SkillLens: Adaptive Multi-Granularity Skill Reuse for Cost-Efficient LLM Agents

2026-05-08

AI Summary

This work addresses the challenge faced by large language model agents in reusing skill libraries, where coarse-grained skills introduce noise while fine-grained rewriting incurs prohibitive computational costs. To reconcile relevance and efficiency, the authors propose SkillLens, a novel framework featuring a four-layer hierarchical skill graph—spanning policies, strategies, procedures, and primitives—combined with hybrid-granularity semantic retrieval, degree-corrected random walks, and a local validation mechanism. This architecture dynamically decides whether to reuse or rewrite skills and enables co-evolution of the skill library and validator. Evaluated on MuLocBench and ALFWorld, SkillLens significantly outperforms strong baselines, improving defect localization Acc@1 by up to 6.31 percentage points and raising task success rate from 45.00% to 51.31%.

Abstract

Skill libraries have become a practical way for LLM agents to reuse procedural experience across tasks. However, existing systems typically treat skills as flat, single-resolution prompt blocks. This creates a tension between relevance and cost: injecting coarse skills can introduce irrelevant or misleading context, while rewriting entire skills is expensive and often unnecessary. We propose SkillLens, a hierarchical skill-evolution framework that organizes skills into a four-layer graph of policies, strategies, procedures, and primitives, and retrieves them at mixed granularity. Given a task, SkillLens first retrieves semantically relevant skill seeds, expands them through degree-corrected random walk over the skill graph, and then uses a verifier to decide whether each visited unit should be accepted, decomposed, rewritten, or skipped. This enables the agent to reuse compatible subskills directly while adapting only locally mismatched components. To improve the system over time, SkillLens further refines multi-granularity skills and verifier in order to improve its routing decisions. We provide theoretical analysis showing that mixed-granularity adaptation incurs sublinear cost under sparse mismatch assumptions and that the evolutionary update rule monotonically improves the validation objective until a local optimum. Across MuLocbench and ALFWorld, SkillLens consistently improves over strong skill-based baselines, achieving up to a 6.31 percentage-point Acc@1 gain for bug localization and raising agent success rate from 45.00% to 51.31%.

Problem

• skill reuse
• cost-efficiency
• multi-granularity
• LLM agents
• skill libraries

Innovation

• multi-granularity skill reuse
• hierarchical skill graph
• mixed-granularity retrieval
• skill adaptation
• LLM agent efficiency