Reproduce “Bian Que: An Agentic Framework with Flexible Skill Arrangement for Online System Operations” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Bochao Liu, Zhipeng Qian, Yang Zhao, Xinyuan Jiang, Zihan Liang, Yufei Ma, Junpeng Zhuang, Ben Chen and 5 more
Kuaishou Technology developed BIAN QUE, an agentic framework designed for online system operations that automates context selection and dynamically updates operational "Skills" using large language models. The system reduced fired alerts by 75% and non-actionable alerts by 95% in production, achieving 80% root-cause analysis accuracy and compressing Mean Time To Resolution by over 50%.
Key insights
- The primary bottleneck in applying LLMs to AIOps is not reasoning capacity but the dynamic orchestration of relevant data and knowledge for specific operational contexts.
- A unified feedback loop driving parallel pathways for knowledge distillation and Skill refinement leads to a more robust, adaptive, and self-evolving autonomous agent system.
- Decomposing diverse operational workflows into canonical agent types combined with flexible, dynamically managed "Skills" provides comprehensive and lightweight coverage of daily maintenance tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.