Entire, led by ex-GitHub CEO Thomas Dohmke, launches a Git forge where agents score every push

What Entire launched

Entire made its Git forge available to all developers on 26 October 2026. In the launch post, co-founder and CEO Thomas Dohmke, who previously led GitHub, describes it as the first forge built for collaboration between agents and humans. It rests on three parts: Trails for review, distributed Git hosting and Continuous Memory for agents.

Dohmke argues that the pull request has become an audit log that only agents read, filled with long descriptions and review comments. Trails puts eval scores at the top of the page instead. When code is pushed, Entire starts review agents that each run in a cloud sandbox and return a score, a rationale and findings with severity levels. Each project ships with agents for code quality, security and slop, and teams can add their own. Gates can require human approval and passing CI before a merge. A Trail can span several repositories.

Hosting, CI and agents under the hood

Entire runs a new Git-compatible source control server. The company claims pushes are up to 89 times faster than GitHub and 6 times faster than Cursor Origin, and says repositories can be replicated across regions for data residency. These speed figures are Entire's own and the post does not describe the test. Developers can mirror a GitHub repository and detach it, or migrate with an entire-migrate plugin.

The review agents use Claude Code and Codex with current models, running on sandboxes from Daytona and E2B. For CI, Entire integrates Buildkite and Depot, both of which offer tools to convert existing GitHub Actions workflows.

Continuous Memory and pricing

Continuous Memory keeps what agents learn in the repository. Two open-source, local-first CLI plugins do the work. entire-graph builds a map of the code with search, definitions, callers and change impact, and Entire says it saved up to 71% of tokens in its tests. entire-brain indexes past agent sessions and links them to that map, so an agent can recall why code changed. Teams can sync these memories to Entire so reviewers see them on each Trail.

Under early access pricing, code hosting, Trails, session storage, memory and search are free. Each user gets a one-time credit, and agent runs then draw on paid credit at model provider rates, billed to the repository owner. The default review model is Claude Opus 5.5 at $4 per million input tokens and $20 per million output tokens. Entire's own estimate puts one review run on a 4,000-line diff at about $0.41.

What to watch next

Entire launches into a market GitHub is reshaping with its own agent features, covered in our Agentic Workflows story. The questions are whether teams will move repositories off GitHub, what Entire charges after early access and whether independent tests back its speed claims. More agent tooling news is on the AI Agents hub.

Questions people ask

What is Entire?

A developer platform led by former GitHub CEO Thomas Dohmke. Its Git forge, launched on 26 October 2026, is designed for teams where agents write and review much of the code.

How much does Entire cost?

During early access, hosting and most features are free. Agent runs are billed at model provider token rates after a one-time free credit.

Can I move a GitHub repository to Entire?

Yes. Entire supports mirroring a GitHub repository and detaching it, migrating with its entire-migrate plugin, or pushing from a local clone.

Sources

1. The Git Forge for Agent-Human Collaboration (Entire, 26 Oct 2026)
2. Pricing (Entire, 28 Oct 2026)