agentleFS
fmindfmind/dot/.agents/skills/dot-release/SKILL.md

dot-release

Skill13 starsChanged 4 days agoMITView on GitHub

Prepare, publish, recover, and verify fmind/dot releases through the repository task.

  • Commits and pushes

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use the checkout's release task as the single owner of preparation and publication.

    line 15
  2. Distinctive:

    Qualify first: run dot-verify on the checkout and resolve its findings.

    line 20
  3. Very specific:

    Run the owner: use the commands below from the repository.

    line 23

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill 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
SKILL.md

Dot Release

Use the checkout's release task as the single owner of preparation and publication. The global release guide owns generic versioning and publication verification; this skill owns dot's preconditions and recovery.

Workflow

  1. Confirm authority: the release command commits, pushes, and refreshes the installed CLI. An explicit /dot-release invocation or release request authorizes the whole flow below, including verification fixes, commits, the push to main, and the delivery fixes and follow-up release below; loading this skill for reference, review, or reconciling a past release does not.
  2. Qualify first: run dot-verify on the checkout and resolve its findings. Stop and report instead of releasing when a finding needs the user's decision or a gate stays red.
  3. Commit the candidate: group verified work into logical Conventional Commits (the changelog is generated from them), then git fetch and push main so HEAD equals upstream; pre-push hooks rerun the network checks and tests. If upstream moved, integrate it without rewriting history only when its commits do not overlap the candidate; otherwise stop.
  4. Know the preconditions: the task stops before writing unless dprint, git, git-cliff, mise, and uv exist, the tree is clean on main, HEAD equals the fetched origin/main, and the next tag is absent locally; without new Conventional Commits it releases nothing. Preserve unrelated work when a precondition fails.
  5. Run the owner: use the commands below from the repository; the task runs from source (uv run --frozen --directory dot python -m dot_tasks.release), not the possibly stale installed CLI.
  6. Read the result: the task bumps dot/pyproject.toml, CHANGELOG.md, and dot/uv.lock (only those files may change), formats the changelog with dprint (CD rejects a tree its formatter changes, so this never depends on Git hooks), gates the bumped commit (check:network, which refreshes the vulnerability data CD rechecks, test:starters, which re-resolves unlocked upstream packages since CI ran, build, and the host-only check:completions), then commits with the Lefthook output visible, tags, pushes both with git push --atomic, and runs mise run deploy.
  7. Verify delivery: the tag triggers cd.yml: a read-only gate job reruns mise run all and extracts the git-cliff notes; a publish job holding only contents: write, with no checkout, creates the GitHub release. Confirm with gh run list --workflow cd.yml --branch <tag> and gh release view <tag>. Local command success does not prove CD completion.
  8. Monitor and fix CI/CD: watch the release commit's CI run (including the macOS job) and the tag's CD run to completion with gh run watch <id> --exit-status. On failure, read gh run view <id> --log-failed, fix the root cause forward on main (never rerun to hide a flake), push, and watch CI again; a failed CD gate then needs the next release per Recovery. Repeat until CI is green and gh release view <tag> succeeds; stop and report when a fix needs the user's decision.
mise run release        # interactive release (after dot-verify)
mise run release -- -y  # non-interactive, within an authorized release

Recovery

Inspect git status --short, the release commit, local tag, and remote state before retrying. release.py owns the flow; its failure cases are exercised in test_release.py.

Failure boundaryNext action
Before the release commitChanges stay in the worktree and index, including concurrent edits. Review git diff and git diff --cached, preserve the intended work, and fix the cause before preparing again.
Local tag already existsPreflight stops before writing files. Delete the leftover tag with git tag -d <tag> after confirming it never reached the remote, then rerun.
Commit created, tagging failedNothing reached the remote. Fix the cause, then run the tag, push, and deploy commands the error prints.
Commit and tag created, atomic push failedInspect remote state before retrying the printed push and mise run deploy. If upstream moved, preserve the local release commit and reconcile the divergence before retrying.
Push accepted, installation refresh failedVerify the remote commit, tag, and CD independently, then retry mise run deploy. An installation error does not undo publication.
CD gate failed on the tagNo release was created. Fix forward on main and release the next version; do not move published tags or rewrite history as an automatic repair.

Documentation

  • Releases: fmind/dot · changelog
  • Companion skills: dot-verify (pre-release qualification), dot-development (implementation and installation proof), commit messages (commit grammar).

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.