Jeudi 09 avril 2026, 14:00 à 15:00

salle de séminaire du département d'informatique

Benedikt Bollig

(LMF, Université Paris-Saclay)

Multi-agent systems built on large language models (LLMs) are difficult to reason about. Coordination errors such as deadlocks or type-mismatched messages are often hard to detect through testing. We introduce a domain-specific language for specifying agent coordination based on message sequence charts (MSCs). The language separates message-passing structure from LLM actions, whose outputs remain unpredictable. We define the syntax and semantics of the language and present a syntax-directed projection that generates deadlock-free local agent programs from global coordination specifications. Through examples such as reflection loops, multi-agent debate, and consensus protocols, we show how coordination properties can be established independently of LLM nondeterminism. This is joint work with Matthias Függer and Thomas Nowak.