Code Reviews
Get deep, LLM-powered code reviews on every commit of every pull request
Our code review AI can run and debug code, catching issues that traditional static code analysis tools can’t. Internally we’ve added a list of obvious best practices that Ellipsis should catch, but since every team is different, we’ve made it easy to add your own rules.
If it finds issues, it will leave comments on the PR. If it doesn’t find any issues, it will leave a “Looks good to me!” comment.
Requesting a review
By default, Ellipsis will automatically review any pull request that is published as or opened in the “ready to review” state. Draft pull requests aren’t automatically reviewed by default (though you can enable this in the config), but you can always ask Ellipsis to review them leaving a comment like @ellipsis review this
. Heads up: there’s no autocomplete.
You have Ellipsis only comment if it finds a problem with the “quiet mode” setting. You can also disable automatic reviews entirely.
Enforcing a Style Guide
You can add your own review rules in either the UI or your ellipsis.yaml
to customize the review behavior.
For example, you might use Ellipsis to make sure your team NEVER logs sensitive information such as user email address. If a developer opens a pull request that does that, Ellipsis will catch it and leave a comment:
Feedback
Ellipsis can learn from your feedback in order to improve. Simply add 👍/👎 emojis to rate review comments.
You can also optionally reply in the thread to be specific (no need to tag Ellipsis) - but remember to give 👍/👎 too. Ellipsis will take this into account in all future reviews (on this PR and on others).
Hide Old Reviews
On long-running PRs, lots of reviews can accumulate and cause clutter. While you can individually click on reviews to hide them, you can also have Ellipsis do it in bulk. If you ask Ellipsis to “hide your old comments,” it’ll hide all its previous reviews to give you a clean slate.
If you’d like to reduce clutter futher, consider quiet mode.
Configuration
x