agent systems Harness Framework A declarative, skill-based harness for rapidly composing LLM agents. Memory Optimization Training-free optimization of agent memory for long-horizon tool use. Context Management Context engineering for long-horizon, multi-step agents. Self-Evolution Agent-native loops that let models keep improving after deployment. E-commerce Agent Landing Taking LLM agents from prototype to production for e-commerce governance. research ReGRPO — Reflection-Augmented RL for Tool-Using Agents Group-relative policy optimization with structured self-reflection for long-horizon tool use. PathComp — Lifelong Vision–Language–Action Adaptation Path-anchored compatibility for continually fine-tuning VLA policies. Ego-centric Predictive Model Two-stage egocentric video prediction conditioned on hand trajectories. Task-Agnostic Compatible Adapter (TaCA) Upgrade visual foundation models without retraining downstream tasks.
