Paper page - SkillNet: Create, Evaluate, and Connect AI Skills

Source: https://huggingface.co/papers/2603.04448 Published on Feb 26

#1 Paper of the day Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

SkillNet introduces an open infrastructure for systematically accumulating and transferring AI skills through a unified ontology, significantly improving agent performance across multiple domains.

CurrentAI agentscan flexibly invoke tools and execute complex tasks, yet theirlong-term advancementis hindered by the lack of systematic accumulation and transfer of skills. Without aunified mechanismforskill consolidation, agents frequently ``reinvent the wheel’’, rediscovering solutions in isolated contexts without leveraging prior strategies. To overcome this limitation, we introduce SkillNet, an open infrastructure designed to create, evaluate, and organize AI skills at scale. SkillNet structures skills within a unifiedontologythat supports creating skills from heterogeneous sources, establishing rich relational connections, and performingmulti-dimensional evaluationacross Safety, Completeness, Executability, Maintainability, and Cost-awareness. Our infrastructure integrates a repository of over 200,000 skills, an interactive platform, and a versatile Python toolkit. Experimental evaluations onALFWorld,WebShop, andScienceWorlddemonstrate that SkillNet significantly enhancesagent performance, improving average rewards by 40% and reducingexecution stepsby 30% across multiple backbone models. By formalizing skills as evolving, composable assets, SkillNet provides a robust foundation for agents to move from transient experience to durable mastery.

View arXiv pageView PDFProject pageGitHub878Add to collection

Get this paper in your agent:

hf papers read 2603\.04448

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/2603.04448 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2603.04448 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2603.04448 in a Space README.md to link it from this page.

Collections including this paper9

Browse 9 collections that include this paper