SkillPyramid: A Hierarchical Skill Consolidation Framework for Self-Evolving Agents
2026-06-02 • Artificial Intelligence
Artificial IntelligenceComputation and Language
AI summaryⓘ
The authors explain that current AI agents have trouble improving over time because they can't efficiently build, save, and reuse skills. They introduce SkillPyramid, a system that organizes skills in a hierarchy and allows AI agents to combine and update skills while working on tasks. This helps agents generalize their skills better and avoid repeating the same work. Tests show that SkillPyramid improves performance and makes task completion faster. Overall, the authors propose turning skills into a dynamic, evolving system instead of a static collection.
AI agentsskill consolidationhierarchical skill topologyskill generalizationself-evolution mechanismtask executionperformance improvementALFWorldWebShopScienceWorld
Authors
Yuan Xiong, Ziqi Miao, Qian Chen, Lijun Li, Yequan Wang, Shizhu He, Jun Zhao, Kang Liu
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.