@inproceedings{huang-etal-2026-skillrubricsrl,
title = "{S}kill{R}ubrics{RL}: Evolving Skill-Grounded Agents with Rubric-Guided Reinforcement Learning",
author = "Huang, Zihe and
Pan, Jianfei and
Zhang, YiFan and
Chen, Tao and
Lin, Jinghao and
Ruan, Kai and
Xu, Xiucheng and
Yao, Jian and
He, Yijun and
Yang, Haihua",
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.471/",
pages = "9870--9885",
ISBN = "979-8-89176-462-0",
abstract = "Long-horizon LLM agents increasingly rely on retrieved procedural skills as external memory. However, existing skill-augmented RL frameworks typically supervise a bundle of retrieved skills only with trajectory-level outcome rewards, leaving two key questions unanswered: whether each skill was executed and whether it helped task completion. We formulate this as a per-skill credit assignment problem. To address it, we propose \textbf{SkillRubricsRL}, which pairs each skill with an applicability-aware trajectory-level rubric and derives two complementary signals: a fulfillment signal for training correct skill execution and a counterfactual utility signal for retrieval and skill-bank evolution. Across ALFWorld and WebShop, SkillRubricsR achieves 97.7{\%} ALFWorld success, 84.4{\%} WebShop success, and a WebShop score of 92.7, improving over the strongest prior multi-skill baseline by 2.8, 2.1, and 1.6 percentage points, respectively, while increasing skill execution rate by over 50 percentage points. Code is available at \url{https://github.com/gminiheimdallr-a11y/SkillRubricsRL}."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="huang-etal-2026-skillrubricsrl">
<titleInfo>
<title>SkillRubricsRL: Evolving Skill-Grounded Agents with Rubric-Guided Reinforcement Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zihe</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianfei</namePart>
<namePart type="family">Pan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">YiFan</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tao</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jinghao</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kai</namePart>
<namePart type="family">Ruan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiucheng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jian</namePart>
<namePart type="family">Yao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yijun</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haihua</namePart>
<namePart type="family">Yang</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>Long-horizon LLM agents increasingly rely on retrieved procedural skills as external memory. However, existing skill-augmented RL frameworks typically supervise a bundle of retrieved skills only with trajectory-level outcome rewards, leaving two key questions unanswered: whether each skill was executed and whether it helped task completion. We formulate this as a per-skill credit assignment problem. To address it, we propose SkillRubricsRL, which pairs each skill with an applicability-aware trajectory-level rubric and derives two complementary signals: a fulfillment signal for training correct skill execution and a counterfactual utility signal for retrieval and skill-bank evolution. Across ALFWorld and WebShop, SkillRubricsR achieves 97.7% ALFWorld success, 84.4% WebShop success, and a WebShop score of 92.7, improving over the strongest prior multi-skill baseline by 2.8, 2.1, and 1.6 percentage points, respectively, while increasing skill execution rate by over 50 percentage points. Code is available at https://github.com/gminiheimdallr-a11y/SkillRubricsRL.</abstract>
<identifier type="citekey">huang-etal-2026-skillrubricsrl</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.471/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>9870</start>
<end>9885</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SkillRubricsRL: Evolving Skill-Grounded Agents with Rubric-Guided Reinforcement Learning
%A Huang, Zihe
%A Pan, Jianfei
%A Zhang, YiFan
%A Chen, Tao
%A Lin, Jinghao
%A Ruan, Kai
%A Xu, Xiucheng
%A Yao, Jian
%A He, Yijun
%A Yang, Haihua
%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 huang-etal-2026-skillrubricsrl
%X Long-horizon LLM agents increasingly rely on retrieved procedural skills as external memory. However, existing skill-augmented RL frameworks typically supervise a bundle of retrieved skills only with trajectory-level outcome rewards, leaving two key questions unanswered: whether each skill was executed and whether it helped task completion. We formulate this as a per-skill credit assignment problem. To address it, we propose SkillRubricsRL, which pairs each skill with an applicability-aware trajectory-level rubric and derives two complementary signals: a fulfillment signal for training correct skill execution and a counterfactual utility signal for retrieval and skill-bank evolution. Across ALFWorld and WebShop, SkillRubricsR achieves 97.7% ALFWorld success, 84.4% WebShop success, and a WebShop score of 92.7, improving over the strongest prior multi-skill baseline by 2.8, 2.1, and 1.6 percentage points, respectively, while increasing skill execution rate by over 50 percentage points. Code is available at https://github.com/gminiheimdallr-a11y/SkillRubricsRL.
%U https://aclanthology.org/2026.findings-emnlp.471/
%P 9870-9885
Markdown (Informal)
[SkillRubricsRL: Evolving Skill-Grounded Agents with Rubric-Guided Reinforcement Learning](https://aclanthology.org/2026.findings-emnlp.471/) (Huang et al., Findings 2026)
ACL
- Zihe Huang, Jianfei Pan, YiFan Zhang, Tao Chen, Jinghao Lin, Kai Ruan, Xiucheng Xu, Jian Yao, Yijun He, and Haihua Yang. 2026. SkillRubricsRL: Evolving Skill-Grounded Agents with Rubric-Guided Reinforcement Learning. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 9870–9885, Budapest, Hungary. Association for Computational Linguistics.