· Oct 13, 2026
Anthropic explains why it built Scion for its growing monorepo
Markdown · JSON · Original source
’s article, published October 13, 2026, argues that ’s coding capacity has shifted the bottleneck in its software work to developer infrastructure. Anthropic reports that its daily merged code changes rose from around 2,500 in May to over 30,000 in September, while its monorepo’s main-branch history grew to almost 100 GB. It built Scion, an in-house version control system, and migrated its monorepo from git in five months. Scion is designed for clients Anthropic says are mostly online agents and jobs: they request data and operations from servers rather than keeping a complete repository locally. Anthropic reports that checkouts became two orders of magnitude faster and its infrastructure footprint became 200x smaller than its previous git-mirroring infrastructure, by and memory requirements. CI alone gained 25% CPU capacity. Anthropic says Claude helped it build Scion, while engineers supplied system-level understanding and made design tradeoffs. It stresses that Scion is not a general replacement for git: its approach suits truly enormous monorepos, and adopting it would require changes to other development tools. Anthropic presents its contribution as sharing ideas that other developers may apply to git, not offering Scion as a broadly suitable replacement.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
· Article
Source details
Original source date: Oct 13, 2026 (2026-10-13T00:00:00.000Z).
Date basis: Date stated on the publisher’s page.
Added to Escape Meridian: 2026-11-05T00:18:29.785Z. First observed: 2026-10-20T22:35:20.777Z. Retrieved: 2026-10-29T23:19:45.741Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 29, 2026; originally published October 13, 2026. The performance figures and descriptions of ’s internal systems are reported by Anthropic; the article does not provide independent verification.
Open question
How well would Scion’s reported scaling benefits and tradeoffs carry over to organizations with different workloads and development tools?
Source-stated
Anthropic says the bottleneck to its software work has shifted from coding to developer infrastructure.
The bottleneck, now, is all of the developer infrastructure built for human scale: source control, builds, and CI.
Source-stated
Anthropic reports that its daily merged code changes grew from around 2,500 in May 2026 to over 30,000 in September.
In May 2026, Anthropic averaged around 2,500 code changes merged per day; as of September, we merged over 30,000 per day.
Source-stated
Anthropic says it built Scion and migrated its monorepo from git in five months.
we built Scion, an in-house version control system, and fully migrated our monorepo from git in five months.
Source-stated
Anthropic reports that Scion checkouts are two orders of magnitude faster than before and that CI gained 25% CPU capacity.
CI and research jobs check out code two orders of magnitude faster (in seconds rather than minutes), which frees up a sizable amount of capacity across the fleet; CI alone realized a 25% CPU capacity win.
Source-stated
Anthropic says a team of seven built Scion, Antique and related systems and completed the migration in five months.
A team of seven people built both Scion and Antique (our code review tool, which we could easily write a whole other blog post about), not to mention half a dozen related systems, and fully migrated to them in five months.
Source-stated
Anthropic says Scion is not a general replacement for git and is intended for truly enormous monorepos.
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
Source-stated
Anthropic argues that its engineers still supplied system-level understanding and made design tradeoffs while working with Claude.
our engineers still supplied the invaluable understanding of the high-level inter-system components and the design-level tradeoffs
Source and contributor context
Anthropic · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
CPU
The central processing unit: a general-purpose processor that carries out a computer's instructions.
Anthropic
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Claude
Claude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Editorial provenance
Version 1. Publication ID: cf86fd10-f2ad-306b-99c1-9998a04d3141.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-05T00:13:29.664Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: c599662b6589ce22315d0dab688a47fe555cfd7d2b09dc1ef2d18215c125466c. Source text hash: b459a1e6b9eab3c47ef99d10d8b6f82a1927c6646bd13cb29452416fbdf6f19d.
Revision and source-check history
Version 1 · 2026-11-05T00:18:29.785Z
Anthropic explains why it built Scion for its growing monorepo
’s article, published October 13, 2026, argues that ’s coding capacity has shifted the bottleneck in its software work to developer infrastructure. Anthropic reports that its daily merged code changes rose from around 2,500 in May to over 30,000 in September, while its monorepo’s main-branch history grew to almost 100 GB. It built Scion, an in-house version control system, and migrated its monorepo from git in five months. Scion is designed for clients Anthropic says are mostly online agents and jobs: they request data and operations from servers rather than keeping a complete repository locally. Anthropic reports that checkouts became two orders of magnitude faster and its infrastructure footprint became 200x smaller than its previous git-mirroring infrastructure, by and memory requirements. CI alone gained 25% CPU capacity. Anthropic says Claude helped it build Scion, while engineers supplied system-level understanding and made design tradeoffs. It stresses that Scion is not a general replacement for git: its approach suits truly enormous monorepos, and adopting it would require changes to other development tools. Anthropic presents its contribution as sharing ideas that other developers may apply to git, not offering Scion as a broadly suitable replacement.
