Skip to content

Introduction

Coding is becoming a solved problem. Workflow is not.

AI coding agents like Claude Code and Codex CLI are getting faster every day. They can write code, fix bugs, and implement features with remarkable speed. But all that velocity means nothing if your workflow can’t keep up.

Autogit is the ultimate companion for AI coding agents. It doesn’t try to replace them — it automates everything around them: commits, branches, and PRs. When you’re running multiple agents across different worktrees, Autogit keeps your git workflow moving as fast as your code. (If you’ve used lazygit, you’ll feel right at home.)

When AI writes your code, reviewing diffs becomes more important than staring at files in an IDE. You need to see what changed, understand the context, and decide whether to commit. The IDE is becoming obsolete — diffs are the new code review.

But most git tools weren’t built for this. They freeze on large repos, lack keyboard shortcuts, and require constant mouse navigation. You spend time context-switching between terminals, writing commit messages, creating branches, opening PRs. These small interruptions add up. They break your flow and slow you down.

Autogit is built for speed. Written in Rust, it handles massive repos (like the Linux kernel with 900k+ commits) without breaking a sweat. Every action is keyboard-driven and async, so you never wait.

With AI-powered workflows, Autogit:

  • Shows diffs in real-time — review AI-generated changes at a glance, stage hunks or individual lines
  • Generates commit messages from your diffs — no more writing “fix bug” or staring at a blank message
  • Creates branches from GitHub issues — one keypress to start working on an issue with a proper branch name
  • Opens draft PRs automatically — link issues, generate descriptions, and stay in the terminal
  • Manages worktrees — run multiple AI agents in parallel, each in isolated workspaces

Autogit connects to ModelRelay for AI capabilities. When you request an AI commit or PR, your staged changes are sent to generate appropriate messages. No API keys required — just authenticate with GitHub CLI.

If you’ve looked at jj (Jujutsu), you know it promises a simpler mental model — no staging area, just work and commit. Autogit delivers similar workflow benefits without requiring your team to learn a new VCS.

When you hit the Smart Commit key, Autogit auto-stages your changes and generates a commit message in one action. For the common case of “I made changes, now commit them,” you don’t think about staging. Same simplicity, familiar git, plus AI-generated messages.