Cursor Launches Origin: A GitHub Rival Built for the Agent Era

Cursor Launches Origin: A GitHub Rival Built for the Agent Era

On August 17, 2026, Cursor rolled out Origin — a code hosting platform built into its AI editor — to paid users. Three and a half hours later, GitHub's status page lit up with what became a six-hour-and-forty-two-minute global outage, with error rates near 20% across pull requests, issues, and the API.

The timing was coincidental. Product launches get locked weeks in advance. But the coincidence dramatized the argument Origin exists to make: after a year of mounting reliability problems, GitHub has left the door open.

What Origin Actually Is

Origin lives inside Cursor as a new Codebase tab. You name a codebase (which becomes part of the URL), push to it over the command line, and get the standard forge machinery: pull requests with timelines, commits, checks, and file diffs. Reviewers read the diff, leave comments, and merge — all without leaving the editor. Pull request conversations on synced repos flow both ways within seconds, per the launch changelog.

The part worth studying is the agent integration. Agents operate in the same surface as the code and the pull requests they're modifying. You can hand an agent a review comment, have it revise the PR in place, or tell it to push a branch — all inside the editor where the code was written. According to RuntimeWire, citing Cursor CEO Michael Truell's February post, 35% of pull requests merged inside Cursor were created by agents running autonomously in cloud VMs. Origin is built around that reality.

The GitHub Sync Strategy

Origin does not ask you to leave GitHub. Connect a GitHub organization, pick repositories, and they appear alongside Origin-native ones. Pushes keep going to GitHub, which stays the source of truth for anything started there. Access permissions mirror GitHub's existing read and write settings. Pull request conversations sync bidirectionally — comment in Cursor and it posts to GitHub; reply on GitHub and it surfaces in Cursor within seconds. All of this matches the launch changelog.

This is a classic wedge strategy. Rip-and-replace source control migration is one of the highest-risk projects an engineering organization can undertake. It touches CI, compliance, audit trails, branch protection rules, every integration in the toolchain, and the muscle memory of every engineer on staff. Almost no CTO approves that for a beta product.

A read-mostly mirror that leaves GitHub authoritative approves itself. It costs nothing to try, breaks nothing if abandoned, and quietly relocates the place developers spend their working hours.

Three Day-One Integrations

The partner choices are telling:

• Vercel spins up a preview deployment for every pull request and ships to production on merge.
• Buildkite and Depot run continuous integration, and critically, both execute existing GitHub Actions workflows unchanged. Buildkite adds native pipelines on top.

These three integrations are confirmed in the launch changelog, which notes more partners are coming.

A forge without deployments and CI is a code viewer. A forge that runs your existing Actions workflows and ships previews to the CDN you already pay for is a candidate. More partners are coming, the company said.

GitHub's Reliability Problem Is Quantified

The timing of Origin's launch overlapped with a pattern, not an incident. An analysis covered by LeadDev counted 257 incidents between May 2025 and April 2026, 48 of them major — roughly one significant disruption per week. February was the worst month on record with 37 incidents. GitHub Actions alone accounted for 57 outages in twelve months.

GitHub's CTO Vlad Fedorov has said the platform "wasn't built for the scale it's now being asked to handle" and must design for 30 times today's load. The company acknowledged in an April engineering post that it "failed to meet its own reliability standards," citing rapid growth, tight architectural coupling, and inadequate load shedding.

The defections started before Origin existed. The Zig programming language moved to Codeberg in November 2025, citing Actions failures. In April, Mitchell Hashimoto announced Ghostty — his terminal emulator — would leave GitHub, pointing to near-daily outages. And The Information reported that OpenAI began building its own GitHub alternative partly because outages left its engineers unable to commit for hours.

The Agent Problem GitHub Hasn't Solved

Google's 2025 DORA report found that 90% of developers now use AI at work. More than 80% said it made them more productive. But AI adoption showed a positive relationship with delivery throughput and a negative one with delivery stability. More output, more breakage.

The numbers tell the operational story. GitHub's Octoverse counted 180 million developers and 43.2 million pull requests merged per month, up 23% year over year. Stack Overflow's developer survey found 84% using or planning to use AI tools, while trust in their accuracy fell to 33% from 43% a year earlier.

