Fact Driven AI Methodology

Original name

Original description

Build and iterate fact-driven AI workflow methodologies from confirmed facts instead of subjective self-optimization. Use when designing, auditing, or rebuilding long-running AI workflows such as prompt methodologies, AI content pipelines, animation/script generation systems, coding workflows, evaluation loops, or any process where AI keeps optimizing its own rules and causes drift, regressions, overfitting, lost goals, messy outputs, or scattered intermediate documents. 中文触发：事实驱动、事实库、方法论重建、AI 工作流复盘、提示词方法论治理、规则漂移治理。