Original source
· Article
Source details
Original source date: Oct 13, 2026 (2026-10-13T00:00:00.000Z).
Date basis: Date stated on the publisher’s page.
Added to Escape Meridian: 2026-11-05T00:18:29.785Z. First observed: 2026-10-20T22:35:20.777Z. Retrieved: 2026-10-29T23:19:45.741Z.
Limits
Text as retrieved on October 29, 2026; originally published October 13, 2026. The performance figures and descriptions of ’s internal systems are reported by Anthropic; the article does not provide independent verification.
Open question
How well would Scion’s reported scaling benefits and tradeoffs carry over to organizations with different workloads and development tools?
Source-stated
Anthropic says the bottleneck to its software work has shifted from coding to developer infrastructure.
The bottleneck, now, is all of the developer infrastructure built for human scale: source control, builds, and CI.
Source-stated
Anthropic reports that its daily merged code changes grew from around 2,500 in May 2026 to over 30,000 in September.
In May 2026, Anthropic averaged around 2,500 code changes merged per day; as of September, we merged over 30,000 per day.
Source-stated
Anthropic says it built Scion and migrated its monorepo from git in five months.
we built Scion, an in-house version control system, and fully migrated our monorepo from git in five months.
Source-stated
Anthropic reports that Scion checkouts are two orders of magnitude faster than before and that CI gained 25% CPU capacity.
CI and research jobs check out code two orders of magnitude faster (in seconds rather than minutes), which frees up a sizable amount of capacity across the fleet; CI alone realized a 25% CPU capacity win.
Source-stated
Anthropic says a team of seven built Scion, Antique and related systems and completed the migration in five months.
A team of seven people built both Scion and Antique (our code review tool, which we could easily write a whole other blog post about), not to mention half a dozen related systems, and fully migrated to them in five months.
Source-stated
Anthropic says Scion is not a general replacement for git and is intended for truly enormous monorepos.
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
Source-stated
Anthropic argues that its engineers still supplied system-level understanding and made design tradeoffs while working with Claude.
our engineers still supplied the invaluable understanding of the high-level inter-system components and the design-level tradeoffs
Source and contributor context
Anthropic · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
CPU
The central processing unit: a general-purpose processor that carries out a computer's instructions.
Anthropic
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Claude
Claude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Editorial provenance
Version 1. Publication ID: cf86fd10-f2ad-306b-99c1-9998a04d3141.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-05T00:13:29.664Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: c599662b6589ce22315d0dab688a47fe555cfd7d2b09dc1ef2d18215c125466c. Source text hash: b459a1e6b9eab3c47ef99d10d8b6f82a1927c6646bd13cb29452416fbdf6f19d.
uncertain · 2026-11-05T00:13:29.664Z · anthropic/claude-opus-5.5
Publication: cf86fd10-f2ad-306b-99c1-9998a04d3141
attribution — pass: This is a single Anthropic article. The reading consistently attributes the bottleneck claim, merge figures, Scion's design, performance numbers and limits to Anthropic as reported claims, not established facts.
In order to support this growth, we built Scion, an in-house version control system
summary — uncertain: The summary is faithful, but central terms are unexplained for non-specialists. 'Monorepo', 'git', 'CI' and 'checkouts' are not defined in the text or definitions, which list only CPU. A reader cannot tell that a monorepo is one shared code repository or what CI checkouts are. Smallest repair: add tooltips or plain glosses for these terms. Fidelity is weakened by omitting the source's pointed claims that engineers now rate Claude as better at writing code and that their role was often to tell Claude no. The wording 'grew to almost 100 GB' also adds a growth framing to a size the source simply states.
All of the engineers who worked on this system would now rate Claude as better at writing code than them.
date — pass: The October 13, 2026 date matches the UTC day of originalPublished, the precision is appropriate, and the kind is publisher-page, so no archive disclosure is needed. Anthropic's six-to-12-month expectation is not presented as an outcome. A modifiedAt of October 20 exists, but editedAfterPublication is not flagged, and the cited passages read as part of the main article.
we anticipate that this class of problems will become more commonplace over the next six to 12 months
support — pass: Each claim's excerpt appears verbatim in the source and supports it. These include the 2,500 to 30,000 daily merges, the five-month migration, the seven-person team, the two orders of magnitude faster checkouts, the 25% CI CPU gain, the 200x smaller footprint and the not-a-git-replacement caveat. The limitation correctly notes there is no independent verification. The question is open and not leading.
Our Scion infrastructure footprint is now 200x smaller than our previous git mirroring infrastructure in terms of CPU and memory requirements.
media — pass: No media was selected and no verified media candidates were supplied, so mediaId null is appropriate.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicClaude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Explore ClaudeThe central processing unit: a general-purpose processor that carries out a computer's instructions.
Definition referenceAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicClaude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Explore ClaudeThe central processing unit: a general-purpose processor that carries out a computer's instructions.
Definition referenceAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore Anthropic