SkillAdaptor: Self-Adapting Skills for LLM Agents from Trajectories

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yu, Zhuoyun, Xie, Xin, Yao, Wuguannan, Wang, Chenxi, Liang, Lei, Qi, Xiang, Deng, Shumin
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916071961788416
author Yu, Zhuoyun
Xie, Xin
Yao, Wuguannan
Wang, Chenxi
Liang, Lei
Qi, Xiang
Deng, Shumin
author_facet Yu, Zhuoyun
Xie, Xin
Yao, Wuguannan
Wang, Chenxi
Liang, Lei
Qi, Xiang
Deng, Shumin
contents Large language model (LLM) agents increasingly rely on reusable external skills to solve long-horizon interactive tasks. Existing training-free skill adaptation pipelines usually update skills from full trajectories or session-level feedback, which makes failure attribution coarse and often produces unstable or overly broad revisions. We propose SkillAdaptor, a training-free step-level skill adaptation framework with explicit failure attribution, and it can plug into OpenClaw-class agent harnesses. Given a failed trajectory, SkillAdaptor identifies a first actionable fault step, links responsibility to candidate skills, and applies targeted updates under explicit acceptance checks while keeping the backbone frozen. We evaluate on WebShop, PinchBench, and Claw-Eval with Kimi-K2.5, GLM-5, and GPT-5.2. SkillAdaptor improves over no-skill and skill-adaptation baselines on all three suites, with the largest single-metric improvements of +1.5 points on PinchBench Avg Score%, +1.8 on Claw-Eval Avg Score, and +1.7 on WebShop success rate. These results indicate that step-level attribution supports more stable and auditable training-free skill maintenance\footnote{The code will be released at https://github.com/zjunlp/SkillAdaptor.}.
format Preprint
id arxiv_https___arxiv_org_abs_2606_01311
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SkillAdaptor: Self-Adapting Skills for LLM Agents from Trajectories
Yu, Zhuoyun
Xie, Xin
Yao, Wuguannan
Wang, Chenxi
Liang, Lei
Qi, Xiang
Deng, Shumin
Computation and Language
Artificial Intelligence
Machine Learning
Multiagent Systems
Large language model (LLM) agents increasingly rely on reusable external skills to solve long-horizon interactive tasks. Existing training-free skill adaptation pipelines usually update skills from full trajectories or session-level feedback, which makes failure attribution coarse and often produces unstable or overly broad revisions. We propose SkillAdaptor, a training-free step-level skill adaptation framework with explicit failure attribution, and it can plug into OpenClaw-class agent harnesses. Given a failed trajectory, SkillAdaptor identifies a first actionable fault step, links responsibility to candidate skills, and applies targeted updates under explicit acceptance checks while keeping the backbone frozen. We evaluate on WebShop, PinchBench, and Claw-Eval with Kimi-K2.5, GLM-5, and GPT-5.2. SkillAdaptor improves over no-skill and skill-adaptation baselines on all three suites, with the largest single-metric improvements of +1.5 points on PinchBench Avg Score%, +1.8 on Claw-Eval Avg Score, and +1.7 on WebShop success rate. These results indicate that step-level attribution supports more stable and auditable training-free skill maintenance\footnote{The code will be released at https://github.com/zjunlp/SkillAdaptor.}.
title SkillAdaptor: Self-Adapting Skills for LLM Agents from Trajectories
topic Computation and Language
Artificial Intelligence
Machine Learning
Multiagent Systems
url https://arxiv.org/abs/2606.01311