Code Review and Quality: Your Agent's Pre-Merge Gate

Where this fits

Your agent just finished a feature branch and you are about to merge. You want a second pass beyond 'it compiles.' This skill slots into that exact moment: before any merge, it runs a structured code review across correctness, readability, architecture, security, and performance.

Why agents benefit

• Agents need an explicit checklist, not vibes. The five axes give the model stable dimensions to evaluate, reducing random omissions.
• The approval standard ('definitely improves overall health') teaches agents to stop gold-plating and approve incremental improvements, matching human reviewer norms.
• It works on code from any source — yourself, another agent, or a human — so it is useful in multi-agent workflows.
• It catches regressions after bug fixes by reviewing the fix and the test together, not just the patch.
• Quality gates are enforceable: you can wire it to a CI step or use it as a pre-merge prompt for your agent.

Practical scenarios

Before merging a PR from another agent

A teammate agent submits a large PR but its summary is thin. Run this skill and ask for a five-axis report. You get a structured verdict, plus specific line-level concerns you can send back for fixes.

After completing a feature with your own agent

Your agent implemented the code and wrote the tests. Instead of trusting its own 'looks good,' have it invoke the skill in self-review mode. It will often find things — dead code, missing edge cases, or a performance trap — that the implementation pass skipped.

When evaluating a human's refactoring

A human developer refactored a critical module and wants a second opinion. Ask your agent to review the diff with this skill. It won't replace a human reviewer, but it gives objective notes on architecture and readability that the human can quickly verify.

Add it to your agent workflow

Install with one command:

npx skills add addyosmani/agent-skills --skill code-review-and-quality

Then instruct your agent, e.g.:

Use code-review-and-quality on the current diff. Return a verdict per axis and a final approve/request-changes decision.

Compare before adopting

Before adding it, look at the skill's quality signals: how recently the repository was updated, how many stars it has, and whether the skill's output in community examples is genuinely specific. Compare it with alternatives like a generic 'review this code' prompt or a linter-only setup. Also assess workflow fit: does your agent already run tests and static analysis? This skill complements those, not replaces them.

Why it is worth tracking

With over 90k stars on GitHub, this skill's repository has strong community momentum. The multi-axis review format is a practical pattern for evaluation-heavy agent tasks, and its approval guidance nudges agents to behave like experienced reviewers instead of pedantic critics. For more context, see the OpenAgentSkill listing at https://www.openagentskill.com/skills/addyosmani-code-review-and-quality. Worth evaluating now if you ship agent-written code to main branch; worth tracking even if you are not yet ready to adopt.