ACLArena: Agent Continual Learning in Multi-stage Post-training

Building general-purpose agents for industrial deployment requires integrating multiple capabilities, each typically acquired at a distinct stage of training. Yet there is currently no well-established recipe for agent continual learning (ACL), with little understanding of the trade-offs among existing integration paradigms. To address this gap, we introduce ACLArena, a framework for comprehensively studying, analyzing, and evaluating ACL. We first build a sequential training pipeline and conduct an in-depth analysis that explains the mechanisms of forgetting and generalization from two complementary perspectives, the model level and the token level. Building on distillation and compensation viewed through these different lenses, we systematically compare three prevailing technical approaches. Through extensive experiments, we develop a detailed understanding of how capabilities transfer across stages; our findings provide the first systematic understanding of capability integration in realistic agent post-training. Finally, we propose a new ACL recipe that combines offline replay over high-quality trajectories with a routed network of multiple LoRA experts each specialized via RL, substantially improving the agent's ability to learn across multiple domains. Comprehensive experiments on four reasoning and agentic tasks, evaluated under both in-domain and out-of-domain settings, demonstrate the value of our analysis and the effectiveness of our approach. Our code is open-sourced at: https://anonymous.4open.science/r/ACLArena-C7B4.