agentsclimarketplace

Code author

Skill kinhluan/rules-quarkus-skills/.agent-skills/code-author

Expert skill for writing high-quality code changes (CLs/PRs) following Google's Engineering Practices. Focuses on small changes, clear descriptions, and professional feedback handling.From its SKILL.md

Install
npx -y skills add kinhluan/rules-quarkus-skills --skill code-author

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
  • fetches URLsInstructs the agent to fetch 3 URLs, including https://google.github.io/eng-practices/review/developer/small-cls.html and 2 more.

SKILL.md

4.2 KB, 935 tokens by cl100k_base, as published. Nobody here has run it

code-author ✍️

Keyword: author | Platforms: gemini,claude,codex

Expert AI Agent Skill for Code Authoring - Best practices for creating, describing, and refining code changes based on Google's Engineering Practices.

🎯 Core Mandates (The Author's Rules)

  • Small CLs Only: Keep changes focused on ONE thing. Small CLs are easier to review, faster to merge, and lower risk.
  • Clear Descriptions: Always explain WHAT is changing and WHY. The "How" should be evident from the code, but the "Why" is critical context.
  • Risk Areas Identification: Explicitly call out parts of the code where you are unsure or where there is a high risk of side effects.
  • No Refactor + Feature Mix: Never mix refactoring with new features or bug fixes. These must be separate CLs.
  • Proactive Communication: Address all reviewer comments, even if you disagree. Explain your reasoning clearly and professionally.
  • Self-Audit First: Run the "Self-Audit Checklist" before sending a CL for review.

πŸ›  Authoring Workflows

1. Planning a Change (CL)

  1. Isolate the Task: Can this be broken into smaller, independent steps?
  2. Draft the Description: Start with a one-line summary. Add context on the problem being solved.
  3. Identify Risks: Note any areas that need extra attention from the reviewer.

2. Writing the Code & Self-Audit

  • Tests First: Ensure the change is covered by automated tests.
  • Style Consistency: Follow Google Java Style.
  • Self-Audit Checklist:
    • Diff Minimization: Revert accidental changes (whitespace, commented-out code).
    • Documentation Check: Update README, JSDoc, or JavaDoc for new exports/logic.
    • Test Alignment: Ensure tests cover logic branches, not just the "happy path."
    • Logic Validation: Mentally trace the execution path for edge cases (null, empty, etc.).

3. Handling Reviewer Feedback

  • Be Grateful: Reviewers are spending time to improve your code.
  • Don't Take it Personally: The goal is to improve the code, not critique the author.
  • Resolve or Disagree: Either apply the change or explain why the current approach is better. Never ignore a comment.

πŸ” CL Description Template (Superpowers Edition)

[Summary Line: Brief one-line description]

### πŸ“ Context & Intent
[Why is this change necessary? What problem does it solve?]

### πŸ›  Key Changes
[Bullet points of the main code modifications]

### ⚠️ Risk Areas & Concerns
[Where should the reviewer focus? Any potential side effects or uncertainties?]

### βœ… Verification Results
[List tests run, coverage achieved, and manual verification steps]

[Related Issues: e.g., Fixes #123]

🌐 Knowledge Sources & Deep Dives

Directive: Use web_fetch to find Google's specific guidance on "Small CLs" or "CL Descriptions" if a change feels too large or complex.

Skill Interoperability

The code-author ✍️ skill ensures that work produced by other skills is packaged correctly:

  • java-expert β˜• & quarkus-expert ⚑: Logic produced is split into reviewable CLs.
  • refactoring-expert πŸ› : Ensures refactors are isolated from feature changes.

What ships with it: 1 file

733 B alongside SKILL.md

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.