Article complete

Ex. GitHub CEO’s Startup Launches ‘Fastest’ Git Hosting Network

Entire says its distributed Git network sustained 570,000 repository clones per hour and 2.1 million Git pushes per hour in early testing, targeting AI coding agents that generate heavy concurrent Git traffic.

A man with glasses and a green sweater stands confidently with arms crossed against a neutral background.

Key takeaways

  • Entire has launched a distributed Git hosting network, claiming significant performance advantages over competitors.
  • The platform can handle high volumes of Git operations, benefiting AI coding agents that generate heavy traffic.
  • Developers can mirror GitHub repositories onto Entire's infrastructure, reducing bottlenecks and improving efficiency.
  • The company aims to decentralise and open source its Git network, promoting a collaborative environment for developers.
  • The service is currently available for existing users of Entire, enhancing their coding experience.

Takeaways drafted with AI from the article text.

The company said early testing put the platform up to 25 times ahead of competing Git hosting performance claims, compared with Cursor Origin's claimed 81,000 pushes per hour.

The service lets developers mirror existing GitHub repositories onto Entire's regional infrastructure, allowing AI coding agents to clone and fetch code from nearby mirrors instead of repeatedly hitting GitHub, reducing read bottlenecks and rate-limit issues during high-concurrency workloads.

“We believe that Git hosting must return to its original promise: a truly distributed network, not a system where the world's software lives in a single location,” says Dohmke. 

Image

(Image: Entire)

He added that the company plans to ‘decentralise and open source our Git network’ over the coming months. 

“By doing so, we enable any developer or agent to host their code in-region, pushing, pulling, and cloning close to where they operate, fast and without bottlenecks, while still part of a global, collaborative network.”

The service is available today to existing Entire users across the US, Europe, and Australia, while new users will join via a waitlist as the company gradually expands capacity.

Entire launched in February with a semantic memory layer that records every AI coding session, prompt, tool call and checkpoint alongside source code. 

The platform integrates with coding agents including Anthropic's Claude Code, OpenAI Codex, Cursor, Factory AI and GitHub Copilot, and has since added features such as Entire Blame, which links code to the AI session and prompt that produced it, Entire Review for multi-agent code reviews, and semantic search across both code and the reasoning behind code changes.

Entire emerged from stealth in February with a $60 million seed round at a $300 million valuation, led by Felicis with participation from Madrona, M12 and Basis Set.

Published

Spotted an error? Tell our editors. Read our editorial standards and corrections.