GitHub Ships Stacked Pull Requests, Built for the Age of AI-Generated Code

GitHub put Stacked Pull Requests into public preview on July 30, 2026. A large change becomes an ordered chain of PRs, each one reviewable as its own focused layer, with automatic rebasing when a lower layer merges.

How it actually works

Each PR in a stack targets the layer below it. A reviewer opens any single PR and sees only that layer's diff. When a lower layer merges, everything above it automatically rebases and retargets, and merging the topmost ready PR merges every unmerged layer below it in one operation. It ships with a CLI extension, a stack-map visualization, and works from github.com, the GitHub CLI, the mobile app, and with Copilot coding agents. Merge queue support for stacks is rolling out over the following weeks.

Why now: agents write more code than review can absorb

TED's CTO, quoted in coverage of the launch, ties the feature directly to a newer pressure: AI-assisted coding started producing more code than the team's existing review process could absorb, and stacked PRs tightened that feedback loop by letting reviewers work on smaller layers in parallel instead of blocking on one giant diff. Early adopters named in GitHub's own rollout include Vercel's Next.js team and TED.

Why a build studio cares

This is the exact bottleneck a studio running Claude Code or Copilot at volume hits first: the agent produces a big, correct diff faster than a human can review it in one sitting. Stacked PRs is GitHub betting the fix is better review ergonomics, not slower agents, which matches how we'd rather solve it too.

Next step: read the GitHub Changelog announcement or itbrief's coverage with customer quotes. If your review pipeline is buckling under AI-generated volume, write to us at hello@gattyworks.com.