DecArena introduces a benchmark for evaluating agents in decentralized environments using graph-constrained communication. Tests with over 100,000 tools show that current LLM agents struggle to discover non-local resources and collaborate without a central orchestrator.
HOW THIS AFFECTS YOU
●
researcherYou can use this benchmark to test how agents perform in decentralized systems with massive toolsets.