KPI alignment, and character realism. It prevents unconvincing product demos by enforcing quality gates before generating final demo data.

Validating a synthetic demo narrative involves checking PDD anchoring, cast realism, anomaly coherence, manifest schema validity, and stakeholder narrative quality. This multi-dimension grading ensures the demo story aligns with project goals.

Hard deduct triggers during synthetic narrative evaluation include critical failures like invalid field paths, missing coaching arcs, or non-anomaly-aligned KPIs. These blocker and warn triggers prevent low-quality demos from progressing to production.

The synthetic narrative evaluation requires no external dependencies. It operates independently to grade artifacts and output structured verdict YAML files per the lib/verdict-schema.ts specification.

You can use this evaluation specifically for Stage 4 of ACE Phase 7 Plan B workflows. It validates synthetic narrative plan artifacts before demo data generation to ensure believable stories for stakeholder presentations.