proomt

Search

Search posts, papers, and topics

All posts

Hugging Face Daily PapersPingchen Lu, Xiangyi Wang, Xiang Li1 min readpaperadvanced

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

Summary

COBRA‑Skills uses a contextual‑bandit loop to selectively evaluate and evolve LLM agent skills, achieving better performance with roughly half the evaluation cost of prior methods. The framework works with limited examples and remains robust across different agent setups and even when the target model creates its own skills.

  • COBRA‑Skills frames skill optimization as a budgeted sequential problem, using a contextual bandit to prioritize which candidate skills to evaluate next.
  • The method cuts evaluation cost by roughly 55‑58% compared to prior SkillOpt while achieving higher average performance across six benchmarks.
  • It requires only about 50 unique optimization examples per benchmark, making it data‑efficient for LLM‑agent development.
  • The skill population is continuously refined from execution feedback, keeping the candidate set dynamic and evidence‑grounded.

LLM‑agent developers who need to optimize reusable skills efficiently will benefit from the cost‑effective, data‑light approach described.

6/10

Related reading

  1. Reflect, Revise, Reuse: Training-Free Skill Evolution for GUI Agents

    EvoSkill‑GUI lets GUI agents revise their procedural skills on‑the‑fly without extra training by using a reflect‑revise‑reuse loop that edits skill packages during execution. The approach yields up to +16.2% improvement on MobileWorld and similar gains on AndroidWorld and OSWorld, and the evolved skills transfer to related tasks.

    Hugging Face Daily Papersarxiv.org1 minpaper
  2. Grounded Skill Synthesis from Code at Scale for Agentic Intelligence

    Code2Skill is an automated pipeline that mines popular GitHub repositories to extract verifiable, implementation‑anchored procedural “skills”. It builds a bank of ~1 M skill records (atomic ops, workflows, patterns) with provenance metadata, verifies each via blind reconstruction, and shows that augmenting LLM‑based agents with these skills yields an average 11.7% performance lift across 72 proto…

    Hugging Face Daily Papersarxiv.org1 minpaper