Once a third of merged changes come from software, the pull request queue stops being a conversation and becomes a scheduling problem. Origin is built around that shift. GitHub's answer — Agent HQ — lets customers orchestrate third-party agents inside GitHub, conceding the agent layer while keeping the substrate. Origin attacks the substrate.

Why Agents Change the Forge

The scheduling problem is only half of it. When agents both write and review code, the review itself becomes an attack surface — a reviewer that runs untrusted code or fetches untrusted context can be steered, which is exactly the agent code-review risk already demonstrated against coding assistants. Origin's Codebase-tab review flow sits squarely in that blast radius: PR review by agents is convenient precisely where it is most exposed.

The SpaceX Ownership Question

In June, SpaceX completed its $60 billion all-stock acquisition of Cursor, weeks before Origin shipped. Cursor now operates as part of SpaceX. The vendor asking to hold your proprietary source code became a unit of a rocket company with its own frontier-model division.

Jason Andersen of Moor Insights & Strategy raised the model-routing question: "xAI's models and treatment of guardrails are very different than what Cursor has stood for." When one company controls the editor where agents write code, the host where that code lives, and the model those agents run on, what governs what it does with the code? That concentration is an agent security posture question, not just a vendor-management one.

Cursor has not published answers on pricing, security architecture, data-handling terms, or migration tooling — its data terms remain unpublished. The changelog says Origin reaches "all paid plan users starting today, except enterprise orgs whose admins opt out." Opt-out, not opt-in.

What to Do Before Trying It

Origin is a beta, not a migration. Treated as one, it's worth evaluating:

1. Check the default. Origin switches on for paid users unless an enterprise admin opts out. Confirm your organization's posture on mirroring proprietary code to a new host.
2. Read the terms. Retention, residency, training use, subprocessors — all unpublished. Until those terms exist in writing, treat Origin as a convenience layer over GitHub, not a system of record.
3. Know the exit. Origin's GitHub-as-source-of-truth design is what makes it safe to adopt. It's also what's most likely to erode as incentives shift toward owning the substrate. Ask what egress looks like now, while the mirror is still a mirror.

The Mirror-Trap Test

Three questions, all answerable from the facts above, separate a safe mirror from a trap. First: what breaks in your workflow the day GitHub sync is cut — CI, deploy, or just the backup? Second: where do CI and deployment actually run today (Vercel, Buildkite, Depot are named in the launch coverage), and would any of them follow you off Origin? Third: what egress does the changelog document — because a mirror with no documented exit is a migration wearing a costume. If all three answers are comfortable, evaluate freely. If any one of them is "unknown," that unknown is the evaluation.

GitHub's failures are engineering problems, and engineering problems close. The question of who holds your source code, what they may do with it, and who they ultimately answer to carries no such timestamp.

Related

• DeepSeek Harness: Open-Source Agent Framework Where Everything Is a Plugin — A composable agent architecture where every layer is replaceable.
• Why AI Writing Pipelines Need a Human Review Step — The case for keeping humans in the loop on AI-generated output.

Sources

• Cursor: Origin code hosting (changelog)
• RuntimeWire: Cursor builds Origin to host code from fleets of AI agents
• TechCrunch: Cursor launches Origin code hosting platform as GitHub outage exposes opening
• The Next Web: Cursor Origin is on by default for paid users, and its data terms are unpublished
• GitHub: Introducing Agent HQ: any agent, any way you work
• Stack Overflow: AI — 2025 Stack Overflow developer survey
• Stack Overflow: AI adoption is accelerating, but trust is declining (2025 survey)
• Google DORA: DORA 2025 state of AI-assisted software development report
• LeadDev: What's gone wrong at GitHub?
• Mitchell Hashimoto: Ghostty is leaving GitHub
• Zig Software Foundation: Migrating from GitHub to Codeberg
• Tom's Hardware: OpenAI building GitHub alternative after outages disrupted engineers
• CNBC: SpaceX to acquire the AI coding startup Cursor for $60 billion
• Tiatra: SpaceX's planned $60 billion deal for Cursor raises questions for CIOs