agentleFS
alunduilalunduil/alunduil-chezmoi/CLAUDE.md

alunduil-chezmoi

CLAUDE.md2 starsChanged 4 days ago0BSDView on GitHub

Chezmoi source directory. Files deploy to $HOME via chezmoi apply; names follow chezmoi rules (dot_ → ., executable_ → +x, .tmpl → Go template, .chezmoiscripts/run_*_before_NN-… → ordered convergent bootstrap). docs/tutorials/bootstrap.md has the bootstrap walkthrough; docs/explanation/architecture.md has the human-facing rationale; docs/reference/bootstrap-reference.md has the lookup tables. chezmoi diff/apply read the applied clone at ~/.local/share/chezmoi, not this working tree. Edits here don't take effect on apply until committed and pulled into the apply clone. Use chezmoi diff --source-path . to preview from this…

  • Commits and pushes

The file

We split this CLAUDE.md into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Write a version pinned inside a URL (the zellij plugins block) out in full and leave it unannotated: Renovate bumps every copy of it.

    line 28
  2. Very specific:

    Every *_VERSION pin carries a # renovate: datasource=… depName=… line directly above it (order: datasource, depName, packageName, …

    line 30

≈ “Common” is an estimate: at least 1 in 5 of the 28,029 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

alunduil-chezmoi

Chezmoi source directory. Files deploy to $HOME via chezmoi apply; names follow chezmoi rules (dot_ → ., executable_ → +x, .tmpl → Go template, .chezmoiscripts/run_*_before_NN-… → ordered convergent bootstrap). docs/tutorials/bootstrap.md has the bootstrap walkthrough; docs/explanation/architecture.md has the human-facing rationale; docs/reference/bootstrap-reference.md has the lookup tables.

Source vs. apply path

chezmoi diff/apply read the applied clone at ~/.local/share/chezmoi, not this working tree. Edits here don't take effect on apply until committed and pulled into the apply clone. Use chezmoi diff --source-path . to preview from this checkout.

Invariants

  • Bootstrap passes in .chezmoiscripts/ converge on host state. Read docs/reference/bootstrap-reference.md before adding a pass, a tool, or a guard, or changing a pass's prefix, phase, or number.
  • The shellcheck hook reads .sh.tmpl files unrendered, so a {{ … }} expression must sit inside quotes or a comment. That is why the package lists arrive via read -ra <<<'{{ … }}' rather than an array literal.
  • Pinned tools live in dot_config/mise/config.toml. After editing it, run just lock-tools and commit mise.lock with the npm sidecars under dot_config/mise/.mise/locks/. chezmoi skips that dot-directory, so symlink_dot_mise.tmpl points ~/.config/mise/.mise at it in the source tree.
  • Write a version pinned inside a URL (the zellij plugins block) out in full and leave it unannotated: Renovate bumps every copy of it.
  • Every *_VERSION pin carries a # renovate: datasource=… depName=… line directly above it (order: datasource, depName, packageName, versioning, extractVersion). One generic manager in renovate.json reads them all — a new pinned tool needs no Renovate config change. An unannotated pin is invisible to Renovate rather than an error, so script/checks/renovate-pins (a prek hook) fails the build on one.
  • The repo requires every action pinned to a full-length SHA, including the actions a third-party action calls. Renovate's digest pinning stops at our own uses: line. Before adopting or bumping a third-party action, read its action.yml at the new SHA and confirm every uses: under runs.steps is a full-length SHA.
  • dot_local/bin/executable_gh shadows system gh to enforce --draft on gh pr create. PRs Claude opens go through this wrapper.

Pull request descriptions

This repo squash-merges with squash_merge_commit_message: PR_BODY, so the description lands in git log verbatim and the branch commits go with the squash. Write it as a commit message: prose wrapped at 72 columns, no H1 repeating the title, no ## section headings, no checkbox lists. A trailer that has to persist goes in the description, since the ones on branch commits are discarded.

The pr-create skill still composes Summary/Gotchas/Verification headings; #778 tracks the divergence. This file wins here.

Sensors

CI is authoritative. Run all sensors locally before claiming done:

just check                     # runs every sensor below, reports all failures

Each runs in its own CI workflow and can be invoked alone:

prek run --all-files           # shellcheck, shfmt, ruff, mypy, check-json
bats --recursive dot_local dot_claude etc script  # shell unit tests
just check-python              # python unit tests (pytest via uv)
script/checks/zellij-config    # zellij KDL validation (needs zellij)
script/checks/theme-contrast   # theme colors meet WCAG AAA
script/checks/chezmoi-apply    # apply round-trip (needs chezmoi + age)

script/checks/fresh-host bootstraps a systemd container from scratch and fails on anything a second apply changes. It installs the whole workstation, so just check leaves it to CI.

Two CLAUDE.md files

  • This file: rules for AI editing the chezmoi source.
  • dot_claude/CLAUDE.md → deploys to ~/.claude/CLAUDE.md. Edits there change Claude's host-wide behaviour on next chezmoi apply.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.