The AI Code Review

How it works

When you have an AI code review available, we read your whole solution and review it the way an experienced developer would. Rather than flagging isolated rule violations, it weighs your code in context: how it's structured, how readable it is, how accessible it is, and how you built it.

You get an overall score out of 10, a short summary, and a breakdown across six dimensions, each with its own score and findings.

Pro members get an AI code review on every submission. Free members get one a month. The Solution Reports Overview explains how this sits alongside the automated checks everyone gets.

Categories and dimensions

The six dimensions are grouped into three categories.

Code Quality

• Best Practices: readability, naming, keeping code DRY, error handling, modern syntax, responsive design, and general code hygiene.
• Architecture: how you structure components, manage state, separate concerns, and apply your framework's patterns. Applies to Intermediate challenges and up.
• File Organization: folder structure, naming conventions, and how well your project layout follows its framework's conventions. Applies to Junior challenges and up.
• Testing: whether your tests make meaningful assertions, cover edge cases, and are well organized. Applies to Advanced challenges and up.

Design & UX

• Accessibility: semantic HTML, ARIA, keyboard support, focus management, color contrast, alt text, and forms, judged across your whole solution.

Documentation

• Commit Quality: your commit messages, how you split work across commits, and repository hygiene. The Commit Quality guide covers this dimension in detail.

Reading your findings

Each dimension's findings are grouped into three types:

• Strengths: things you did well and should keep doing.
• Suggestions: improvements that would raise the quality of your code.
• Needs attention: clearer problems worth fixing.

Findings point to the specific file and line they apply to, and carry tags for the technology and skill involved, so you can connect each piece of feedback to the exact part of your code.

Scores and rating bands

Every score, from the overall down to each dimension, comes with a rating band so you can read it at a glance: Needs Work, Fair, Solid, Great, and Exceptional.

A lower score shows where the most useful practice is right now. Even strong solutions get feedback, so treat the review as a guide for your next step rather than a final grade.

How difficulty changes the review

We calibrate the review to the challenge difficulty, and this works in two ways.

First, which dimensions apply. Simpler challenges are reviewed on fewer dimensions, and more come into play as challenges get harder:

• Newbie: Best Practices, Accessibility, Commit Quality
• Junior: adds File Organization
• Intermediate: adds Architecture
• Advanced and Guru: adds Testing

Dimensions that don't apply to a challenge show as locked rather than scored, and they don't count toward your overall score.

Second, the bar within each dimension. We expect more polish from an Advanced solution than a Newbie one, so the same code can read as solid at one level and as a gap at another.

Improving your score

• Start with anything under "Needs attention", then work through the suggestions.
• Use your strengths to see what's already working, so you keep doing it.
• Regenerate the report after making changes to see your updated scores.

Across several challenges, look for patterns. A dimension that's consistently lower than the rest is the clearest signal of where to focus next.

Want to go deeper on a specific dimension? Start with the Commit Quality guide.