A terminal multiplexer and window manager in Go with tiling panes, workspaces, persistent sessions, and an Inbox for coding agents running in each pane.
feder-cr/git-ai
A Go server and CLI that reimplements a server-centric, agent-oriented Git design, with changes landed by compare-and-swap and AI conflict resolution.
- Stars
- 1,280
- Forks
- 124
- Open issues
- 2
- Last commit
- 3 hours ago
What is it?
gitai is an open-source, independent implementation of a server-centric version control design for coding agents. Instead of having each client clone a full repository and push after long work sessions, clients stay online, fetch only the objects and sparse paths they need, and ask the server for history questions such as blame, path history and code search. Repository state is backed by normal Git repositories underneath.
For concurrent writes, changes are rebased onto the current branch tip and landed with a compare-and-swap rather than through a merge queue, and simultaneous lands on one branch are batched into a single reference update. Conflicts can be handed to a pluggable resolver (an LLM provider, a deterministic tool or none). A model's proposed resolution only lands if a configured verify command passes on the exact resulting tree.
The project also includes a changeset log that clients can list or watch, per-agent tokens and branch policies, semantic review rules, and a storage core that can use SQLite or PostgreSQL for metadata with file, S3, GCS or Azure object storage. The Go-based server and graft CLI are built with make and require Linux or macOS.
Key features
- 01Clients fetch single objects and sparse trees instead of full clones.
- 02Blame, path history, ancestry and code search run on the server and are cached per commit.
- 03Changes rebase onto the branch tip and land via compare-and-swap, with concurrent lands batched.
- 04Conflicts can be resolved by a pluggable resolver, with model proposals verified by a configured command.
- 05Stores metadata in SQLite or PostgreSQL and packs in file, S3, GCS or Azure object storage.
Use cases
- Running many coding agents that commit to the same branches at once.
- Hosting a self-managed Git service where agents need sparse checkouts.
- Experimenting with AI-assisted merge conflict resolution behind a verification gate.
- Adding an auditable changeset log of who changed which reference.
Quick start
- 1
Clone the repository
git clone https://github.com/feder-cr/git-ai.git - 2
Enter the project directory
cd git-ai - 3
Build the server, CLI and tools
make build - 4
Run the quickstart example, which starts a server, a cache daemon, two agents and a land in a temporary directory
examples/graft/quickstart.sh
Things to know
- Supports only Linux and macOS.
- Requires Go 1.26, a C toolchain, Meson and Ninja to build, because it compiles a bundled Git.
- Version 0.3.0; interfaces may still change before 1.0.
- Known gaps such as garbage collection of packs in the storage core are not yet implemented.
Topics
FAQ
- What is git-ai?
- A Go server and CLI that reimplements a server-centric, agent-oriented Git design, with changes landed by compare-and-swap and AI conflict resolution. gitai is an open-source, independent implementation of a server-centric version control design for coding agents. Instead of having each client clone a full repository and push after long work sessions, clients stay online, fetch only the objects and sparse paths they need, and ask the server for history questions such as blame, path history and code search. Repository state is backed by normal Git repositories underneath.
- Is git-ai free to use? What license does it have?
- git-ai is open source under the Other license. Check the license terms before using it in commercial products.
- Is git-ai still maintained?
- Actively maintained. The last commit to git-ai was on Oct 21, 2026.
- What can I use git-ai for?
- Running many coding agents that commit to the same branches at once. Hosting a self-managed Git service where agents need sparse checkouts. Experimenting with AI-assisted merge conflict resolution behind a verification gate. Adding an auditable changeset log of who changed which reference.
- How do I install git-ai?
- Clone the repository: git clone https://github.com/feder-cr/git-ai.git Enter the project directory: cd git-ai Build the server, CLI and tools: make build Run the quickstart example, which starts a server, a cache daemon, two agents and a land in a temporary directory: examples/graft/quickstart.sh
- Where can I get git-ai?
- The source code and installation instructions are on GitHub: https://github.com/feder-cr/git-ai
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A self-hosted, all-in-one software development service written in Go, covering Git hosting, code review, issues, packages and CI/CD.
A SQL database that supports Git-style fork, clone, branch, merge, push and pull on tables, and connects like MySQL.
A Markdown filesystem for AI agents and teams, with full-text and vector search, Git versioning, and a built-in web UI in one Go binary.
A simple terminal UI for git commands that makes staging, committing, rebasing and branch work faster from the keyboard.
A painless, self-hosted Git service written in Go, with issues, pull requests, wiki and webhooks, distributed as a single binary.