Claude Code Plugins

Community-maintained marketplace

Feedback

agileflow-commit-messages

@projectquestorg/AgileFlow
3
0

Formats git commit messages following Conventional Commits and respects user's attribution preferences from CLAUDE.md

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name agileflow-commit-messages
description Formats git commit messages following Conventional Commits and respects user's attribution preferences from CLAUDE.md

Conventional Commits Skill

Automatically formats git commit messages following the Conventional Commits specification while respecting the user's attribution policy defined in CLAUDE.md.

When to Use

This skill activates when:

  • Creating or staging git commits
  • Discussing code changes that need committing
  • Keywords: "commit", "git commit", "conventional commit", "commit message"

What This Does

  1. Checks CLAUDE.md for attribution policy (respects user preferences)
  2. Analyzes staged changes to determine commit type and scope
  3. Formats message using Conventional Commits format
  4. Shows message to user for approval before executing

Instructions

  1. Check attribution policy:

    • Read CLAUDE.md for "Git Commit Attribution Policy"
    • If "DO NOT ADD AI ATTRIBUTION" found → skip footer
    • Otherwise → prepare attribution footer

  2. Analyze changes:

    • Run git diff --cached --stat to see staged files
    • Determine primary change type (feat, fix, refactor, etc.)
    • Identify affected scope (api, ui, auth, db, config, etc.)

  3. Format message:

    <type>(<scope>): <subject>

<body>

<footer>

  4. Show for approval: Display formatted message (diff-first pattern)

  5. Execute after confirmation: Run git commit with user's YES

Conventional Commits Format

Types: feat, fix, docs, style, refactor, perf, test, chore, ci, revert

Scope (optional): (api), (ui), (auth), (db), (config), etc.

Subject rules:

  • Imperative mood: "add feature" not "added feature"
  • Lowercase, no period, max 50 characters

Body (optional):

  • Explain WHAT and WHY (not HOW)
  • Wrap at 72 characters per line

Footer (optional):

  • Issue references: Closes #123
  • Breaking changes: BREAKING CHANGE: description
  • Attribution: Add if CLAUDE.md allows

Output Examples

WITH attribution (default):

feat(auth): add two-factor authentication

Implemented TOTP-based 2FA with QR code generation
for enhanced user account security.

Closes #234

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

WITHOUT attribution (if CLAUDE.md specifies):

feat(auth): add two-factor authentication

Implemented TOTP-based 2FA with QR code generation
for enhanced user account security.

Closes #234

Breaking changes:

feat(api)!: change authentication endpoint structure

BREAKING CHANGE: The /auth/login endpoint now returns
a different response format.

Old: { token: "..." }
New: { accessToken: "...", refreshToken: "...", expiresIn: 3600 }

Closes #456

Quality Checklist

Before suggesting commit message:

  • Type is appropriate (feat/fix/refactor/docs/chore)
  • Subject uses imperative mood
  • Subject is lowercase, <50 chars, no period
  • Body explains WHY, wrapped at 72 chars
  • Attribution policy checked in CLAUDE.md (CRITICAL)
  • No secrets in staged files (.env, tokens, credentials)
  • Breaking changes marked with ! and BREAKING CHANGE: footer

Integration

  • agileflow-changelog: Commit messages inform changelog entries
  • agileflow-story-writer: Reference story IDs in commits (Closes STORY-042)
  • agileflow-adr: Link ADRs in footer (Refs: ADR-0005)

Notes

  • ALWAYS check CLAUDE.md first for attribution preferences
  • For large commits, suggest breaking into smaller logical commits
  • Keep line lengths: subject <50 chars, body <72 chars
  • Check for secrets before committing (.env, credentials.json, API keys)
  • Security check: Warn user if sensitive files are staged