SkillGraph: Skill-Augmented Reinforcement Learning for Agents via Evolving Skill Graphs
For LLM agents performing compositional tasks, SkillGraph addresses the limitations of flat skill libraries by providing structural cues for skill dependencies and library maintenance.
SkillGraph introduces a directed skill graph with typed edges to represent dependencies between skills, enabling better retrieval and maintenance for compositional tasks. It achieves state-of-the-art performance on ALFWorld, WebShop, and seven QA tasks, with significant gains on complex multi-step tasks.
Skill libraries enable large language model agents to reuse experience from past interactions, but most existing libraries store skills as isolated entries and retrieve them only by semantic similarity. This leads to two key challenges for compositional tasks. Firstly, an agent must identify not only relevant skills but also how they depend on and build upon each other. Secondly, it also makes library maintenance difficult, since the system lacks structural cues for deciding when skills should be merged, split, or removed. We propose SKILLGRAPH, a framework that represents reusable skills as nodes in a directed graph, with typed edges encoding prerequisite, enhancement, and co-occurrence relations. Given a new task, SKILLGRAPH retrieves not just individual skills, but an ordered skill subgraph that can guide multi-step decision making. The graph is continuously updated from agent trajectories and reinforcement learning feedback, allowing both the skill library and the agent policy to improve together. Experiments on ALFWorld, WebShop, and seven search-augmented QA tasks show that SKILLGRAPH achieves state-of-the-art performance against memory-augmented RL methods, with especially large gains on complex tasks that require composing multiple skills.