Cursor Origin Guide

The table separates what the beta ships from what Cursor has only shown or promised. Stacked pull requests, Code Tours and a merge queue have shipped since launch, while the June demo's intent-aware conflict resolution is still not documented.

Cursor's release notes and announcements say what changed. They don't say how a team should act on it. The steps below are our suggested order, and the next section lists what to watch out for along the way.

1. 1Sync one non-critical repo from GitHub at cursor.com/codebase. Keep GitHub as the source of truth while you try review and merge on Origin.
2. 2List what you would need to prove before moving source hosting: identity, audit, branch protection, CI, reviews and data policy.
3. 3Compare Origin with GitHub on review and agent workload, not only on repository hosting.
4. 4Recheck cursor.com/docs/origin and the changelog as Cursor adds agent-native features.

Each item below is either a fact about the release, such as its beta status or plan access, or our advice for using it safely. Plan around each one, and recheck the facts when Cursor ships an update.

• Don't treat an early beta as a GitHub replacement. Origin's own docs say its permission and protection controls are still being redesigned.
• Get security, legal and platform review before you move source hosting, because the move changes who holds your code.
• Don't assume your GitHub integrations, checks or protection rules work the same way in Origin. Actions workflows and secrets do not sync.

Cursor announced Origin at its Compile event in June 2026 (press coverage gives June 16) and shipped it in early beta on August 17, 2026. Cursor describes Origin as a git forge designed for the volume of code that agents now produce, and coverage of its Compile demo reported thousands of parallel agents pushing on the order of 22.6 commits per second. The beta ships the essentials, such as repos, pull requests, code browsing and GitHub sync, and Cursor writes a Code Tour for every pull request to walk a reviewer through the change.

This page answers the main question: what Origin is and whether to care. The Cursor Origin deep-dive pillar covers each sub-question in more depth: why git hosting is being rebuilt for agents, how stacked diffs change review and what reviewing agent code looks like. It also covers how parallel agents avoid merge pile-ups, what a GitHub migration would involve and who holds your code.

It points to review before hosting. Cursor signed a definitive agreement to acquire Graphite, a company built on stacked pull requests and a merge queue, on December 19, 2025. With that team inside Cursor, you could expect Origin to ship the review layer first, because Graphite has built that layer once already.

The early beta ran the other way, at least at first. It opened on August 17, 2026, with hosting, pull requests, GitHub sync and three apps (Vercel, Depot and Buildkite), and without the review layer the team is known for. That layer has been arriving since: stacked pull requests, a Code Tour that Cursor writes for every pull request and, in Cursor's October release notes, a merge queue.

The ecosystem has grown too. Cursor's docs now list a marketplace of Origin apps, open to submissions, with CI, deployment and code review apps among others. A team can also build a private app that calls the Origin API. Marketplace apps work only on repos hosted on Origin, not on repos mirrored from GitHub.

The lineage still tells you what to evaluate. The review model is the piece with a track record, so judge Origin on review and merge, and treat repository hosting as a baseline it meets or doesn't. Several vendors solved hosting years ago. Nobody has settled review at agent volume, and the beta doesn't settle it either.

One caveat: Cursor's docs list what the beta does today and publish no roadmap. So every sentence about what Origin will be good at, including the ones above, is a read on the team. That read is worth something as the only evidence available, but it doesn't justify a migration plan.

Probably not yet, because Origin targets a problem a small team doesn't have. It aims at review and merge throughput when agents produce a high volume of change. With six engineers and a couple of agents running, that is not your binding constraint. Your constraint is more likely that review is one person's spare attention on a Thursday.

What a small team can take from Origin today is the practice. Reviewing small dependent changes in sequence is a way of working that predates any particular tool, and it is available on the host you already pay for. That makes it a cheap experiment: run it for two weeks and see whether your reviews get faster or you just get more of them.

At a few hundred engineers, the calculation changes, because the problem Origin targets shows up in your own numbers:

• Time from review-ready to merged.
• How many changes are waiting on a person at any moment.
• What share of the queue an agent opened.
• How long the oldest waiting change has waited.

Start collecting those on your current host now. They are what you would need to tell whether a move helped, and you can't collect them retroactively after the move.

Don't treat Origin as a hedge against your current host, because nobody keeps a spare source host. Either the review model earns a migration on its own terms, or the answer is no for another year.

Throughput figures from demo coverage. The figure quoted above reached us through press coverage of the Compile demo, not through a Cursor document, so treat it as a claim. Figures also lose detail in transit: eesel's coverage noted that a widely shared clones figure had its unit and timeframe cut off in the original headline. The commits-per-second page works through what the figure does and does not mean.

Use a rule that survives a security review. If a number isn't in cursor.com/docs/origin, the changelog or a Cursor engineering post, label it as reported in the deck or leave it out. That sounds fussy until someone quotes your slide back to you in a procurement meeting.

Apply the same discipline to dates. The early beta arrived on August 17, 2026, a month ahead of the fall window that coverage had settled on. Cursor has published no general-availability date, so a plan that assumes general availability this fiscal year rests on a guess. That can be fine. Write the assumption down where whoever inherits the plan can find it, along with what happens if it slips two quarters.