Alden ranks the PRs waiting on you and writes each one a briefing with the callers and checks the diff leaves out. Every place you mark useful or not useful teaches it a little more about how you review.
GitHub shows you the lines that moved. Alden reads your local clone and follows them out to every call site, so you see that three of seven callers don't catch the new error before you approve.
TypeScript · JavaScript · Python · Go · PHPMark a place useful and Alden gives that kind of flag more weight next time. Mark it not useful and it backs off. Open some PRs first and they rank higher. Everything it learns is listed, with where it came from, and you can edit or forget any of it.
Alden works from your local clone and your own GitHub sign-in. You pick the model and decide how much it can spend.
We're letting in a few teams at a time. Expect a short call each week while we tune it to how you work.
Ask to join the beta