COBRA-Skills: Contextual Bandit-Guided Evolution for Agent Skill Optimization

Hugging Face Daily Papers Papers
agent-skill-optimization contextual-bandits evolutionary-algorithms ai-research llm-optimization neural-predictor budget-allocation

Summary

COBRA-Skills introduces a method for optimizing agent skills using contextual bandits and evolutionary operators, achieving 55-58% lower optimization costs across multiple AI agent benchmarks.

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 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.
Original Article
View Cached Full Text

Cached at: 09/14/26, 02:33 AM

Paper page - COBRA-Skills: Contextual Bandit-Guided Evolution for Agent Skill Optimization

Source: https://huggingface.co/papers/2609.11682

https://huggingface.co/papers/2609.11682#cobra-skills-contextual-bandits-for-efficient-agent-skill-optimization-%F0%9F%9A%80COBRA-Skills: Contextual Bandits for Efficient Agent Skill Optimization πŸš€

How can we optimize Agent Skills without repeatedly spending large amounts of computation on weak candidates and costly LLM-based refinement?

COBRA-Skillstreats skill optimization as a sequential budget-allocation problem:

Which skill is most worth evaluating next?

Instead of exhaustively trying and refining candidate skills, COBRA-Skills maintains a population of skills and uses:

  • a neural predictor to estimate skill utility,
  • LinearUCBto balance exploration and exploitation,
  • target-agent execution feedback to update the bandit,
  • scheduled evolution operators (regeneration, rollout mutation, and crossover) to evolve the skill population.

https://huggingface.co/papers/2609.11682#resultsResults

We evaluate COBRA-Skills on6 diverse agent benchmarks Γ— 3 target models, covering search QA, spreadsheets, document understanding, mathematical reasoning, social reasoning, and embodied tasks.

COBRA-Skills achieves:

  • πŸ† the best average performance among the compared methods,
  • πŸ’°~55–58% lower optimization costthan SkillOpt,
  • πŸ“Š only50 optimization examples per benchmark,
  • πŸ”§ consistent effectiveness underCodex and Claude Codeharnesses,
  • πŸ€– strong performance even when thetarget model itselfgenerates and refines skills.

An interesting finding is that the cost reduction does not mainly come from reducing target-agent executions. A large part comes from avoiding repeated LLM-based trajectory analysis and skill rewriting.

πŸ“„ Paper:https://arxiv.org/abs/2609.11682 πŸ’» Code:https://github.com/Jerry-LuP/COBRA-Skills

Feedback and discussions are very welcome!

Similar Articles

SkillOpt: Executive Strategy for Self-Evolving Agent Skills

Hugging Face Daily Papers

SkillOpt introduces a systematic text-space optimizer for agent skills that trains skills as external agent state with stable updates and zero deployment inference overhead, achieving superior performance across multiple benchmarks and execution environments.

Bayesian-Agent: Posterior-Guided Skill Evolution for LLM Agent Harnesses

Hugging Face Daily Papers

Bayesian-Agent presents a framework that treats reusable skills and SOPs as hypotheses, using Bayesian inference to guide agent behavior and improve task performance through posterior-guided harness optimization. It achieves significant improvements on multiple benchmarks with deepseek-v4-flash.

SkillOpt-Lite: Better and Faster Agent Self-evolution via One Line of Vibe

Hugging Face Daily Papers

SkillOpt-Lite proposes a minimal viable pipeline for skill optimization in autonomous agents, achieving better and faster self-evolution by treating all components as editable code and integrating into production coding agents. It formalizes skill optimization via Zeroth-Order optimization and outperforms prior methods on benchmarks.