From Memory to Skills: Evidence-Grounded Co-Evolution Governance for Long-Horizon LLM Agents

A new training-free framework, Memory--Skill Co-Evolution (MSCE), enables long-horizon LLM agents to convert passive memory traces into reusable skills, outperforming state-of-the-art baselines on EvoAgentBench and LoCoMo benchmarks. The method, detailed in arXiv:2607.16621v1, uses evidence-grounded policies and reflection-weighted value backfilling to govern co-evolution of memory and skills, demonstrating strong cross-domain transferability.

arXiv:2607.16621v1 Announce Type: new Abstract: 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.