Cursor takes on GitHub with its own code hosting platform
Cursor challenges GitHub with its own code hosting platform, extending its AI coding business beyond the editor.
Most developer teams still juggle three separate tools: one for hosting code, one for reviewing it, and one for running AI agents against it. Cursor is now trying to collapse all three into the editor developers already work in.
On 17 August 2026, Cursor announced Origin, a cloud code hosting service that sits inside the Cursor client. It is rolling out in early beta to all paid plans, except enterprise organisations whose admins opt out. The timing turned into an unplanned advertisement: hours after the rollout began, GitHub suffered a global outage lasting close to seven hours, with error rates approaching 20% across services and nearly 50% on archive and raw file downloads.
Origin is also Cursor's first major product launch since its $60 billion all-stock sale to SpaceX, announced on 16 June 2026 and closed on 14 August, three days before Origin shipped.
It is built on Git, the version control system that lets teams track changes and work safely in parallel, so nothing about how you version code actually changes. What changes is where everything lives.
Code and agents come together
Origin lives in a new Codebase tab in the Cursor client, and is also available through a dedicated command line tool. Repositories get URLs in the format cursor.com/codebase/name. Teams can create and clone repositories, browse and search code, and open, review and merge pull requests with full timelines, commits, checks and file diffs.
Cursor's AI agents can work directly on those repositories. Developers can ask questions about code they are browsing, and the agent can answer, make changes, update a pull request or push a branch.
There are two types of agents. Desktop agents run on a developer's own computer, while cloud agents operate inside a managed sandbox. Cloud agents can continue working on long running tasks even when a developer's computer is switched off, which suits larger refactors, maintenance work and other multi step jobs that do not need constant supervision.
The code review piece did not appear from nowhere. Cursor acquired Graphite, a startup built around pull request workflows, in December 2025.
GitHub still has a role
Developers do not have to leave GitHub to use Origin. Connect a GitHub organisation, pick repositories, and they appear in Cursor alongside Origin native ones. Synced repositories update in real time, and you can browse, search and pull from the Origin copy.
The sync is not symmetric, and that distinction matters. Pushes keep going to GitHub, which remains the source of truth for anything started there. What does sync both ways is pull request activity: a comment left in Cursor posts to GitHub, and a reply or reaction on GitHub shows up in Cursor within seconds. A review assigned on GitHub can be handled and merged from Cursor.
That gives teams a way to test Origin's review and agent capabilities without migrating their development workflow.
More tools are joining Origin
Origin launched with three integrations. Vercel, connected from a repository's Apps tab, gives every pull request a preview deployment where changes can be tested and commented on before merging.
Depot and Buildkite handle continuous integration. Both run a team's existing GitHub Actions workflows, and Buildkite also runs its own native pipelines. Cursor says more integrations are coming, along with agent native features it has not yet detailed.
That last point is the one to watch. As AI agents take on larger parts of the software development process, having code, reviews and agents in one place could let agents move between tasks with less manual coordination.
Why Git remains important
Origin is built around Git, which was created by Linus Torvalds. One of Git's core features is branching, which allows developers to create separate versions of a project, make changes independently and merge them back into the main codebase.
By keeping Git at the centre, Cursor is building on a workflow developers already understand rather than introducing a new way to manage code. The bigger change is the addition of AI agents directly alongside those existing workflows.


