chore(worktree): keep Claude Code worktrees out of the working tree

[anilcancakir]

What

.gitignore: ignore .claude/worktrees/ (only that subdirectory; the rules under .claude/ stay tracked).

Why

Worktrees were opened beside the checkout as <repo>-<slug>, so they landed in the workspace directory next to the real repositories, and Claude Code's EnterWorktree could not be used at all: it writes to a fixed .claude/worktrees/<slug>. With them inside the repository, git has to be told to ignore them.

No .worktreeinclude here, because this repository keeps no pubspec_overrides.yaml; there is nothing gitignored that a worktree would need. Sibling repositories that do carry one get that file too, see fluttersdk/magic_starter#124.

Testing

Config only. No Dart, YAML or lib/ file is touched, so analyze, format and test have nothing to say about this diff.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[kodizm]

Note

Kodizm (AI-generated). May contain mistakes; verify before acting.

A five-line .gitignore addition that does exactly what it says; the ignore rule is correctly scoped and nothing downstream is affected.

Minor

.gitignore:41-42 — The comment says "the rest of .claude/ is tracked, because the rules under it are part of the repository", but this repository tracks nothing under .claude/ today (git ls-files .claude returns 0 files). The rule itself is right; the justification describes a sibling repository's state, not this one. Maintainability only. (Also worth knowing: .pubignore:52 already excludes all of .claude/, so nothing needs adding there for publishing.)

Tests

Config-only change with no test surface, which matches the description. Verified behaviourally instead, see below.

Checks I ran

• git check-ignore -v .claude/worktrees/foo → .gitignore:43:.claude/worktrees/, so the pattern matches, and .claude/rules/x is not matched: the scoping is correct.
• git ls-files .claude | wc -l → 0, the basis for the Minor above.
• git status --porcelain → clean, no tracked file becomes ignored by this change.
• Simulated a worktree at .claude/worktrees/slug/ (with its own pubspec.yaml, as a real git worktree of this repo would have) and ran dart analyze and dart format --output=none --set-exit-if-changed . at the root: neither reported a single path under worktrees/, because the nested pubspec makes it a separate analysis context. So placing worktrees inside the checkout does not pollute local analyze or format runs. CI is unaffected regardless, since .github/workflows/ci.yml:27,30 scopes both to lib/ test/.
• The diff also removes the missing trailing newline on the old last line, which is a tidy-up rather than a behaviour change.

[anilcancakir]

Fixed. git ls-files .claude returns nothing here, so the comment described a sibling repository rather than this one. It now says only that subdirectory is ignored, so anything else landing under .claude/ stays visible to git, and that nothing is tracked there today.

Also noted your .pubignore:52 point: nothing needed adding there.

[kodizm]

Note

Kodizm (AI-generated). May contain mistakes; verify before acting.

Since my last review the ignore rule itself is unchanged and still correct; the new commit only rewrites the comment above it, and two things in that new prose are off.

The .claude tracking claim I raised is fixed, thanks. Reviewing only the comment rewrite in 6e4a5ba.

Minor

.gitignore:42 — Stray # glued to the end of Nothing is tracked there today.#. It looks like an intended blank comment separator that lost its newline; harmless to git, but it reads as a typo in a file whose whole purpose here is explanation. (maintainability)

.gitignore:45-46 — The claim "A worktree's .git is a file rather than a directory, so git's nested-repo guard does not stop it" is the wrong way round: git's guard resolves gitfiles too, so plain git clean -xdf does skip a nested worktree and leaves it intact. The hazard the comment describes only appears at -xdff. As written it warns about the wrong command, so someone reasoning from it would draw the wrong conclusion about which invocation is safe. (maintainability)

Tests

Comment-only change, no test surface. Verified the claim behaviourally instead, below.

Checks I ran

• Built a throwaway repo with a worktree at .claude/worktrees/slug (git 2.43.0) and ran git clean -xdf → Skipping repository .claude/worktrees/slug, and the worktree checkout plus .git/worktrees/slug both survived. Re-running with git clean -xdff → Removing .claude/, which does destroy it. That pair is the basis for the second finding.
• git check-ignore -v .claude/worktrees/foo → .gitignore:47:.claude/worktrees/; .claude/rules/x still not matched, so the scoping is unchanged and correct.
• git status --porcelain → clean; git ls-files .claude → 0, consistent with the new comment's "nothing is tracked there today".

[anilcancakir]

You are right and I had it backwards. Reproduced your probe independently before changing anything:

$ git clean -xdf --dry-run
Skipping repository .claude/worktrees/slug

$ git clean -xdf        # worktree intact, its untracked scratch file too
$ ls .claude/worktrees/slug/
untracked.txt

$ git clean -xdff       # gone
$ git worktree list --porcelain | grep -c prunable
1

So the guard keys on the gitlink, exactly as you said, and my stated reason (that a file-form .git slips past it) was wrong as well as the command. The comment now says -xdf leaves them alone and names -xdff as the one that does not, with the "Skipping repository" line quoted so the next reader can recognise it.

The stray # is fixed in the same commit; it came from joining two comment blocks without the separating newline, and it went out to every repository in this sweep, so the fix did too.

[kodizm]

Note

Kodizm (AI-generated). May contain mistakes; verify before acting.

Both findings from my last review are fixed and the new comment now matches the behaviour I measured; nothing else to raise.

Since 6e4a5ba the only change is the comment block above the ignore rule. The stray # is gone, and the git clean prose is now the right way round: -xdf skips a nested worktree with "Skipping repository ...", -xdff is the one that wipes it and leaves a registration for git worktree prune. That matches the reproduction in my previous review and yours. The ignore pattern itself is untouched.

Tests

Comment-only change on a config file, so no test surface, as the description says.

Checks I ran

• git check-ignore -v .claude/worktrees/foo → .gitignore:49:.claude/worktrees/; .claude/rules/x still not matched, so the scoping is unchanged and correct.
• git status --porcelain → clean; git ls-files .claude → 0, consistent with "Nothing is tracked there today."
• tail -c 1 .gitignore | od -c → \n, so the file still ends with a newline.
• I did not re-run the git clean -xdf / -xdff probe this round (the sandbox declined the temp-repo command); I am relying on the run from my previous review, which produced exactly the "Skipping repository" line the new comment quotes.