GitHub Ships PR Limits to Fight the 3.6x AI-Driven Code Flood
GitHub now lets maintainers cap how many open PRs any outside contributor can hold at once, with a bypass list for trusted contributors — a direct response to AI-fueled PR spam.
- New feature: GitHub now lets maintainers cap open PRs per external contributor in a repository, available now in Settings.
- AI-driven surge: Monthly merged PRs on GitHub jumped from ~25M (Jan 2023) to 90M+ today, overwhelming maintainer review capacity.
- Bypass list: Trusted contributors can be exempted from the cap without being granted full write access to the repo.
- Draft PRs excluded: Draft pull requests don't count toward the limit, so contributors can still stage work-in-progress freely.
- Roadmap includes: PR archiving, issue limits, automated trust signals (e.g. prior merged PR, account age), and cross-repository rate limiting.
- Early feedback positive: AutoGPT and Homebrew maintainers report the feature is already making review queues more manageable.
Open source maintainers have been quietly drowning. The number of pull requests merged on GitHub has gone from roughly 25 million per month in early 2023 to over 90 million today , a 3.6x surge driven in large part by AI-assisted code generation. The cost to open a PR has collapsed. The cost to review one hasn't moved. GitHub is now shipping tooling to address that asymmetry head-on.
Pull request limits are a new repository-level setting that lets maintainers cap how many open PRs a contributor without write access can hold at any given time. Hit the cap, and you have to close or merge an existing PR before opening a new one. It's a simple constraint with a surprisingly strong behavioral effect.
How it actually works
Maintainers of open source repositories are dealing with an ever-growing volume of pull requests, including repeated low-quality or drive-by contributions that can slow triage and overwhelm review queues. The new setting lets you define a maximum number of open PRs that users without write access may have open in your repository at one time. The limit applies to all external contributors by default, including AI agents , pull requests opened by Copilot or another AI agent count toward the limit.

A few important nuances in the mechanics:
- If a user without write access reaches the limit, they'll need to close or merge an existing pull request before opening another one.
- You can add trusted contributors to a bypass list, allowing them to exceed the limit without being granted full collaborator access.
- Draft pull requests do not count towards a user's limit. So contributors can still stage work-in-progress without burning a slot.
- These new pull request limits are persistent and configurable , unlike GitHub's existing interaction limits, which are temporary cooldowns.
The behavioral shift is the real feature
The mechanical cap is straightforward, but the deeper effect is on contributor behavior before a PR even lands in the queue. When anyone can open a pull request in seconds, a polished change and a rough draft look the same in the queue. But when only a few pull requests can be open at once, a contributor must be selective and prioritize which contributions they want reviewed. That first judgment call happens before the pull request reaches you, and a smaller pool makes good work easier to spot.
Early adopters are already seeing this play out. Nicholas Tindle from AutoGPT noted that "it's helped us want to review pull requests again. Knowing that someone hasn't just opened 5-10 pull requests that are slop makes it much easier to want to look. Going forward we expect it to help us manage our backlog." Mike McQuaid from Homebrew echoed the sentiment, pointing out that AI had accelerated the problem and that the new limits allow outside contribution to continue while gating volume to a level maintainers can actually handle.
What it doesn't solve (yet)
The current implementation has a clear blind spot: it's scoped to a single repository. A per-repository cap helps with repeated activity in one project, but it does nothing when someone opens pull requests across hundreds of repositories at once. GitHub is exploring ways to catch contributors who spray pull requests across multiple repositories, whether through trust signals, rate limiting, or other cross-repository controls.
The bypass list is also manual for now. You have to curate it yourself, which adds overhead for large projects with many trusted contributors.
What's coming next
GitHub has outlined a roadmap of follow-on features building on the same foundation:
- Archiving pull requests (shipping soon): Admins will be able to hide low-quality or spammy PRs from the main view without permanently deleting them , important for organizations with legal or compliance requirements around record retention.
- Issue limits (in development): The same per-user cap mechanism will extend to issues, with a bypass list and an option to restrict issue creation to collaborators only.
- Smarter bypass signals (up next): Instead of curating a bypass list by hand, maintainers could let contributors clear a limit automatically based on real signals: a previously merged pull request in the repo, account age, or organization membership. That frees maintainers from curating lists by hand.
Context: this is part of a broader maintainer push
This isn't GitHub's first move on the maintainer burnout problem. In February 2026, GitHub added features that allow users to completely disable pull requests or restrict submissions to collaborators only. You can now turn off pull requests entirely from your repository's Settings, just like wikis or issues. When disabled, the pull requests tab is not visible, meaning no one can see existing pull requests or open new ones , particularly useful for mirror repositories or read-only codebases. Pull request limits sit between that nuclear option and doing nothing, giving maintainers a middle ground that keeps the door open for new contributors while managing the volume.
This update can help reduce noise in busy repositories, lower unnecessary review and CI overhead, and make it easier for maintainers to focus on high-signal contributions from their community. The feature is available now in repository settings. It's free, with no plan restrictions mentioned. You can enable it under Settings > Moderation options > Interaction limits > Pull request limits, and share feedback in the GitHub community discussion.