Skill optimization improves by using existing task knowledge effectively
Retrieval-Augmented Skill Optimization via Cross-Harness Adaptation
Artificial Intelligence
Summary
Optimizing how agents perform tasks usually requires lots of trial and error, which is costly and slow. The authors propose a method called RASO that helps agents improve their task-solving skills by learning from a large collection of existing skills instead of starting from scratch. RASO uses two steps: first, it creates an initial skill by retrieving relevant knowledge without needing costly trials; second, it refines that skill by incorporating feedback while continuing to gather helpful information. Experiments show this approach works better than methods that do not use such retrieval techniques.
What this means in practice
- •For robotics developers: Improve robot task performance by initializing and refining skills from large skill libraries rather than costly trial runs alone.
- •For software automation engineers: Enhance software agent efficiency by adapting existing skill sets from different domains to new automation tasks without starting optimization from zero.
Authors
Jaewon Chu, Ji Soo Lee, Jihwan Park, Dohwan Ko, Jeehye Na, Seunghun Lee, Taehoon Lee, Minseo Yoon, Minseok Joo, Yunyang Xiong, Hyunwoo J. Kim
Abstract
An agent skill is a reusable, actionable natural-language artifact that guides an agent to perform a task effectively under a given harness. Recent studies have explored the optimization of agent skills, contributing to a growing collection of publicly available skills spanning diverse tasks, domains, and harnesses. Despite millions of publicly shared skills, existing skill optimization methods largely overlook this accumulated knowledge, instead relying solely on expensive agent rollouts to iteratively refine skills for a target task. To address this, we propose \textbf{Retrieval-Augmented Skill Optimization (RASO)}, a framework that leverages an external skill corpus as prior knowledge throughout skill optimization. RASO retrieves relevant knowledge from existing skills and adapts it to the target task and harness via Cross-Harness Adaptation, accounting for mismatches in both domain and harness. RASO comprises two complementary stages: \textbf{Retrieval-Augmented Skill Initialization (RASI)} constructs a knowledge-grounded initial skill without requiring agent rollouts, while \textbf{Retrieval-Augmented Skill Update (RASU)} iteratively refines the skill by retrieving external knowledge guided by execution feedback. Across four agent benchmarks and two models, extensive experiments show that RASO consistently outperforms baselines without retrieval-augmented skill initialization and updating.