Trackers

Every tracker is its own history. root owns . — whatever no more specific tracker claims, .newgit/ included — and starts as your project's Git history; access, mounted at .newgit/trackers/, holds who may read what, which only the project's owner changes by default (maintainers = "owner"). Carve out a tracker in Git with newgit tracker create electron --git --path apps/electron: one commit that mounts it there, with its history starting at the carve. Carved inside another mounted tracker, it is that tracker's child, defined in that tracker's own .newgit/trackers/: bound only where its parent is, and brought along with it unless its own priority is negative. So a large directory can stay out of workspaces while the parent's own repository still holds it. Trackers are named like files: t3code/repos from the top of a project, repos from inside apps/t3code, .. to go up, and a leading / for the full path, such as /remote/<owner>/<project>/t3code/repos. Without --git a tracker stays outside Git: captured, never committed.

An instance works on what it binds:

newgit spawn fix-db --use staging-db --drop electron   # one per role; the rest as you say
newgit tracker drop electron --stash   # out of the working tree, history untouched
newgit tracker use electron            # back, with the stashed work
newgit status                          # ready / starting / not ready, and why

Trackers sharing a role are alternatives (dev-db, staging-db); a spawn picks one by --use, then your own default in ~/.newgit/config.toml, then priority. A tracker can bring resources (they run only while it is bound) and require others.