SkillOpt-Lite: Better and Faster Agent Self-evolution via One Line of Vibe

Hugging Face Daily Papers Papers
skill-optimization autonomous-agents zeroth-order-optimization agent-self-evolution llm-agents code-agents

Summary

SkillOpt-Lite proposes a minimal viable pipeline for skill optimization in autonomous agents, achieving better and faster self-evolution by treating all components as editable code and integrating into production coding agents. It formalizes skill optimization via Zeroth-Order optimization and outperforms prior methods on benchmarks.

While skill optimization for autonomous agents has gained traction, existing methods rely on complex pipelines. This leaves a fundamental question unaddressed: What constitutes a minimal viable pipeline for skill optimization, where every component is justified by theory or empirical necessity? We formalize skill optimization via Zeroth-Order (ZO) optimization, mapping classical counterparts (central difference, trust regions) to recent literature. Noting that unlike blind numerical perturbations in classical ZO, skill trajectories serve as interpretable debugging feedback. Grounded in Claude Code philosophy and PAC learning, we establish three principles for convergence and generalization: file-system-based trajectory exploration, consensus attribute mining, and independent validation gating. Eliminating redundancies, we propose SkillOpt-Lite. It accelerates convergence and outperforms full SkillOpt: improving LiveMath by +8.8 points on GPT-5.5 and +25.4 points on GPT-5.4-nano, allowing the nano model to surpass standard GPT-5.4 optimized by SkillOpt. Finally, we integrate our framework into production coding agents like VSCode Copilot, enabling developers to evolve agent skills via one line of vibe. Because our framework treats all agent components simply as standard editable code, this minimal pipeline naturally generalizes to full harness optimization (HarnessOpt). On SpreadsheetBench, HarnessOpt enables GPT-5.4-nano to achieve 0.7758 accuracy, outperforming the larger GPT-5.5 running standard pipelines (0.7620). Code is available at https://github.com/EvolvingLMMs-Lab/SkillOpt-Lite.
Original Article
View Cached Full Text

Cached at: 07/08/26, 02:47 AM

Paper page - SkillOpt-Lite: Better and Faster Agent Self-evolution via One Line of Vibe

Source: https://huggingface.co/papers/2607.03451

Abstract

A minimal viable pipeline for skill optimization is proposed through Zeroth-Order optimization formalization, eliminating redundancies while maintaining convergence and generalization through trajectory exploration, consensus mining, and validation gating principles.

Whileskill optimizationfor autonomous agents has gained traction, existing methods rely on complex pipelines. This leaves a fundamental question unaddressed: What constitutes a minimal viable pipeline forskill optimization, where every component is justified by theory or empirical necessity? We formalizeskill optimizationvia Zeroth-Order (ZO) optimization, mapping classical counterparts (central difference, trust regions) to recent literature. Noting that unlike blind numerical perturbations in classical ZO, skill trajectories serve as interpretable debugging feedback. Grounded in Claude Code philosophy and PAC learning, we establish three principles forconvergenceandgeneralization: file-system-basedtrajectory exploration,consensus attribute mining, and independentvalidation gating. Eliminating redundancies, we proposeSkillOpt-Lite. It acceleratesconvergenceand outperforms full SkillOpt: improving LiveMath by +8.8 points on GPT-5.5 and +25.4 points on GPT-5.4-nano, allowing the nano model to surpass standard GPT-5.4 optimized by SkillOpt. Finally, we integrate our framework into production coding agents like VSCode Copilot, enabling developers to evolve agent skills via one line of vibe. Because our framework treats all agent components simply as standard editable code, this minimal pipeline naturally generalizes to full harness optimization (HarnessOpt). On SpreadsheetBench,HarnessOptenables GPT-5.4-nano to achieve 0.7758 accuracy, outperforming the larger GPT-5.5 running standard pipelines (0.7620). Code is available at https://github.com/EvolvingLMMs-Lab/SkillOpt-Lite.

View arXiv pageView PDFProject pageGitHub14Add to collection

Get this paper in your agent:

hf papers read 2607\.03451

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2607.03451 in a model README.md to link it from this page.

Datasets citing this paper1

#### cy0307/awesome-loop-engineering Viewer• Updatedabout 10 hours ago • 380 • 1.06k • 2

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.03451 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

SkillOpt: Executive Strategy for Self-Evolving Agent Skills

Hugging Face Daily Papers

SkillOpt introduces a systematic text-space optimizer for agent skills that trains skills as external agent state with stable updates and zero deployment inference overhead, achieving superior performance across multiple benchmarks and execution environments.

@AlphaSignalAI: https://x.com/AlphaSignalAI/status/2069064122218717387

X AI KOLs Timeline

This article explores how AI agents can automatically write and optimize their skill files using techniques like SkillOpt from Microsoft Research, which treats skill documents as trainable state and delivers significant performance improvements. It addresses the challenge of manual skill tuning and presents frameworks like GEPA and EvoSkill as evolutionary approaches.

OpenSkill: Open-World Self-Evolution for LLM Agents

Hugging Face Daily Papers

OpenSkill is a framework for LLM agents to self-evolve skills and verification signals from open-world resources without target-task supervision, achieving high performance across benchmarks.