branch-and-commit

Branch and Commit

Automates the workflow of analyzing uncommitted changes, grouping them into atomic commits with proper notation, and creating a clean feature branch ready for PR.

When to Use This Skill

Use this skill when you have uncommitted changes and need to:

• Group changes into logical, atomic commits
• Apply proper risk notation (Arlo's Commit Notation)
• Organize commits on current branch OR create a new feature branch
• Prepare changes for pull request review

Do NOT use for:

• Branch splitting or rebasing existing commits
• Amending published commits
• Reorganizing pushed commits

Overview

This skill transforms messy working directory changes into a clean commit history by:

1. Analyzing all uncommitted changes (file types, diff content, intentions, security)
2. Interviewing you with in-depth questions to uncover missing changes and improvements
3. Grouping changes by intention and risk level (after understanding context from interview)
4. Ordering commits optimally (provably safe changes first, features last)
5. Deciding whether to use current branch or create new feature branch
6. Committing with proper commit messages
7. Preparing for PR creation (optional)

Workflow

Phase 1: Analysis (Read-Only)

The skill detects all changes and categorizes each file:

git status --porcelain
git diff HEAD --stat
git diff HEAD --cached --stat

For each changed file, it determines:

• File category: TEST, DOCS, CONFIG, SOURCE
• Intention: F (feature), B (bugfix), R (refactoring), D (docs), T (test), E (environment), A (automated), C (comment)
• Change nature: Import-only, comment-only, mixed changes

Security Checks:

Before proceeding, scan diffs for sensitive data:

• API keys: Long alphanumeric strings (32+ chars)
• AWS credentials: AKIA[A-Z0-9]{16} pattern
• Private keys: -----BEGIN.*PRIVATE KEY-----
• Passwords: Common variable names in plain text (like password = or PASSWORD =)
• Tokens: patterns like token = or auth_token = or bearer =