{"id":"arxiv/2608.19047","url":"https://arxiv.org/abs/2608.19047","tags":["cs.AI","math.NT"],"score":0,"title":"Eureka: Task-Conditioned Meta-Agent Orchestration for Scientific Discovery","domain":"cs.AI","source":"arxiv","authors":["Alizer Wong","Heng Cui","Yi Tan","Xiongchao Zhan","Liang Lin","Yuxiang Guo","Zhaorong Dai","Zixin Zeng","Wenyuan Li"],"reviews":0,"subject":"cs.AI","summary":"We present Eureka, a task-conditioned Meta-Agent architecture that compiles long-horizon tasks into dynamic obligation graphs with explicit acceptance semantics. During execution, Eureka forms Macro-Agents with specialized state, memory, operators, tools, verifiers, and local topology via receding-horizon planning, architecture promotion, and minimal-sufficient compilation. When bottlenecks recur, cost-benefit-gated evolution updates the local architecture under constraints. Theoretically, we establish results on regret, planning invalidation, amortization, subtree interfaces,…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-08-19","reactions":{}}