SkillVine: Agent Skill Evolution via Branching Exploration

Kaiwei Liu; Jiqian Dong; Liran Dong; Shuai Mao; Mingming Zhao; Bufang Yang; Jie Chuai; Zhitang Chen

cs.AISep 26, 2026

SkillVine: Agent Skill Evolution via Branching Exploration

Authors: Kaiwei Liu, Jiqian Dong, Liran Dong, Shuai Mao, Mingming Zhao, Bufang Yang, Jie Chuai, Zhitang Chen,

Organizations: The Chinese University of Hong Kong · Noah’s Ark Lab, Huawei Technologies

Abstract

Agent skills encapsulate reusable procedural knowledge that enables LLM agents to perform tasks, and they can be improved automatically using trajectories from interactions with the environment. This is the classic problem of skill evolution. Existing approaches predominately follow a linear evolution paradigm, in which updates are sequentially applied to the latest skill-library version. As a result, they inevitably fall into local optima, leaving many promising evolution paths unexplored. We propose SkillVine, an automatic skill-evolution framework that formulates skill evolution as a graph search problem and employs a branching exploration strategy. Equipped with a trunk-branch collaborative searching mechanism, an intelligent parent-node selector, and an adaptive-granularity update rule, SkillVine achieves a balance between exploration and exploitation. We evaluate SkillVine on 5 benchmarks with two LLMs. Results show that SkillVine discovers better skill-library versions along branches than along the linear trunk and achieves the best test performance in nine of ten benchmark-model combinations.

Figures & tables

Figure 1: Overview of SkillVine. SkillVine organizes skill library versions into a graph, maintains a continuous trunk, selects branch nodes based on quality, exploration, and complementarity, and adapts branch-update granularity based on whether the trunk update is accepted.

Panel of Figure 2 — Algorithm 1 SkillVine

Model	Method	Benchmark Score
		Spreadsheet	LiveMath	SearchQA	OfficeQA	TABVERSE
DeepSeek-V4 -flash	Linear	48.04	43.40	73.36	47.30	63.88
	Best-of-3	64.41	50.00	73.14	45.27	64.49
	Random	53.38	46.23	74.14	52.70	64.08
	SkillOpt	62.99	47.17	73.50	46.62	63.47
	SkillOpt-Lite	66.19	50.48	73.14	37.84	61.84

Table 1: Main results across benchmarks. Higher values indicate better performance.

Model	Ablation	Benchmark Score
		Spreadsheet	LiveMath	SearchQA	OfficeQA	TABVERSE
	w/o Selector & AGBU	53.38	46.23	74.14	52.70	64.08
	w/o Selector	67.26	46.23	72.93	47.30	63.47
	w/o AGBU	65.84	50.00	74.57	52.03	63.27
	w/o Comp	71.53	53.77	73.36	55.41	60.82
DeepSeek-V4 -flash	SkillVine	71.53	53.77	74.43	55.41	67.35

Table 2: Ablation study results. Higher values indicate better performance.

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Model	Benchmark	Evolution Path
DeepSeek-V4 -flash	SpreadsheetBench	M:v0001 $\rightarrow$ D:v0004 $\rightarrow$ R:v0007 $\dagger$ $\rightarrow$ D:v0026 $\dagger$ $\rightarrow$ D:v0030
	LiveMath	M:v0001 $\rightarrow$ D:v0004 $\rightarrow$ R:v0006 $\rightarrow$ D:v0011 $\dagger$
	SearchQA	M:v0001 $\rightarrow$ D:v0004 $\rightarrow$ D:v0010 $\rightarrow$ D:v0041 $\dagger$ $\rightarrow$ R:v0187 $\dagger$
	OfficeQA	M:v0001 $\rightarrow$ R:v0009
	TABVERSE	M:v0001 $\rightarrow$ D:v0004 $\rightarrow$ D:v0010 $\dagger$
Qwen3.8 -Flash	SpreadsheetBench	M:v0001 $\rightarrow$ R:v0003 $\rightarrow$ R:v0013 $\dagger$ $\rightarrow$ D:v0027 $\dagger$

Appendix

Table 3: Evolution paths to the final best nodes across the ten benchmark–model combinations. Each vXXXX is a skill library version number within that run. M:vXXXX denotes a node on the trunk; D:vXXXX denotes a side-branch node generated using a discovery selector; R:vXXXX denotes a side-branch node generated using a refinement selector.

