agentleFS
helmut-hoffer-von-ankershoffenhelmut-hoffer-von-ankershoffen/heinzel/.claude/skills/ship-small/SKILL.md

ship-small

Skill2 starsChanged 16 days agoMITView on GitHub

Ship a small, low-risk change (a copy or docs fix, a small web UI tweak) from intent to a validated pull request with one command per stage, safe for a small model (Haiku). Use for typos, wording, doc fixes, link fixes and small component or style tweaks; anything else escalates by itself.

  • Commits and pushes

2 other projects use 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, 0 are distinctive, and 3 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.

  • 3

    Very specific

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

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

    3 STOP: this is a correct outcome, not an error.

    line 25
  2. Very specific:

    Never write a name of a person, employer, customer or private repository into a file.

    line 33
  3. Very specific:

    If the change turns out bigger than the job you were given, run S escalate "one sentence why" and stop.

    line 35

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Ship a small change

A script decides what is small and what is not; you never judge. Run the stages in order. Every command prints one JSON line; do what its next field says, literally. S below is the uv run --script .../scripts/ship.py prefix that next prints; always cd into your worktree first (your shell forgets it between commands).

StageCommandDoes
1S start SLUGNew worktree and branch (SLUG: 3-41 chars, lowercase, digits, dashes)
2S classify --files PATH...Before editing: is this file allowed? Exit 0 = edit it, exit 3 = not allowed
3edit the filesSmallest edit that does the job. No new or deleted files
4S classifyReads your real diff: lane low or STOP
5S checkChecks for exactly your files; fixes whitespace and formatting itself
6S pr "type(scope): summary" --body "why"Commits, pushes, opens the PR labelled risk: low, auto-merge off
7S waitWaits for CI; add --budget 540 and a 600000 ms shell timeout to wait longer
8S merge PRMerges your pull request once CI OK is green at its exact head; it re-checks the diff itself and refuses anything not small
9S verifyAfter the merge: Docs run green and the live page shows your text

S <stage> --help prints a stage's usage, an example and the exit codes. Exit codes: 0 go on with next. 1 this stage failed: fix exactly what detail names, then repeat the stage. 3 STOP: this is a correct outcome, not an error. Send the JSON line to whoever started you and run nothing else. 4 still running: run the same command again. 2 you used the command wrongly: read msg.

Rules that no command can enforce:

  • The second failure of any stage stops the change by itself (exit 3). Never retry a third time, never work around.
  • Merge only with S merge, never touch auto-merge, never run git push, git commit, gh pr merge or make pr yourself, never use --no-verify. The stages do all of it; a pull request that is not small is the orchestrator's.
  • Never write a name of a person, employer, customer or private repository into a file. If check reports a confidential term, reword the text and never repeat the term anywhere.
  • If the change turns out bigger than the job you were given, run S escalate "one sentence why" and stop.

The rules (paths, diff checks, limits) live in scripts/ship.py; background in docs/SHIP_SMALL.md.

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.