DigestAI news desk

Cut through the AI noise.

Enterprise & Industry16 min read

Anthropic replaces git with in-house system Scion for agent-scale code

Anthropic has migrated its entire monorepo from git to a custom version control system called Scion, driven by the explosive growth in AI-generated code. The company reports that daily merged code changes surged from around 2,500 in May 2026 to over 30,000 by September. This volume strained traditional infrastructure, as the main branch history alone approached 100 GB, and daily CI checkouts…

1 source primary source HN 76

Key points

  • Anthropic migrated its monorepo from git to an in-house system called Scion in five months.
  • Daily merged code changes increased from 2,500 in May 2026 to over 30,000 in September.
  • Scion uses a client-server model designed for always-online agents, replacing git's decentralized offline approach.

The new system, Scion, was designed specifically for agents that are always online, abandoning git’s decentralized, offline-first model. Instead, Scion uses a client-server architecture where the server handles commits, merges, and history queries, while clients maintain only sparse, on-demand working copies. This approach eliminates the need for large local repositories and allows for horizontal scaling without the maintenance burden of packfiles or internal mirrors.

Anthropic built Scion, along with a code review tool named Antique, in five months. The company notes that while few organizations face such extreme scale today, the pressure points identified are likely to become common as AI coding tools become more widespread. The shift highlights how developer infrastructure must evolve to support autonomous software development at machine speed.

Full story from anthropic.com · by Anthropic · via Hacker News primary sourceOpen source ↗

Redesigning Source Control at Anthropic

anthropic.com · 13 October 2026

Loading the full article…

This text was published by anthropic.com and written by Anthropic. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Coverage and discussion

1source
Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Enterprise & Industry

All →

Related stories