@inproceedings{tang-etal-2026-memory,
title = "From Memory to Skills: Evidence-Grounded Governance for Long-Horizon {LLM} Agents",
author = "Tang, Bo and
Zhang, Yang and
Zhuang, Guomian and
Wei, Wenqiang and
Zheng, Gaoyang and
Xie, Lindong and
Tan, Yanchao and
Xiong, Feiyu and
Yang, Qingyu and
Chung, Edward and
li, Zhiyu",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1637/",
pages = "34826--34847",
ISBN = "979-8-89176-462-0",
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 show that MSCE achieves the best aggregate performance among the evaluated skill-augmented and memory-driven baselines, with statistically supported aggregate gains, positive cross-domain transfer, and continual improvement as online experience accumulates. Our code is available at https://github.com/MemTensor/MemOS/tree/main/apps/memos-local-plugin."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="tang-etal-2026-memory">
<titleInfo>
<title>From Memory to Skills: Evidence-Grounded Governance for Long-Horizon LLM Agents</title>
</titleInfo>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yang</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guomian</namePart>
<namePart type="family">Zhuang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wenqiang</namePart>
<namePart type="family">Wei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gaoyang</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lindong</namePart>
<namePart type="family">Xie</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanchao</namePart>
<namePart type="family">Tan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Feiyu</namePart>
<namePart type="family">Xiong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qingyu</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Edward</namePart>
<namePart type="family">Chung</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiyu</namePart>
<namePart type="family">li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<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 show that MSCE achieves the best aggregate performance among the evaluated skill-augmented and memory-driven baselines, with statistically supported aggregate gains, positive cross-domain transfer, and continual improvement as online experience accumulates. Our code is available at https://github.com/MemTensor/MemOS/tree/main/apps/memos-local-plugin.</abstract>
<identifier type="citekey">tang-etal-2026-memory</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1637/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>34826</start>
<end>34847</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T From Memory to Skills: Evidence-Grounded Governance for Long-Horizon LLM Agents
%A Tang, Bo
%A Zhang, Yang
%A Zhuang, Guomian
%A Wei, Wenqiang
%A Zheng, Gaoyang
%A Xie, Lindong
%A Tan, Yanchao
%A Xiong, Feiyu
%A Yang, Qingyu
%A Chung, Edward
%A li, Zhiyu
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F tang-etal-2026-memory
%X 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 show that MSCE achieves the best aggregate performance among the evaluated skill-augmented and memory-driven baselines, with statistically supported aggregate gains, positive cross-domain transfer, and continual improvement as online experience accumulates. Our code is available at https://github.com/MemTensor/MemOS/tree/main/apps/memos-local-plugin.
%U https://aclanthology.org/2026.findings-emnlp.1637/
%P 34826-34847
Markdown (Informal)
[From Memory to Skills: Evidence-Grounded Governance for Long-Horizon LLM Agents](https://aclanthology.org/2026.findings-emnlp.1637/) (Tang et al., Findings 2026)
ACL
- Bo Tang, Yang Zhang, Guomian Zhuang, Wenqiang Wei, Gaoyang Zheng, Lindong Xie, Yanchao Tan, Feiyu Xiong, Qingyu Yang, Edward Chung, and Zhiyu li. 2026. From Memory to Skills: Evidence-Grounded Governance for Long-Horizon LLM Agents. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 34826–34847, Budapest, Hungary. Association for Computational Linguistics.