agentleFS
epamepam/ai-dial-chat/.claude/skills/git-ship/SKILL.md

git-ship

Skill501 starsChanged 15 days agoApache-2.0View on GitHub

Use this skill whenever the user wants to commit, push, or ship changes in a git repository. Triggers include: "commit changes", "push my changes", "ship it", "commit and push", "create a branch and commit", "make a PR", "open a pull request", "create draft PR", or any variation of committing/pushing work. Always use this skill when the user mentions committing — even casually — as it handles the full cycle: branch → add → commit → push → (optionally) PR, with Conventional Commits format and automatic area detection.

  • Commits and pushes

4 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

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    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
  • could be automatic: a setting could enforce it
  1. Distinctive:

    If there are no changes at all — report and stop.

    line 71
  2. Very specific:

    If the area argument was given, use it.

    line 77
  3. Distinctive:

    If push fails (no permissions, rejected, conflict) — report the full output and stop.

    line 111
  4. Very specific:

    If the user requested a PR or draft PR, read .agents/skills/git-ship/examples/pr-body.md and follow the repo PR template …

    line 118
  5. Distinctive:

    Always finish with a concise confirmation:

    line 126

≈ “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

Git Ship Skill

Full cycle for ai-dial-chat: checkout new branch from development → stage → commit (Conventional Commits) → push → optional PR. Base branch is development.

Supporting files (read them when the step says so — don't inline everything up front):

  • .agents/skills/git-ship/areas.md — how to resolve the commit area (scope) for this repo.
  • .agents/skills/git-ship/examples/commit-messages.md — type table, message + branch examples.
  • .agents/skills/git-ship/examples/pr-body.md — PR body template, rules, and gh pr create usage.

Step 0 — Detect mode (new branch vs. update existing)

git rev-parse --abbrev-ref HEAD                 # current branch
gh pr view --json number,url,state 2>/dev/null  # open PR for current branch (if any)
  • New-branch mode — current branch is development (or another base branch), or there's no PR for it. Run the full cycle: Steps 1–6.
  • Update-existing mode — you're already on a feature branch and it has an open PR (or the user says "push to the existing PR" / "update my PR"). Do not create a new branch and do not run gh pr create. Stay on the current branch, then:
    1. Step 2 — review the changes.
    2. Step 4 — generate the commit message (reuse the existing PR's ticket; keep the same area/type).
    3. Step 5 — skip the git checkout lines; just git add → git commit → git push.
    4. Skip Step 6's gh pr create. Pushing updates the open PR automatically. Only refresh the body (gh pr edit --body) if the user asks or the description is stale/missing.
    5. Step 7 — report the existing PR link, not a new one.

Step 1 — Gather required context

  • Ticket number — use the ticket argument if passed. Else look in conversation context. If still not found, ask: "What is the ticket number?"
  • Area — if the area argument is passed, use it directly in Step 3.
  • Draft PR? — if the user said "draft", note it for Step 6.

This skill runs in a forked context (context: fork), so it does not see the main conversation. Prefer passing arguments: /git-ship 7432 catalog.


Step 2 — Understand the changes

git status
git diff HEAD

If there are no changes at all — report and stop.


Step 3 — Determine area

If the area argument was given, use it. Otherwise read .agents/skills/git-ship/areas.md and map the changed files to an area using its resolution rules. If the resolved area is new, follow that file's append-only self-extend rules so the taxonomy grows (the edit is staged in Step 5 and ships in the same commit).


Step 4 — Generate commit message

Format: <type>(<area>): <short description> (Issue #<ticket>).

Read .agents/skills/git-ship/examples/commit-messages.md for the type table, message examples, and the branch-name rule. Analyze the diff to pick the most accurate type and write a concise, imperative description. Breaking change → add ! before the colon.


Step 5 — Execute

# 1. Create branch from base (skip in update-existing mode)
git checkout development
git checkout -b <type>/<short-slug>

# 2. Stage all changes (includes any areas.md self-extend edit)
git add .

# 3. Commit
git commit -m "<type>(<area>): <description> (Issue #<ticket>)"

# 4. Push
git push origin <type>/<short-slug>

If push fails (no permissions, rejected, conflict) — report the full output and stop. Do not force push or rebase automatically.


Step 6 — Pull Request (if requested)

If the user requested a PR or draft PR, read .agents/skills/git-ship/examples/pr-body.md and follow the repo PR template (.github/pull_request_template.md) — fill every placeholder, then run the gh pr create command shown there with --base development.


Step 7 — Summary

Always finish with a concise confirmation:

Branch:  <type>/<short-slug>
Commit:  <type>(<area>): <description> (Issue #<ticket>)
Push:    ✅ succeeded  /  ❌ failed — <reason>
PR:      <link> (created)  /  <link> (updated existing)  /  skipped
Area:    <area>  (note "added to taxonomy" if Step 3 self-extended areas.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.