Your first scan

zap scan only reads. Run it inside any repository:

cd ~/dev/my-app
zap scan

Each row is one linked worktree. zap leaves the main checkout (~/dev/my-app itself) out of the list and never removes it.

Column	What it tells you
WORKTREE	Where the folder is. Paths inside the repository are shown relative to it.
BRANCH	The branch it has checked out, or a commit hash if it is detached.
AGENT	Which agent made it, guessed from the folder: .claude/worktrees is Claude Code, ~/.codex/worktrees is Codex, and so on. - means zap can't tell, which usually means you made it.
STATUS	merged, empty, active and a few others. See How zap decides.
SIZE	How much disk removing it would free.
DEPS	The part of SIZE in node_modules, .venv, target and other dependency or build folders.
AGE	When the worktree was created.
ACTIVE	When anything in it last changed.
NOTES	Extra facts, such as dirty(1) for one uncommitted file or detached.

In this example, the two Claude Code worktrees are merged into main, the Codex one never got a commit of its own, and search has an uncommitted change. zap would offer the first three for removal and leave search alone.

Every repository at once

zap scan --all

--all searches your scan roots (~/dev, ~/code, ~/src, ~/projects and the folders agents use) for repositories. To change where it looks, see Scan roots.

The last lines total it up: how much space linked worktrees use, how much of that is merged or empty, and how much free space your disk has.

Faster scans

Measuring sizes takes most of the time on a big disk. Skip it with --no-size.

Next: Why zap chose that status