SkillPyramid: A Hierarchical Skill Consolidation Framework for Self-Evolving Agents

2026-06-02

AI Summary

Current AI agents lack systematic mechanisms for skill acquisition and transfer, resulting in redundant development, limited reusability of learned experience, and constrained generalization. To address this, this work proposes SkillPyramid, a novel framework that transforms static skill repositories into dynamic, evolving systems. By organizing skills into a hierarchical topological structure, SkillPyramid enables composition, validation, internalization, and cross-task transfer of skills, thereby facilitating continuous self-evolution of agents. Evaluated on ALFWorld, WebShop, and ScienceWorld, the framework achieves an average reward improvement of 38.0% and reduces execution steps by 27.7%, significantly outperforming baseline methods.

Abstract

Recent AI agents can flexibly invoke skills to solve complex tasks, but their long-term improvement is fundamentally constrained by a lack of systematic skill construction, accumulation, and transfer. In particular, without a unified framework for skill consolidation, agents tend to redundantly construct similar capabilities across different tasks, are unable to effectively transform experience into reusable assets, and struggle to generalize task-specific skills to novel scenarios. To address this limitation, we propose SkillPyramid, a skill consolidation framework that reuses existing skill experience for broader task generalization. Operating on a hierarchical skill topology, SkillPyramid further introduces a self-evolution mechanism that enables agents to compose, validate, and incorporate new skills during task execution. Experiments on ALFWorld, WebShop, and ScienceWorld across four backbone models show that SkillPyramid substantially increases the average reward by 38.0% and reduces execution steps by 27.7%. Overall, our method transforms a skill collection from a static resource pool into a dynamic evolution system.

Problem

• skill consolidation
• skill generalization
• self-evolving agents
• hierarchical skill framework
• experience reuse

Innovation

• Skill Consolidation
• Hierarchical Skill Topology
• Self-Evolving Agents
• Skill Reuse
• Dynamic Skill Evolution