Table of Contents

1. Understand the Context
2. Review Small, Frequent PRs
3. Focus on Functionality First
4. Check for Consistency
5. Prioritize Security and Performance
6. Test the Changes Locally
7. Provide Constructive Feedback
8. Use Automated Tools
9. Encourage Discussion
10. Balance Thoroughness with Efficiency
11. AI Coding Tools for PR Reviews

---

1. Understand the Context

Before diving into the code, take a moment to understand the purpose of the PR. Read the description carefully, and if available, check out related tickets or documentation. Knowing the context helps you focus on the important aspects of the code.

2. Review Small, Frequent PRs

Encourage submitting smaller, more frequent PRs rather than large, monolithic ones. Smaller PRs are easier to review, less prone to errors, and allow for quicker feedback and iteration.

3. Focus on Functionality First

Start by reviewing the functionality. Does the code achieve the intended outcome? Ensure that the logic makes sense and that the feature works as described before delving into the finer details.

4. Check for Consistency

Look for consistency in code style, naming conventions, and architecture. Consistent code is easier to read, maintain, and scale. Ensure that the changes align with the existing codebase's standards.

5. Prioritize Security and Performance

Evaluate the code for potential security vulnerabilities and performance bottlenecks. Consider how the changes might impact the overall system's security and efficiency.

6. Test the Changes Locally

If possible, pull the branch and test the changes locally. Running the code yourself can help you spot issues that aren't immediately obvious from the code alone, such as unexpected side effects or integration problems.

7. Provide Constructive Feedback

When pointing out issues or suggesting changes, be constructive and specific. Offer explanations and alternatives rather than just highlighting problems. This fosters a positive, collaborative environment.

8. Use Automated Tools for AI Code Review

Leverage automated tools to catch common issues such as syntax errors, formatting problems, and simple bugs. Tools like linters, static analysis tools, and automated tests can save time and ensure consistency. AI-powered tools like Fine are great options for catching such issues automatically, giving you more time to focus on functionality and design.

9. Encourage Discussion

Use the PR review process as an opportunity to discuss the code with the author and other team members. Engage in meaningful conversations about design decisions, potential improvements, and alternative approaches.

10. Balance Thoroughness with Efficiency

Aim to be thorough in your review, but also be mindful of the time it takes. Focus on critical areas first and avoid getting bogged down by minor issues that can be addressed in subsequent PRs. Remember that the goal is to improve the codebase, not to achieve perfection in a single review.

AI Coding Tools for PR Reviews

By connecting AI tools to your tech stack, the process of reviewing PRs becomes much easier. To start off with, have the AI create a summary of every PR before you review it. That way, no matter how long or short the PR is, you’ll know what it’s meant to do and how, before you begin.

Next, you can have the AI Coding tool review the PR before you do. This can help on both ends of the spectrum: On the one hand, it will catch basic fixes, allowing the developer to fix them before your final review, saving your time as a manager. On the other hand, it will identify edge cases that you may not have considered, thereby improving the quality of your code.

Fine is an AI Coding tool that not only reviews and summarizes PRs when directly asked to, but also offers automated workflows. Set it up so that any time a new PR is created (in your codebase or a specific repository), it reviews and summarizes it, sending you a Slack message when it’s done and ready for your sign-off. Here’s how it works.