When speaking about CodeRabbit, I often emphasize the importance of code review guidelines. Code review guidelines represent a shared understanding of how code reviews should be conducted within a team or organization.
Without such guidelines, the following issues tend to arise:
- It is unclear what perspectives to use when reviewing
- Review perspectives vary from person to person
- Review criteria change depending on mood or circumstances
To prevent this kind of inconsistency, code review guidelines are essential.
A helpful resource when creating code review guidelines is Code Review Developer Guide | google-eng-practices-ja. This is a guideline published by Google and serves as the official explanation of Google’s code review process and policies. Note that the original repository has been archived, so it may not reflect the latest practices.
I usually recommend reading this guide and adapting it to your own organization, but creating guidelines from scratch is time-consuming. To address this, I built a web application that allows you to create your own code review guidelines in a wizard-based format, using Google’s guidelines as a foundation.
About Code Review Guideline Creator
The technologies used in Code Review Guideline Creator are as follows:
- Cloudflare Workers
- React Router
- Tailwind CSS
The source code is open source and released under the MIT License.
goofmint/review-guideline-creator
Types of Guidelines You Can Create
With Code Review Guideline Creator, you can create the following two types of guidelines:
- For reviewers (those who review code)
- For reviewees (those whose code is reviewed)
The guidelines clarify what reviewers should focus on during reviews and what reviewees should pay attention to before submitting a pull request.
About the Wizard
The process is simple: you answer a series of questions. Default values are provided wherever possible.
The questions are defined in Markdown files, which are loaded and reflected in the wizard.
By editing these files, you can customize the wizard itself. Note that there are around 60 questions, so the process is relatively extensive.
Data Storage
No data is stored on the server. Intermediate progress is saved in SessionStorage and used to restore state on page reload. There is no authentication mechanism.
Supported Languages
The default language is English, and Japanese is also supported. Additional languages can be added by providing new wizard files.
AI Code Review
Google’s code review guidelines do not cover AI-assisted code reviews, so I added dedicated sections for this. You can define how AI code review tools (such as CodeRabbit) should be handled for both reviewers and reviewees.
About the Generated Output
The completed guidelines can be downloaded as Markdown or PDF files. However, since the output directly reflects the input answers, it may feel somewhat hard to read as-is.
To address this, the app also outputs a prompt for LLMs. By copying this prompt and pasting it into tools like ChatGPT or Claude, you can generate a cleaner and more readable coding guideline.
Summary
Having code review guidelines helps reduce stress for those new to code reviews and helps pull request authors prepare mentally before submitting their work.
To improve team productivity and psychological safety, I encourage you to give it a try.
Top comments (0)