sou350121

Adversarial research analysis framework that uses structured Bull/Bear/Arbiter debates to help users make better research judgments. Maintains a belief graph as backend engine, applies statistical calibration discipline, tracks phase transitions, and detects biases. MANDATORY TRIGGERS: Use this skill whenever the user asks to analyze a research paper, evaluate a research direction, make a strategic research decision, assess technology trends, review academic papers, or asks "what should I work on / invest in / bet on" in a research context. Also trigger when the user mentions "paper review", "research direction", "trend analysis", "technology forecast", "belief update", or wants structured pro/con analysis of any technical topic. Even casual requests like "what do you think about this paper" or "is X going to be important" should trigger this skill.

VLA (Vision-Language-Action) 领域专家。332+ 篇论文压缩记忆、信念网络、收敛地图、产业分析。 何时使用：深度分析、比较、判断或决策涉及 VLA 领域的内容时。例如：分析论文价值、比较架构优劣、判断研究方向、评估产业趋势、做技术选型。 何时不使用：随口提到 VLA/机器人/π0 但不需要深度分析时（闲聊、简单事实问答）。