Anthropic replaces Git with custom version control system Scion as daily merges top 30,000
Anthropic has migrated its monorepo from Git to an in-house system called Scion after daily code merges surged tenfold to more than 30,000, driven by autonomous coding by Claude.
Anthropic announced that it has replaced Git with an in-house version control system dubbed Scion, completing a full monorepo migration over a five-month span. The AI lab explained on its engineering blog that its previous developer infrastructure, built for human rather than agent scale, became a bottleneck as daily merged code changes jumped from around 2,500 in May to more than 30,000 in September.[2][1][4]
According to data shared from Anthropic's monorepo tracking January through September 2026, daily merged changes reached a single-day peak of 30,401 on September 9. The data reveals a composition shift from traditional pull requests toward Antique change requests, aligning with reports that Anthropic also created a GitHub replacement named Antique to manage agent-driven workflow.[6][7][3]
The rapid increase in merge activity followed Claude taking over major software engineering tasks, with Anthropic noting that the model can autonomously iterate on research loops, micro-optimize complex programs, and build new services from scratch at unprecedented scale. Scion itself was built by a team of seven engineers working with Claude.[7][5][1]
Key facts
- Anthropic developed an in-house version control system named Scion and migrated its monorepo from Git in five months.
- Daily merged code changes in Anthropic's monorepo rose from around 2,500 in May 2026 to more than 30,000 in September 2026.
- Monorepo merges peaked at 30,401 changes in a single day on September 9, 2026.
- Scion was built by seven engineers using Claude over a five-month period.
- Anthropic also created an internal GitHub counterpart named Antique to support agent-driven changes.
- Anthropic said existing developer tooling became a bottleneck because it was built for human rather than agent scale.
Sources · 6 sources
- TE
Techmeme@TechmemePost on X ·
Anthropic details Scion, a git replacement for its monorepo, built by seven engineers with Claude in five months, as daily merges rose from 2.5K in May to 30K+ (Anthropic) (Visit Techmeme dot com for the link and full context!)
Open source - AN
Anthropic@AnthropicAIPost on X ·
New on our Engineering blog: Creating in-house source control. In May, we merged around 2,500 code changes per day. In September, we merged more than 30,000 per day. As a result, we became bottlenecked by developer infrastructure built for human (rather than agent) scale. Here, we share how we built a new version control system—and fully migrated our monorepo from git—in five months: https://t.co/wzNkwHs7IN
Open source - KC
Kent C. Dodds 🐨@kentcdoddsPost on X ·
Anthropic built its own git and github replacement called scion and antique. Built for agents. https://t.co/oyWB9LZDpc
Open source - CB
Crypto BriefingArticle ·
Anthropic replaces Git with custom source control system Scion Anthropic's shift to Scion highlights the growing need for scalable, efficient systems to manage AI-driven code production, despite potential risks. The post Anthropic replaces Git with custom source control system Scion appeared first on Crypto Briefing .
Open source - CB
Crypto BriefingArticle ·
Anthropic’s daily code merges jump 10x since May as Claude takes over the keyboard Anthropic's shift to Scion highlights the growing need for advanced infrastructure as AI-driven coding accelerates, reshaping software development. The post Anthropic’s daily code merges jump 10x since May as Claude takes over the keyboard appeared first on Crypto Briefing .
Open source - MT
MTS@MTSlivePost on X ·
SITUATION DETECTED: Anthropic says its code merges per day have grown tenfold and Claude is now able to autonomously iterate on research loops at a speed and scale they have not seen before. https://t.co/bkxgPJ1vO1
Open source - AC
Andrew Curran@AndrewCurran_Post on X ·
Anthropic's code changes merged per day has 10x since May. 'Claude is able to autonomously iterate on research loops, micro-optimize complex programs, and build new services from the ground up at a speed and scale we haven’t seen before.' https://t.co/KsUDDAfkMM
Open source