Skip to content

fcakyon/claude-codex-settings / plugins/github-dev/skills/commit-staged

commit-staged

Verified

This skill should be used when user asks to "commit these changes", "write commit message", "stage and commit", "create a commit", "commit staged files", or explicitly invokes "commit-staged".

FileKindSize
SKILL.mdskill md3.6 KB
scripts/verify-stagedexecutablescript1.0 KB
---
name: commit-staged
description: This skill should be used when user asks to "commit these changes", "write commit message", "stage and commit", "create a commit", "commit staged files", or explicitly invokes "commit-staged".
---

# Commit Staged

Complete workflow for creating commits following project standards.

When explicitly invoked with extra text, treat that text as additional context about the
changes and include it in commit planning and commit messages.
When session history includes findings, motivation, or rationale, include the strongest
points in the commit message body instead of relying on the diff alone.

## Process

**First, run the `/simplify` skill on the staged diff and apply its findings before committing, docs-only commits included.**

1. **Preferred execution**
   - If subagents are available, use `github-dev:commit-creator` for the full workflow.
   - Pass along any extra invocation text as additional context.
   - Otherwise follow the manual steps below.

2. **Analyze staged files only**
   - Check all staged files: `git diff --cached --name-only`
   - Read diffs: `git diff --cached`
   - When a relevant check could pick up unstaged or untracked code, use the bundled `scripts/verify-staged COMMAND [ARG...]` from the repository. It runs the command in a temporary export of the staged tree and prints its Git tree ID on success. Supply a command that builds or starts its own candidate there. Dependencies, ignored files, submodule checkouts and Git history are outside the export. Confirm a previously tested tree still matches `git write-tree` before reusing its result.
   - Completely ignore unstaged changes

3. **Commit message format**
   - First line: `{type}: brief description` (max 50 chars)
   - Types: `feat`, `fix`, `refactor`, `docs`, `style`, `test`, `build`
   - Use plain language. Avoid jargon, buzzwords, and repo shorthand unless an exact command or tool name is needed.
   - Use findings and motivation from session history when available.
   - Focus on 'why' not 'what'
   - 1 sentence conventional style + 1-2 short motivation/findings sentences if possible
   - For complex changes, add bullet points after blank line
   - After a blank line, end each commit message with one lowercase `verification:` trailer. Name the relevant checks actually completed for the staged changes and their results. If none ran, use `verification: none (<reason>)`. Include a material failure or blocker without claiming it passed. Keep future checks out of the trailer. If the project uses `project-qa`, its `run --verification` output supplies case IDs and observed statuses for the trailer. For a journey, include the environment and observed behavior, for example `verification: local Chrome mailbox refresh kept one sync request and manual sync started a second`. [Source](https://claude.dev/blog/how-the-claude-code-team-thinks-about-verification)

4. **Message examples**
   - `feat: add sign-in flow`
   - `fix: stop duplicate jobs on save`
   - `docs: add skills install snippets`

5. **Documentation update**
   - Check README.md for:
     - New features that should be documented
     - Outdated descriptions no longer matching implementation
     - Missing setup instructions for new dependencies
   - Update as needed based on staged changes

6. **Execution**
   - Commit uses HEREDOC syntax for proper formatting
   - Verify commit message has correct format

## Best Practices

- Analyze staged files before writing message
- Keep first line under 50 chars
- Use active voice in message
- Use simple words that still stay accurate
- Prefer session findings over repeating raw diff details
- One logical change per commit
- Ensure README reflects implementation
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 469kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall