Graphite Reviewer is now Diamond

Enhancing code quality with AI

Greg Foster
Greg Foster
Graphite software engineer
Try Graphite

As codebases grow and release cycles tighten, maintaining high code quality is becoming both more critical and more difficult. Manual reviews take time. Linters catch only the basics. Developers need support systems that help uphold standards without slowing down development. In this guide, we’ll explore how to use AI to assess quality of code in practical ways.

There are several areas where AI is already proving useful for improving software quality:

AI-powered code reviewers go beyond simple diff comparisons. They analyze logic, detect potential bugs, and offer suggestions that take context into account. Unlike basic linters, these systems can flag things like:

  • Unintended logic (e.g., inverted conditionals)
  • Unclear or incorrect documentation
  • Edge cases that could break in production
  • Performance issues based on known patterns

Example: a developer writes a shell script using readlink -f, which fails on macOS. An AI reviewer can catch this and suggest a cross-platform solution, which is something traditional tooling might miss.

Most teams have guidelines, but enforcement is often inconsistent. AI tools can help:

  • Enforce naming conventions and design patterns
  • Flag TODOs without tracking tickets
  • Prevent secrets from being committed
  • Recommend best practices (e.g., prefer const over let)

This doesn’t just keep the code clean—it also makes it easier to onboard new team members who can rely on AI-powered feedback for real-time learning.

Some AI models can identify potential bugs before they cause failures. Think of them as an extra pre-commit sanity check.

For example, AI can flag suspicious regex usage, brittle logic, or recursive calls that might lead to stack overflow. In some cases, they can even propose refactored alternatives that are more readable or performant.

​Integrating AI into your development workflow can significantly enhance code quality by automating reviews, detecting vulnerabilities, and ensuring adherence to best practices. Once you're ready to bring AI into your workflow, the next step is picking tools that align with your development stack and team needs.

Here are a few to consider:

Diamond is an AI-powered code review tool that integrates seamlessly with GitHub. Its key features include:

  • Automated code reviews: Diamond automatically scans pull requests for bugs, logical errors, and other technical pitfalls, providing immediate, actionable feedback.
  • Codebase awareness: It understands the entire codebase context, ensuring that feedback is relevant and accurate
  • Customizable rules: Teams can define and enforce custom patterns with AI prompts and regular expressions, tailoring the tool to their specific coding standards.

By automating the detection of issues and providing context-aware feedback, Diamond helps maintain high code quality and reduces the manual effort required during code reviews.

Codacy offers automated code reviews with a focus on static analysis and code quality metrics. Its features include:

  • Language-specific checks: Supports various programming languages, providing tailored feedback and ensuring adherence to language-specific best practices.
  • Integration into CI/CD pipelines: Easily integrates into continuous integration and continuous delivery pipelines, allowing for automated checks at every stage of development.

Codacy's automated reviews and metrics help maintain consistent code quality across projects and teams.

DeepCode, now part of Snyk, utilizes machine learning to identify code issues, particularly security vulnerabilities. Its capabilities include:

  • Real-time analysis: Provides immediate feedback as code is written, allowing developers to address issues promptly.
  • Security focus: Specializes in detecting security vulnerabilities, helping teams proactively address potential threats.

By focusing on security and leveraging machine learning, DeepCode assists in maintaining both code quality and security.

Amazon Q Developer is an AI-driven tool that enhances various aspects of the development process. Its features include:

  • Automated code reviews: Detects and resolves code quality issues such as logical errors, anti-patterns, code duplication, and security vulnerabilities.
  • Automated documentation: Generates in-depth documentation within source code, including data flow diagrams, facilitating faster understanding of codebases.

Amazon Q Developer's features support developers in writing high-quality, secure, and well-documented code.

Integrating these AI tools into your development workflow can lead to more efficient processes, higher code quality, and reduced manual effort in code reviews and maintenance.

Integrating AI into your development flow isn't just a technical shift—it’s a cultural one. The goal isn’t to offload all responsibility to a machine, but to make high-quality code the default.

Here are a few best practices:

  • Use AI for early feedback, before PRs are even opened
  • Pair AI feedback with custom team rules and templates
  • Regularly audit what the AI flags to tune thresholds and reduce false positives
  • Encourage developers to treat AI suggestions as input, not mandates

AI isn't a replacement for developers, but it can help teams write better code, faster. By combining smart automation with human insight, developers can focus more on solving meaningful problems—and less on catching off-by-one errors or enforcing line length rules. Whether you use Diamond, Codacy, or another platform, the end goal is the same: higher quality code, with less friction and more confidence.

Git inspired
Graphite's CLI and VS Code extension make working with Git effortless.
Learn more

Built for the world's fastest engineering teams, now available for everyone