Urgent.News

What's breaking now, across thousands of outlets.

AI

Anthropic Reinvents Source Control As Claude Floods the World With Code

Anthropic says Claude-generated code has scaled so quickly that its existing Git-based infrastructure could not keep up. In response, the company built a custom centralized source-control system called Scion, along with AI-assisted review and search tools. The Register reports: "In May 2026, Anthropic averaged around 2,500 code changes merged per day; as of September, we merged over 30,000 per…

Anthropic's Claude AI has generated code at such a rapid pace that the company's previous Git-based system could not keep up. To address this, Anthropic developed a custom centralized source-control system named Scion, complete with AI-assisted review and search capabilities. In May 2026, Anthropic averaged around 2,500 code changes merged daily; as of September, the number had surged to over 30,000 per day.

The company expects this growth rate to continue unabated, as there is a virtually limitless demand for merging code, and each bottleneck removal leads to further increases in code production. After months of model prompting and migration, Anthropic created Scion, a monorepo-like system akin to Subversion. Scion is tailored for AI agents, with client-side work offloaded to the server.

Queries are accessible via API for internal tools, and while synchronization is reduced, server speed becomes more crucial. Anthropic's changeset, a commit unit in Scion, stores files changed relative to the first parent and includes content-addressed pointers to compressed file contents in blob storage. Although designed for internal use, Anthropic believes similar source control challenges will emerge for other organizations, prompting the sharing of Scion's approach.

Written by urgent.news from Slashdot's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at developers.slashdot.org →

More in AI

More from Wednesday 14 October →