From Memory to Skills: Evidence-Grounded Co-Evolution Governance for Long-Horizon LLM Agents
For LLM agents requiring long-horizon tasks, this work provides a method to evolve memory into skills without training, addressing a key bottleneck in agent autonomy.
MSCE introduces a training-free framework that converts LLM agent experiences into reusable skills with evidence links, achieving significant improvements over state-of-the-art baselines on EvoAgentBench and LoCoMo, with strong cross-domain transferability.
Existing memory systems for long-horizon LLM agents often retrieve prior traces as passive context rather than converting them into executable capabilities. In this paper, we propose MSCE, a training-free Memory--Skill Co-Evolution framework that organizes agent experience into grounded step traces, reusable procedural policies, and declarative environmental cognition. MSCE crystallizes evidence-backed L2 policies with positive estimated gain into callable skills that retain evidence links, applicability boundaries, decision guidance, verification rules, and reliability estimates. It further introduces reflection-weighted value backfilling, which propagates sparse terminal feedback through dense local self-reflections to produce evidence-calibrated trace values for governing memory and skill evolution. Experiments on EvoAgentBench and LoCoMo demonstrate that MSCE significantly outperforms state-of-the-art skill-augmented and memory-driven agent baselines, exhibiting strong cross-domain transferability and lifelong-evolution capabilities.