Most PR descriptions leave out the one thing a reviewer actually needs: why the change was made. And AI review bots that live natively inside GitHub tend to solve that with noise instead, dropping comments a reviewer then has to sort through to find the two that matter.

What is GitKraken Code Review?

GitKraken Code Review, released as part of June’s set of updates, is a new way to read a pull request. It runs on gitkraken.dev and pairs an AI-generated summary of a PR’s intent with semantic diffing, so a reviewer sees what actually changed and why, instead of scrolling through every line.

Comment without leaving the review

Reviewers can comment directly on a line, @mention teammates, and react, the same shorthand developers already use on GitHub, just inside a layout built specifically for review instead of general-purpose issue tracking.

Shipped outside the usual release cycle

Because GitKraken Code Review runs on gitkraken.dev rather than inside a downloadable client, the team can iterate on it without waiting for a full GitKraken Desktop release. Features like line commenting can ship as soon as they are ready.

Bonus: GitBench, a report card for AI models on Git tasks

Also worth knowing about: GitBench, a benchmark suite built by developer advocate Chris Griffing, now live at gitbench.gitkraken.com. GitBench measures how different AI models perform on specific Git tasks, like identifying which commits in a messy history should be squashed together. Results vary more than most developers would expect. In early testing, one model handled commit squashing well in structured JSON output but performed worse with plain text output, and results got worse, not better, as its reasoning effort was pushed higher. That is the kind of detail no general-purpose benchmark surfaces, and it is a useful, evidence-based way to decide which model to reach for on a given task rather than guessing.

Frequently asked questions

Where can I access GitKraken Code Review? Go to gitkraken.com and select GitKraken Code Review.

What makes GitKraken Code Review different from GitHub’s built-in AI review comments? It uses semantic diffing and one AI-generated summary of intent per PR, rather than a stream of individual line comments a reviewer has to sort through.

What is GitBench? A benchmark suite that tests how well different AI models handle specific Git tasks, such as commit squashing, so developers can pick a model based on evidence instead of assumption.

Try GitKraken Code Review at gitkraken.dev, and check gitbench.gitkraken.com to see how the model in your current workflow actually performs on Git-specific tasks.