commit_skill

Overview

This skill analyzes code changes and generates clean, conventional commit messages. It identifies change types and scopes, runs quality checks for sensitive content and TODOs, and can create commits without adding AI tool signatures to history. The goal is reliable, professional commit messages that fit team rules.

How this skill works

The skill scans the diff to detect changed files, file types, and affected modules to infer scope. It classifies the change as feat/fix/docs/etc., composes a conventional-format message, and optionally runs configured quality checks. When requested, it can execute a commit using the generated message while ensuring no AI footprint is inserted.

When to use it

• Preparing a well-formed commit message from staged or unstaged changes
• Enforcing commit message conventions across a team or CI pipeline
• Running pre-commit quality checks for secrets, TODOs, or policy violations
• Automating single-command commits for straightforward changes
• Auditing changes to determine affected modules and impact

Best practices

• Keep generated messages concise and review scope suggestions before committing
• Enable qualityChecks in CI to catch secrets or TODOs early
• Use autoScope for consistent scope detection, but override manually when needed
• Prefer conventional messageFormat for tooling compatibility (changelog, release tools)
• Run analyze before create to validate the suggested message and checks

Example use cases

• Developer stages multiple files and runs analyze to get a conventional commit message suggestion
• CI pipeline invokes check to validate PR commits for secrets and missing scopes
• Automated bot composes a commit message for dependency bumps and creates the commit without AI attribution
• Repo maintainers enforce commit convention by failing CI when qualityChecks report issues
• Feature branch work where autoScope identifies the module impacted and generates the appropriate scope