\dagger

means that AGBU rejects at least one edit during the update.

Case 1: DeepSeek-V4-flash $\times$ SpreadsheetBench
Version Node	Skill Capability	Unique / Lost	Typical Validation Samples	Final Gain
Trunk M:v0002 Branch R:v0007	R:v0007 adds Python-based calculation beyond M:v0002 ’s workbook inspection and preservation. M:v0002 rules: Inspect headers and cell types; read cached formula values while preserving the workbook. R:v0007 rules: Compute in Python and write values directly; infer the intended operation from the actual table when the supplied formula is flawed.	M:v0002 0 / 7 R:v0007 7 / 0	10747: Two-key profit lookup M:v0002 : None (fail) R:v0007 : -10600 (pass) 51556: Conditional average M:v0002 : None (fail) R:v0007 : 82 (pass) The branch also solves all 21 examples solved by the trunk.	Final best D:v0030 outperforms the best trunk M:v0028 by 20.51 on validation and 23.49 on test.
Case 2: DeepSeek-V4-flash $\times$ SearchQA
Version Node	Skill Capability	Unique / Lost	Typical Validation Samples	Final Gain
Trunk M:v0031 Branch D:v0010	D:v0010 handles wordplay and answer forms missed by M:v0031 ; M:v0031 better resolves associated entities in the examples shown. M:v0031 rule: For a bare entity name, identify the associated category or relationship from context. D:v0010 rules: Match the source answer form; for word transformations, return both the original and transformed terms.	M:v0031 5 / 9 D:v0010 9 / 5	“Adam Levine” M:v0031 : Maroon 5 (pass) D:v0010 : Adam Levine (fail) Holes on postage stamps M:v0031 : perforation (fail) D:v0010 : perforations (pass) Hawaiian wreath becoming a sheltered area M:v0031 : lee (fail) D:v0010 : lei & lee (pass)	Final best R:v0187 outperforms the best trunk M:v0134 by 5.00 on validation and 1.07 on test.

Appendix

Table 4: Two examples of selected complementary nodes.

Case 1: DeepSeek-V4-flash $\times$ TABVERSE
Version Node	Rejected Trunk Update	AGBU-Filtered Branch Update	Accepted Gain
Mainline M:v0002 $\rightarrow$ M:v0005 (rejected) Branch D:v0004 $\rightarrow$ R:v0007 $\dagger$	Proposed rule: Return the full cell text, such as “Middelkerke, Belgium”, rather than one part. Result: The update included this rule. It lost three previously solved samples and only gained two: 45/70 $\rightarrow$ 44/70 ( $-1.43$ ), so it was rejected.	AGBU kept one edit and rejected two . The kept edit selects the correct column and computes durations from date ranges: 47/70 $\rightarrow$ 49/70. An edit with the same full-cell rule then reduced the score to 43/70 ( $-8.57$ ), so AGBU rejected it.	The accepted child R:v0007 $\dagger$ improved over D:v0004 : 47/70 $\rightarrow$ 49/70, +2 samples ( +2.86 ).
Case 2: Qwen3.8-Flash $\times$ SearchQA
Version Node	Rejected Trunk Update	AGBU-Filtered Branch Update	Accepted Gain
Mainline M:v0020 $\rightarrow$ M:v0058 (rejected) Branch M:v0002 $\rightarrow$ D:v0066 $\dagger$	Proposed rule: Shorten names: use “Webster” instead of “Noah Webster” and “the Colorado” instead of “Colorado River”. Result: The update included this rule. The score fell from 152/200 to 148/200 ( $-2.00$ ), so it was rejected.	AGBU kept two edits and rejected two . An edit with the same name-shortening rule reduced the score from 147/200 to 141/200 ( $-3.00$ ), so it was rejected. The kept edits select the specific answer type and use “&” for paired answers such as “Nina & Pinta”. Together, they raised the score to 151/200.	The accepted child D:v0066 $\dagger$ improved over M:v0002 : 147/200 $\rightarrow$ 151/200, +4 samples ( +2.00 ).

