AgentEvolver

A wave of research shows that AI agents which manage their own context window using learned policies outperform those relying on fixed truncation rules. Approaches like AgentSwing and budget-aware rei…