一句话：提出 LintCFG，用 DSL 结构化表示编码规范，并通过 LLM 编译流水线自动生成 linter 配置。
编码规范对代码质量至关重要，但要把规范落实为 linter 配置仍需人工完成，过程繁琐、易出错且难以维护。本文提出 LintCFG，一个从自然语言编码规范自动生成 linter 配置的工具：它先用一种领域特定语言（DSL）结构化地表示编码规则，再通过基于 LLM 的编译流水线把规则编译为面向具体工具的目标配置。作者在 Checkstyle（Java）与 ESLint（JavaScript）上进行了评估，结果显示该工具准确率超过 70%，精确率是现有基于 LLM 的基线的两倍以上，表明其有效性与跨语言、跨工具的通用性。论文还公开了源码、SKILL 及演示视频。
Coding standards are essential for code quality, but configuring linters to enforce them remains manual, error-prone, and hard to maintain. We present LintCFG, a tool that automates linter configuration generation from natural language coding standards. It leverages a domain-specific language (DSL) to structurally represent coding rules and an LLM-based compilation pipeline to produce tool-specific configurations. Our evaluation shows that the tool achieves over 70% accuracy and more than double the precision of existing LLM-based baselines for Checkstyle (Java) and ESLint (JavaScript), confirming its effectiveness and generality. Source code is available at https://github.com/crowdedplus/gen-linter-config, SKILL is available at https://github.com/idiomaticrefactoring/LintConfig/, and a video demonstration is available at https://www.youtube.com/watch?v=95wk0jsogAY.
@inproceedings{paperbot12329,
title = {LintCFG: DSL-Driven Linter Configuration Generation for Coding Standards},
author = {Yi Tang and Zejun Zhang and Zhenchang Xing and Minxue Pan and Tian Zhang and Xuandong Li},
booktitle = {ASE},
year = {2026},
doi = {10.1145/3832783.3834632}
}