Appendix

Table 5: Two examples of harmful edits rejected by AGBU.

Figure 6: Evolution path of DeepSeek-V4-flash on SpreadsheetBench. Nodes are labeled with their version numbers and validation scores. Gray dashed nodes denote rejected candidate versions. Blue arrows denote accepted trunk updates, while orange and green arrows denote accepted side-branch updates using the discovery and refinement selectors, respectively.

\dagger

marks updates for which AGBU rejects at least one edit.

Model	Method	Validation Score
		Spreadsheet	LiveMath	SearchQA	OfficeQA	TABVERSE
DeepSeek-V4 -flash	Linear	56.41	54.29	77.00	61.22	70.00
	Best-of-3	71.79	57.14	77.00	65.31	71.43
	Random	64.10	54.29	77.00	67.35	70.00
	SkillOpt	71.79	60.00	78.00	57.14	70.00
	SkillOpt-Lite	71.79	45.71	76.00	53.06	68.57

Appendix

Table 6: Validation results across benchmarks. Higher values indicate better performance.

LLM Agent Skill Learning

Large language model agent skill learning studies how agents acquire, represent, reuse, and refine procedures for multi-step tasks.

See all 402 papers

MultiEvoSkill: Structure-Aware Multi-Agent Skill Evolution for Generalizable Skill Artifacts
Oct 25, 2026Kounianhua Du, Jianxing Liu, Ruoshui Shan +4LLM Agent Skill Learning Large Language Model-Guided Optimization
Shanghai Jiao Tong University Shanghai, China · Purdue · Tencent
SkillHEX: Improving Agent Skills via Hypothesis-Driven Autonomous Exploration and Exploitation
Aug 6, 2026Yuru Feng, Yaoqi Chen, Beidi Zhao +7Test-Time Adaptation LLM Agent Self-Improvement
Microsoft · University of California, San Diego · University of Science and Technology of China +1
SkillGrad: Optimizing Agent Skills Like Gradient Descent
May 26, 2026Hanyu Wang, Yifan Lan, Bochuan Cao +2Large Language Model-Guided Optimization LLM Agent Self-Improvement
College of Information Sciences and Technology The Pennsylvania State University University Park, PA, USA

Oct 25, 2026J/K move · Enter open · S save

MultiEvoSkill: Structure-Aware Multi-Agent Skill Evolution for Generalizable Skill Artifacts

Kounianhua Du, Jianxing Liu, Ruoshui Shan, Yucheng Zhang+3

Shanghai Jiao Tong University Shanghai, China · Purdue · Tencent

LLM Agent Skill Learning Large Language Model-Guided Optimization Multi-Agent LLM Systems

Open paper AskPDF arXiv

Skill documents are becoming the default adaptation surface for frozen large language model (LLM) agents, but current skill generators face two key limitations. First, humans and different LLMs differ in how they organize procedural knowledge and approach task solving. Skills authored from a human or single-model perspective may therefore fail to align with the reasoning behaviors of other backbones, limiting cross-model generalization. Second, sequential skill descriptions inadequately represent procedures with parallel branches and explicit dependencies, while existing skill optimizers primarily operate on prose and lack explicit support for graph-structured skill evolution. We introduce MultiEvoSkill, a skill-optimization framework comprising three core components: 1) Skill compiler casts the skill as a typed graph with tailored operators, producing a compact, informative representation with explicit procedural topology; 2) Heterogeneous agent system proposes revision advices using collective intelligence, where minimal and informative issues are identified based on multiple sets of experiences. 3) Structure-aware debate and evolution mechanism organizes candidate edits into a heterogeneous graph, resolves them under dependency and importance constraints, and distills the collective proposal into the artifact. Multi-dimensional gating ensures the information gain and topological correctness of each evolution. Optimal skills are then obtained when the system converges at a consistent state. Various experiments and ablations against SOTA methods validate the effectiveness and generalization of MultiEvoSkill.

Abstract

LLM Agent Skill Learning

Explore similar work

MultiEvoSkill: Structure-Aware Multi-Agent Skill Evolution for Generalizable Skill Artifacts

SkillHEX: Improving Agent Skills via Hypothesis-Driven Autonomous Exploration and Exploitation

SkillGrad: Optimizing Agent Skills Like Gradient Descent