Overview

How newgit works and how to drive it, in more depth than the README. For every tracker and resource key, its default, and which template variables each hook sees, run newgit reference: the definition format travels with the binary.

The central idea shows up in the code shape: branch instances bind source state to user-defined trackers and resources, instead of baking env files, installs, processes, databases, or external resources into special internal lanes.

Using newgit

• The workflow: the whole loop, on one machine and with a remote
• Trackers: what owns which files, and what an instance binds
• Adopting it in a project: init, what to commit, and what installs cost
• Behavior worth knowing: export, GitHub, tags, [[render]], cleanup
• Running your own remote: newgit serve, and pointing a project at it
• Commands: every command, in one list

With your agent

• T3 Code: a newgit instance for every thread
• Claude Code: a newgit instance for every worktree

Working on newgit

• Development: building, testing, and the repository's layout
• The design documents explain why it is shaped the way it is: the design, the tracker model, and the remote.

The wiki is the Markdown in docs/wiki/. Every page has an edit link, and a change is live within a minute of merging.