EmoMAS: Emotion-Aware Multi-Agent System for High-Stakes Edge-Deployable Negotiation with Bayesian Orchestration

EmoMAS addresses a complex niche: enabling high-stakes negotiation on edge devices where large-scale LLMs are too costly or privacy-prohibitive. By using Bayesian orchestration to manage 'emotional' states in SLMs, it attempts to compensate for the reduced reasoning capacity of smaller models. Quantitatively, the project is in its infancy with 0 stars and a 5-day lifespan, identifying it as a research artifact rather than a viable product or community-driven project at this stage. The defensibility is low (2/10) because the implementation is likely a reference for the associated paper rather than a production-ready library. The primary competitive threat comes from two sides: 1) Frontier labs improving SLM (e.g., Phi-4, Gemma-3) native reasoning and emotional intelligence to the point where external Bayesian orchestration becomes redundant, and 2) Established agent frameworks like Microsoft's AutoGen or LangChain's LangGraph adding emotional/Bayesian modules as standard components. Given the 1-2 year horizon of model improvement, this specific orchestration technique risks being absorbed into the weights of next-gen edge models or being superseded by general-purpose agent libraries.