Concepts

Parallel work in worktrees

Run several LoopCode sessions on the same repository at once, each in its own git worktree and branch, without touching each other's files.

#Why worktrees

Two agents editing the same checkout trip over each other. A git worktree is a second checkout of the same repository, on its own branch, sharing the history. LoopCode can start a session in a fresh worktree so its changes never touch the checkout you are working in.

#Start a session in a worktree

loopcode -w                  # a generated name
loopcode -w payments-fix     # a name of your choice

LoopCode creates .loopcode/worktrees/<name> on a new branch loopcode/<name>, and works there. The folder is kept out of git status automatically.

In a running session, /worktree starts or continues a conversation in a new worktree, and lists and manages existing ones.

#Bring along ignored files

Files git ignores but the checkout needs to run — .env, local configuration — are not in a new worktree. List them in a .worktreeinclude file at the repository root, in gitignore syntax, and LoopCode copies them in:

.env
.env.local
config/local.json

#When you're done

When you leave a worktree session, LoopCode looks at what is in it:

• Nothing changed — the worktree is removed.
• Changes or commits — it asks: keep it, merge it into the main checkout, or remove it (with its branch). Keep is the default.

A kept worktree can be continued later:

cd .loopcode/worktrees/payments-fix && loopcode -c

LoopCode refuses to remove a worktree with uncommitted changes or unmerged commits unless you confirm it.

#Subagents in worktrees

LoopCode can also give a subagent its own worktree, so an Agent working on a separate change does not touch your files. Its changes come back as a branch to review and merge.