Show HN: 深度代码研究 – AI 扫描 10 个相似代码库，用于审查你的代码
1 分•作者: WindChimeRan•6 个月前
Deep Code Research 是一款命令行工具，可以自动查找与你的代码库相似的仓库，分析它们处理问题的方式差异，并生成对比报告。
它不会给出像“添加错误处理”这样的泛泛建议，而是会确切地展示相关项目是如何做的，以及你的代码与之有何不同。 每次发现都包含来自你的代码库和参考代码库的并排代码片段，并附有具体的文件路径和行号。
该工具使用多代理架构：一个主代理通过 GitHub 搜索发现相关的代码库，然后生成并行子代理来分析每个代码库。结果被整合到一个优先排序的发现列表中。
我构建这个工具是因为代码“文献综述”非常耗时。在构建某些东西之前，你应该调查类似的项目，以学习模式并避免已知的陷阱。但是，阅读 10 多个代码库需要数小时。这个工具将这个过程自动化了。
[https://github.com/WindChimeRan/deep_code_research](https://github.com/WindChimeRan/deep_code_research)
查看原文
Deep Code Research is a CLI tool that automatically finds repositories similar to yours, analyzes how they handle things differently, and produces a comparative report.<p>Instead of generic advice like "add error handling," it shows you exactly what related projects do and where your code differs. Each finding includes side-by-side code snippets from both your repo and the reference repo, with specific file paths and line numbers.<p>The tool uses a multi-agent architecture: a main agent discovers related repos via GitHub search, then spawns parallel sub-agents to analyze each one. Results are synthesized into a prioritized list of findings.<p>I built this because code "literature review" is time-consuming. Before building something, you should survey similar projects to learn patterns and avoid known pitfalls. But reading through 10+ repos takes hours. This automates that process.<p><a href="https://github.com/WindChimeRan/deep_code_research" rel="nofollow">https://github.com/WindChimeRan/deep_code_research</a>