<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wan-etal-2026-skillforge">
    <titleInfo>
        <title>SkillForge: Learning from Success–Failure Boundaries for Consistent Long-Horizon Agents</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Guangya</namePart>
        <namePart type="family">Wan</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Hao</namePart>
        <namePart type="family">Wang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Xiao</namePart>
        <namePart type="family">Ling</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Mingyang</namePart>
        <namePart type="family">Ling</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Xiaoqi</namePart>
        <namePart type="family">Ren</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Sheng</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
    </relatedItem>
    <abstract>Long-horizon LLM agents often fail from procedural errors such as forgetting constraints, misordering tool calls, or stopping prematurely. Because these slips are unpredictable, the same agent can succeed on one run and fail on an identical task later. The resulting execution trajectories contain rich signals about which procedural choices lead to success or failure, yet existing approaches that extract reusable guidance from agent trajectories either introduce noise through imprecise distillation or depend on external retrievers for deployment. We propose SkillForge, a training-free framework that rethinks both the construction and deployment of procedural guidance: it pairs successful and failed attempts on the same task to identify critical procedural choices, consolidates and filters candidate guidance to remove low-quality entries, and loads relevant guidance on demand at inference time only when the agent’s current context signals its relevance. Across agent benchmarks including τ²-Bench and Terminal-Bench, SkillForge improves repeated-run consistency, reducing the consistency gap by up to 37% and lowering total token usage by up to 23.6% compared to baselines, while moderately improving single-trial performance. These results suggest a practical path toward distilling high-quality procedural guidance from past trajectories for more consistent and efficient long-horizon agents.</abstract>
    <identifier type="citekey">wan-etal-2026-skillforge</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.1954/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>44486</start>
            <end>44510</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T SkillForge: Learning from Success–Failure Boundaries for Consistent Long-Horizon Agents
%A Wan, Guangya
%A Wang, Hao
%A Ling, Xiao
%A Ling, Mingyang
%A Ren, Xiaoqi
%A Li, Sheng
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F wan-etal-2026-skillforge
%X Long-horizon LLM agents often fail from procedural errors such as forgetting constraints, misordering tool calls, or stopping prematurely. Because these slips are unpredictable, the same agent can succeed on one run and fail on an identical task later. The resulting execution trajectories contain rich signals about which procedural choices lead to success or failure, yet existing approaches that extract reusable guidance from agent trajectories either introduce noise through imprecise distillation or depend on external retrievers for deployment. We propose SkillForge, a training-free framework that rethinks both the construction and deployment of procedural guidance: it pairs successful and failed attempts on the same task to identify critical procedural choices, consolidates and filters candidate guidance to remove low-quality entries, and loads relevant guidance on demand at inference time only when the agent’s current context signals its relevance. Across agent benchmarks including τ²-Bench and Terminal-Bench, SkillForge improves repeated-run consistency, reducing the consistency gap by up to 37% and lowering total token usage by up to 23.6% compared to baselines, while moderately improving single-trial performance. These results suggest a practical path toward distilling high-quality procedural guidance from past trajectories for more consistent and efficient long-horizon agents.
%U https://aclanthology.org/2026.emnlp-main.1954/
%P 44486-44510

Markdown (Informal)

[SkillForge: Learning from Success–Failure Boundaries for Consistent Long-Horizon Agents](https://aclanthology.org/2026.emnlp-main.1954/) (Wan et al., EMNLP 2026)

• SkillForge: Learning from Success–Failure Boundaries for Consistent Long-Horizon Agents (Wan et al., EMNLP 2026)

ACL

• Guangya Wan, Hao Wang, Xiao Ling, Mingyang Ling, Xiaoqi Ren, and Sheng Li. 2026. SkillForge: Learning from Success–Failure Boundaries for Consistent Long-Horizon Agents. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 44486–44510, Budapest, Hungary. Association for Computational Linguistics.