
Scion – Is Git's End Near?
I don’t want to waste your time, so I’ll answer right away: in my opinion, no. However, if you’re interested in why, I encourage you to stick around and read the article to the end.
以下正文同步自 Hacker News 最新,版权归原站所有,已转换为易读排版。
elszczepano
Nov 17, 2026
Share
I don’t want to waste your time, so I’ll answer right away: in my opinion, no. However, if you’re interested in why, I encourage you to stick around and read the article to the end.
Importantly, I’m presenting only my own opinion here. I’m not focusing on comparing the two tools in terms of performance or efficiency. In this post, I’m focusing on what a new version control system might change and how it affects Git’s position.
Scion is a version control system created by Anthropic. Anthropic’s engineers aren’t the first to conclude that Git doesn’t meet their requirements and that they need their own version control system. Meta had previously made this decision when it created Sapling, as had Google when it built Piper.
As the pace of code delivery increased, the development infrastructure – CI, the build process, and change management became a problem. In September 2026, Anthropic engineers were merging over 30,000 commits per day. There is no indication that this pace will slow down. On the contrary, since May 2026, the number of commits merged has increased more than tenfold, and the upward trend suggests the pace will continue to accelerate.
The pace of code deployment in Anthropic’s infrastructure. Source: Anthropic
The second problem was a disconnect between the principles underlying Git and the way Anthropic engineers use it. Git is a distributed version control system in which each repository clone is fully independent, and there is no centralized single source of truth. Using an agent-based approach to code delivery generated unnecessary costs and delays. Agents and CI systems unnecessarily had to clone the repository each time and perform millions of unnecessary checkouts. A centralized approach with a single source of truth simply works better for Anthropic.
A third argument in favor of migrating to a proprietary solution is the development of a custom code review tool (Antique). A proprietary VCS allows for much better integration of both systems than using off-the-shelf solutions.
To solve these problems, Anthropic decided to build its own version control system called Scion. At the time of this article’s publication, Scion is not publicly available and remains an internal system at Anthropic.
First and foremost, because as of the publication of this article, Scion is not a publicly available project. Furthermore, Anthropic itself acknowledges that Scion is not a replacement for Git or a general-purpose tool:
Scion is by no means a suitable replacement for git for general use. The concepts we’ve outlined here really only make sense for truly enormous monorepos, with either tens of thousands of engineers or swarms of agents prolifically writing code. [...]
Practically speaking, it’s difficult to offer Scion in a form that would be broadly useful to other organizations. It diverges significantly from git, meaning companies would need to migrate every piece of the development lifecycle, and it requires its own code review, code search, and CI. Our internal versions of these components are highly bespoke, so they would be difficult for other organizations to adopt, and we worry that attempting to build a public version that tries to please everyone wouldn’t actually be very useful.
For the sake of further discussion, let’s assume that in the near future Anthropic changes its mind and decides to launch Scion on a large scale, in an attempt to gain a foothold in the version control market as well.
Git has been on the market for over 20 years and has achieved a dominant position during that time. Migrating to a different version control system is a task that cannot be accomplished overnight. Over the years, a vast ecosystem of tools, integrations, and extensions has grown up around Git, one that would be difficult to replace. I believe that even if coding agents significantly simplify the creation of new tools, replacing existing ones may still be difficult and time-consuming. Git is currently the industry standard.
The widespread use of tools like GitHub and GitLab means that starting a new project practically assumes the use of Git from the get-go. The vast majority of open-source projects are hosted on platforms that use Git. A mass migration to any other tool is a complex process. The world first heard about Scion just one month before the publication of this article. That’s far too short a time to be able to say that the solution will prove successful in the long term. This is especially worth considering now, when the pace of change and innovation in AI is so rapid. The current flood of innovations doesn’t convince me that anyone will even remember Scion 20 years from now.
Another argument in Git’s favor is its widespread use and familiarity among programmers. The barrier to entry for Git is relatively low, and virtually every programmer can use it at a basic level. As for Scion, there are currently no educational resources available, and this won’t change until the project is made public. Even if educational materials become available, it will take some time before best practices and patterns for working with this tool are established. If I were you, I’d consider whether your project is the right one to experiment with.
Finally, it’s worth asking yourself: “Why?” Scion was created in response to a specific problem that engineers at Anthropic encountered. The same for other Git alternatives. If Git has become a bottleneck in your project and you’ve already exhausted all optimization possibilities, it makes sense to look for alternatives. However, to get a sense of whether your project operates on a similar scale, it’s worth comparing it to the situation that led Anthropic to migrate:
The history of our main branch alone is almost 100 GB, and a compressed working copy is over 25 GB. We perform on the order of 25 million code checkouts per day for CI, before counting developer environments, agent sandboxes, and research jobs. Fetched naively, that would be over 600 PB of git traffic per day.
Another valid reason for testing alternatives to Git might simply be the desire to learn something new and broaden your horizons. However, if you don’t have a specific problem that alternative solutions address better, I consider implementing them in non-hobby projects to be Resume-Driven Development.
Last but not least, I simply don’t trust Anthropic. It’s not a charity. If the company decides to make Scion publicly available, it must have some vested interest in doing so. I don’t believe that if the solution hits the market, it will be in the form of open source. If we don’t pay for Scion directly, we’ll pay with our data. Git is free, open source, and has a broad community of contributors.
A scenario in which Git is abandoned en masse in favor of new alternatives is possible, but in my opinion, extremely unlikely. In addition to Scion, there are also more accessible alternatives to Git that are competitive with it, such as the open-source Jujutsu project.
Personally, however, I rate the chances of this happening very low. Even if a viable alternative were to gain traction and earn industry recognition, the migration itself would likely stretch out over many years or not happen at all.
However, if Anthropic decides to release Scion, I encourage you to carefully consider whether you want to jump on this hype train.
Anthropic Blog - Redesigning source control at Anthropic
Gergely Orosz - The Pulse: Anthropic builds its own source control system
Jujutsu - A Git-compatible VCS that is both simple and powerful
sunshowers - Anthropic’s Scion is heavily inspired by Mononoke
sunshowers - More on Anthropic’s Scion ripping off Mononoke
正文由 探How 从来源站点 RSS 同步,内容未经改写;遇到排版缺失或需要图片、视频时请以原文为准。