Cursor launches Origin, GitHub alternative

That gap between the pitch and the shipped product drove most of the reaction. People kept coming back to the same point: hosting git is not the hard part. GitHub’s moat is the surrounding system of pull requests, checks, issues, integrations, discovery, and the fact that every other developer tool already treats GitHub as the default. Several comments framed GitHub less as a forge than as a social and workflow network. That makes Origin’s biggest challenge ecosystem migration, not raw storage or even uptime. GitHub’s recent outages did give Cursor an opening, and a few people said a credible competitor could force Microsoft to improve. But that only helped so much because many readers do not see Origin as credible yet. The launch felt timed to GitHub’s bad week, while shipping an intentionally familiar beta with little new functionality. The promised future is "GitHub for agents". The released product is mostly "GitHub, but elsewhere". Trust dominated everything else. A large share of comments flatly rejected putting private code into a Musk-linked service at all. The concern was not only politics. It was custody. People repeatedly connected Origin to recent reports about Grok-related code upload behavior and to the broader fear that an AI company hosting your repository has every incentive to use repository contents, histories, fixes, and review loops as training data or product telemetry. Some asked for terms of service and enforceable guarantees. Those questions went largely unanswered in the comments, which made the trust problem worse. The more constructive thread underneath all that was about where code hosting goes next. Some argued the integrated stack is the real productivity win, so centralization or at least a widely adopted standard will keep winning. Others said AI lowers the cost of bespoke integrations enough that decentralization becomes more practical again, whether through self-hosting, Forgejo, Radicle, SourceHut, Tangled, or simpler git-plus-CI setups. The useful conclusion was not that one of those is about to replace GitHub. It was that AI may weaken integration lock-in at the margin, while making data custody a first-order product decision instead of a background infrastructure choice.