aiqbench analysis
COBRA-Skills introduces a novel framework for optimizing LLM agent skills by integrating contextual bandit strategies to minimize costly execution evaluations. This approach significantly reduces resource requirements while maintaining robust performance across diverse benchmarks and target models, addressing key efficiency challenges in agent development.
COBRA-Skills: Contextual Bandit-Guided Evolution for Agent Skill Optimization
arXiv:2609.11682v1 Announce Type: new Abstract: Large language model (LLM) agents can benefit from reusable skills distilled from prior task experience, yet existing skill optimization methods often rely on costly execution-based evaluation and substantial task data. We introduce \textbf{COBRA-Skills}, an efficient framework that formulates skill optimization as budgeted sequential optimization over a dynamically evolving candidate space. COBRA-Skills couples contextual-bandit-guided prioritization with evidence-grounded skill evolution, selectively allocating evaluations to promising or informative candidates while continually refining the skill population from execution feedback. Across six heterogeneous agent benchmarks and three target models, COBRA-Skills consistently achieves the strongest average performance among compared methods, while reducing optimization cost by 55--58\% relative to SkillOpt and using only 50 unique optimization examples per benchmark. Further analyses show that COBRA-Skills remains robust to changes in the agent harness and performs effectively when the target model itself is used for skill generation and refinement.
Abstract:Large language model (LLM) agents can benefit from reusable skills distilled from prior task experience, yet existing skill optimization methods often rely on costly execution-based evaluation and substantial task data. We introduce \textbf{COBRA-Skills}, an efficient framework that formulates skill optimization as budgeted sequential optimization over a dynamically evolving candidate space. COBRA-Skills couples contextual-bandit-guided prioritization with evidence-grounded skill evolution, selectively allocating evaluations to promising or informative candidates while continually refining the skill population from execution feedback. Across six heterogeneous agent benchmarks and three target models, COBRA-Skills consistently achieves the strongest average performance among compared methods, while reducing optimization cost by 55--58\% relative to SkillOpt and using only 50 unique optimization examples per benchmark. Further analyses show that COBRA-Skills remains robust to changes in the agent harness and performs effectively when the target model itself is used for skill generation and refinement.
Artificial Intelligence (cs.AI)
arXiv:2609.11682 [cs.AI]
(or arXiv:2609.11682v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2609.11682 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 234 words
Related coverage
MAGE: Multi-Agent Self-Evolution with Co-Evolutionary Knowledge Graphs
arXiv · Oct 7, 2026
10 Ways AI Helps You Build What Your Business Needs
AppDirect · May 15, 2026
Fusion is the New Mutation: Bandit-Guided Evolution on Workflow Graphs
arXiv · Oct 7, 2026
ThinkFlow: Self-Evolving Probabilistic Latent Memory for Lifelong Conversational Agents
arXiv · Sep 16, 2026