Building Reliable Long-Horizon Agents: A Survey

Building Reliable Long-Horizon Agents surveys how long-horizon capability should be defined, measured, evaluated, and engineered across the full agent system. It covers model design, harness design, environments, benchmarks, reliability, recovery, and evaluation protocols.

Visit the project page.