version: "1.0.1" name: wine-blending-fusion description: 'Run prompt on models with different strengths; fusion pass extracts best elements selectively — not averaging. Use when models complement (reasoning + phrasing + facts). Scope boundary: single-model routing → emergency-triage-compute.' compatibility: ChatGPT (web, desktop, mobile via plugins) and Codex (desktop, CLI, IDE). metadata: host: chatgpt-codex ported_from: Cursor_Skills
Wine blending multi-model fusion
#28 · Domain: Enology · Category: refinement · Difficulty: 🔴 High
Core principle
Selective composition — each varietal covers others' weaknesses.
AI problem addressed
Route-to-one or average ensembles lose complementary strengths.
Implementation
Model A: structure/skeleton Model B: phrasing/metaphor Model C: citations/facts Fusion: compose strengths; cover weaknesses — do not average.
Boundaries
- Prototype as prompt scaffold (🟢) before full pipeline middleware (🟡/🔴)
- Category router: `ai-transfer-refinement`
- Catalog: `ai-transfer-ecosystem-primer`
- Runtime plugin id:
wine_blending