agentleFS
SajidisrarSajidisrar/ios-claude-skills/skills/ios-commit-and-pr/SKILL.md

ios-commit-and-pr

Skill1 starsChanged 14 days agoMITView on GitHub

Write commits and pull request descriptions for this iOS project to the team's conventions, including ticket linkage and an AI-code review gate. Use when asked to commit, stage changes, write a commit message, open or describe a pull request, or summarise a change for review.

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
  1. Very specific:

    Never commit to the default branch.

    line 10
  2. Distinctive:

    Never commit without being asked to.

    line 13
  3. Very specific:

    Check git status and git diff --staged and read what you are about to commit.

    line 15
  4. Distinctive:

    Not this: "fixed stuff", "WIP", "address review comments" (say what changed), or a message that restates the diff line by line.

    line 47
  5. Distinctive:

    Treat AI-generated code as untrusted until a named human has reviewed it and taken ownership, and require a clean test run and SAST pass …

    line 89

≈ “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

Commit and pull request conventions

Before committing anything

  • Never commit to the default branch. Branch first: feature/TASK-123-offline-sync, fix/TASK-456-stuck-spinner, chore/lint-config.
  • Never commit without being asked to. Producing the change and committing it are separate decisions.
  • Check git status and git diff --staged and read what you are about to commit. Watch for: a stray print, a .xcodeproj (git-ignored — if it appears, the ignore file was bypassed), a secret in an xcconfig, a fixture with real data, a debug flag left on.

Commit messages

<type>(<scope>): <imperative summary, under 72 characters>

Why this change is needed, if it is not obvious from the summary.
What was decided and what was deliberately left out.

Refs: TASK-123

Types: feat, fix, refactor, test, docs, chore, perf, build. Scope is the feature or Core area: widgets, auth, networking, sync.

Good:

feat(widgets): queue widget creation while offline

Users on a poor connection lose signal mid-session and were losing
unsaved creates. Creates now fall back to the outbox and appear
immediately with a pending badge. Conflict handling is still
last-write-wins — see docs/DECISIONS.md D12.

Refs: TASK-123

Not this: "fixed stuff", "WIP", "address review comments" (say what changed), or a message that restates the diff line by line.

One logical change per commit. A refactor and a behaviour change in one commit cannot be reviewed or reverted separately.

Pull request description

## What
One or two sentences. What now works that did not before.

## Why
The problem, the task or ticket, and the decision if one was made.

## How
The approach, and anything a reviewer would otherwise have to reverse-engineer.
Name any layer boundary the change touches.

## Testing
- Unit tests added: <which behaviours, including failure paths>
- Manual: <device, iOS version, and the states checked — including offline,
  expired session, empty and error states where relevant>

## Risk and rollback
What could go wrong, who is affected, how to revert.

## Checklist
- [ ] Builds for iOS 16 simulator
- [ ] SwiftLint clean
- [ ] Tests pass locally and in CI
- [ ] SAST clean
- [ ] No secrets, credentials or personal data in the diff
- [ ] Accessibility checked (VoiceOver + largest text size) for UI changes
- [ ] Docs updated (ARCHITECTURE / DECISIONS / SECURITY) if a decision changed
- [ ] Named human reviewer assigned

Attribution and the AI-code gate

Where a change was written with AI assistance, say so in the pull request body — one line is enough ("Drafted with Claude Code; reviewed and tested by <name>"). Treat AI-generated code as untrusted until a named human has reviewed it and taken ownership, and require a clean test run and SAST pass before merge — adjust this to whatever your own project's policy requires. Do not describe a change as "ready to merge" on your own authority — describe it as ready for review.

Changes touching authentication, token storage, transport or logging need a second reviewer. Say so in the description rather than leaving it to be noticed.

Verification

  • Branch is not the default branch.
  • One logical change per commit.
  • Message says why, and references the task or ticket.
  • Diff read through before staging; no debug leftovers, no secrets, no real data.
  • Pull request body includes testing done and risk.
  • AI assistance disclosed where applicable.
  • Second reviewer flagged for security-relevant changes.

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.