SkillMOO: Multi-Objective Optimization of Agent Skills for Software Engineering

2026-04-10

AI Summary

Manually tuning the skill sets of large language model (LLM)-based coding agents is costly and yields fragile performance. This work proposes SkillMOO, a novel framework that introduces multi-objective evolutionary optimization to this task for the first time. By integrating the NSGA-II algorithm, a failure-analysis-driven skill editing mechanism, and an iterative evaluation pipeline, SkillMOO automatically evolves skill compositions that balance success rate, computational cost, and runtime. Experimental results demonstrate that streamlined, focused skill sets outperform verbose or redundant instructions, achieving up to a 131% improvement in pass rates on three software engineering tasks from SkillsBench while reducing inference costs by 32%. Moreover, the optimization overhead remains practically manageable.

Technology Category

• Search and Optimization: Evolutionary Computation
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Agent skills provide modular, task-specific guidance for LLM- based coding agents, but manually tuning skill bundles to balance success rate, cost, and runtime is expensive and fragile. We present SkillMOO, a multi-objective optimization framework that automatically evolves skill bundles using LLM-proposed edits and NSGA-II survivor selection: a solver agent evaluates candidate skill bundles on coding tasks and an optimizer agent proposes bundle edits based on failure analysis. On three SkillsBench software engineering tasks, SkillMOO improves pass rate by up to 131% while reducing cost up to 32% relative to the best baseline per task at low optimization overhead. Pattern analysis reveals pruning and substitution as primary drivers of improvement, suggesting effective bundles favor minimal, focused content over accumulated instructions.

Problem

• multi-objective optimization
• agent skills
• LLM-based coding agents
• skill bundles
• software engineering

Innovation

• multi-objective optimization
• LLM-based coding agents
• skill bundle evolution
• NSGA-II
• automated skill tuning