Guide · Agent skill
How to use the Code Review Workflow agent skill (omc-review)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The omc-review skill, published in the oh-my-claudecode project as skills/review, teaches an AI agent to act as a reviewer for work that already exists. It reads a change against what it was meant to do, checks it for defects, risk and simplification, and reports findings from most to least severe. This guide covers what the skill does, how to install it, how to ask for a review in plain language, and where its limits are.
Code Review WorkflowYeachan-Heo/oh-my-claudecode · MIT Download the skill Quick start
- 1Download the ZIP. Get the omc-review skill ZIP from the page where you found this guide.
- 2Unzip into a skills folder. Extract it into ~/.claude/skills/ for personal use, or into the project's .claude/skills/ folder.
- 3Or upload it in the Claude apps. Go to Settings → Capabilities → Skills and upload the skill.
- 4Make the change available. Have the branch, pull request or diff you want reviewed checked out in the repository.
- 5Ask for a review in plain language. Describe what to review and what it was meant to do, such as a ticket or spec.
- 6Read the findings. Work through findings most-severe first, checking the file, line and failure scenario for each.
- 7Decide what to act on. