multi-agent-patterns

SKILL.md

Multi-agent architectures distribute work across multiple language model instances, each with its own context window. When designed well, this distribution enables capabilities beyond single-agent limits. When designed poorly, it introduces coordination overhead that negates benefits. The critical insight is that sub-agents exist primarily to isolate context, not to anthropomorphize role division.

Multi-agent systems address single-agent context limitations through distribution. Three dominant patterns exist: supervisor/orchestrator for centralized control, peer-to-peer/swarm for flexible handoffs, and hierarchical for layered abstraction. The critical design principle is context isolation—sub-agents exist primarily to partition context rather than to simulate organizational roles.

Effective multi-agent systems require explicit coordination protocols, consensus mechanisms that avoid sycophancy, and careful attention to failure modes including bottlenecks, divergence, and error propagation.

Cette compétence doit être utilisée lorsque l'utilisateur demande de « concevoir un système multi-agents », « d'implémenter un modèle de superviseur », de « créer une architecture en essaim », de « coordonner plusieurs agents » ou mentionne des modèles multi-agents, l'isolation du contexte, les transferts d'agents, les sous-agents ou l'exécution d'agents parallèles. Source : guanyang/antigravity-skills.

Voir l'original