hundredschools

name	hundredschools
description	Multi-dimensional agent thinking framework mapping classical philosophical schools to LLM control strategies. Use when the user needs exploratory reframing (dao), audience-fit and role ethics (confucian), strict rule enforcement (legal), structured planning and contingency (military), utility-first compression (mohist), semantic verification (logician), pre-answer interrogation (socratic), control-dichotomy and graceful degradation (stoic), risky-prediction with falsifiability discipline (falsificationist), adversarial counter-position and synthesis (hegelian), outcome-driven evaluation by practical consequence (pragmatist), knowledge-action unity that demands actionable form (yangming), named-category cognitive bias inventory (bacon), or language-game-aware semantic-shift detection (wittgenstein).
license	MIT
metadata	{"author":"FanBB2333","version":"0.4.0"}

HundredSchools / 诸子百家控制框架

HundredSchools maps classical philosophical schools — six from pre-Qin China and eight later additions spanning early-modern empiricism, the Mediterranean tradition, modern Western philosophy, and Ming-dynasty Neo-Confucianism — into distinct LLM control stances. Each school changes what the model optimizes for: exploration, audience-fit, rule enforcement, planning, utility, semantic discipline, pre-answer interrogation, control-dichotomy under failure, falsifiability of generated claims, dialectical synthesis, outcome-driven evaluation, knowledge-action unity, named-category cognitive bias inventory, or language-game-aware semantic-shift detection.

HundredSchools 将经典哲学流派映射为不同的 LLM 控制立场——六家来自先秦中 国，另外八家是后续拓展（早期近代经验主义、地中海传统、近代西方哲学， 以及明代新儒学）。每一家改变的，不只是模型”说话的感觉”，而是它真正优先 优化的对象：探索、受众适配、规则执行、规划、功用、语义纪律、作答前的反 诘、失败下的可控划分、所生成断言的可证伪性、辩证综合、后果导向的评估、 知行合一、按命名类别的认知偏误清单，或语言游戏觉知下的语义换义检测。