Entire, the developer startup led by former GitHub CEO Thomas Dohmke, has opened its Git forge to every developer today. The Entire Git forge is built for a world where AI coding agents push code faster than people can read it. It swaps the classic pull request for something called a Trail.
The launch follows a year of building, and the company says the goal is simple: help teams merge changes as fast as agents create them.
What the Entire Git Forge Does
A Git forge is the place where teams host code, review changes and run tests. Most developers know GitHub as the default. Entire argues that this model was designed for humans typing code, not for agents producing thousands of commits.
Entire rests on three main ideas, according to its launch post.
- Trails, which replace the pull request as the place where code gets reviewed.
- Distributed Git hosting, built on a new Git-compatible server.
- Continuous Memory, which keeps agent sessions and prompts next to the code.
How Trails Replace the Pull Request
In a normal pull request, comments stack up in one long timeline. Entire says that with coding and review agents posting long notes, the timeline has become a log that mostly only agents read.
A Trail puts scores at the top of the page instead. Review agents run on every push, and each returns a score and a short rationale, much like judges scoring a skating routine. Default evals cover code quality, security and what the company calls slop.
Loops Until the Scores Turn Green
If a score looks bad, you can tell your coding agent to fix the problem and push again. Each Trail shows the last five runs plus the current one, so you can see whether the results have settled. Teams can also set Gates that require human approval or passing tests before a merge.
Under the hood, Entire says its review agents use Claude Code and Codex models in isolated cloud sandboxes from Daytona and E2B.
Why the Entire Git Forge Hosts Faster
Entire says its server pushes code up to 89 times faster than GitHub and six times faster than Cursor Origin, based on open-source benchmarks the company published. Treat those numbers as the company’s own claims until independent tests appear.
Repositories can be copied across several regions. The service is live in the US, the EU, Australia and India, which matters for teams with data residency rules. You can create a new repository or mirror an existing GitHub one.
| Feature | Classic pull request flow | Entire’s approach |
|---|---|---|
| Review unit | Pull request with a diff and comments | Trail with agent-scored evals |
| Merge decision | Human reads the changes | Scores, findings and Gates guide it |
| Agent context | Lost when a session ends | Saved in the repo as Continuous Memory |
| Hosting | Centralized servers | Distributed, multi-region |
Continuous Memory for Coding Agents
Most of what an agent learns disappears when a session ends. Entire’s answer is two open-source plugins: entire-graph maps how your code fits together, and entire-brain ties past agent sessions to that map.
The company says the code map saved up to 71% of tokens in its own testing. For anyone paying per token, that is a real cost angle.
Partners and Compatibility
Entire works with Buildkite and Depot for continuous integration, so pipelines keep up with agent speed. Both can convert existing GitHub Actions workflows. The Entire CLI also works with tools such as Claude and Grok, saving their sessions into your Git history.
What This Means for You
If you write code with agents every day, this launch is worth watching, even if you never leave GitHub. Here is how to think about it.
- Try it on a side project first. A mirrored repository lets you keep GitHub as your source of truth.
- Check the speed claims yourself, since the 89x figure comes from the vendor.
- Look at your own review habits. If you already approve agent output without reading every line, evals and gates may fit your workflow.
- Keep an eye on session data. Storing prompts in your repo is useful, but decide who should be able to see it.
The bigger story is that coding platforms are being rethought around agents. Tools like Claude Code have pushed teams to merge more code, and the bottleneck has moved to review and infrastructure. You can follow more stories like this in our AI news section.
Sources
Primary source: Entire’s launch announcement. Additional reporting from ZDNet.
Conclusion
Entire is betting that the pull request is the wrong tool for the agent era, and it is now open to all developers. Whether Trails and distributed hosting win over teams will depend on independent speed tests and how well the platform handles real projects. For now, it is a serious new option for anyone whose agents write most of the code.
