If you've ever opened a PR with 47 changed files and a diff so long GitHub just gives up and shows you "Load Diff" seventeen times, this one's for ...
For further actions, you may consider blocking this person and/or reporting abuse
Agents made the mega PR problem worse before GitHub shipped this. A coding agent will happily touch forty files in one pass and the summary will make it sound coherent, then you open the diff and realize three of those files are unrelated cleanup it decided to help with. Stacking forces the review unit back down to something a human can actually finish. The part I care about most is that each PR only shows the delta it introduces, because that is the only surface where you catch the silent test rewrite or the path you never asked for.
Yess
This is actually a really practical way to look at stacked PRs. The part about smaller diffs leading to better reviews makes a lot of sense, especially now that AI agents can generate huge changes so quickly. I also like that GitHub is making the dependency between PRs visible instead of forcing developers to keep track of it themselves. Definitely something Iβd like to try on a real project.
Yes yes, you must try!
Github just invented commits
xD
good writeup. the before/after numbers are what convinced me to try it.
Thanks bud
Isn't this why, they introduce Stacked PRs?
This is a really useful breakdown of stacked PRs. π The distinction between Gitβs branches and GitHubβs new first-class stack experience is especially interesting, and the real terminal walkthrough makes the concept much easier to understand.
I also think the connection to AI-assisted development is spot onβsmaller, dependency-aware PRs can make AI-generated changes much easier for humans to review and maintain.
Great practical write-up! This would also be a valuable topic to share with the developer community on CodeCan.net.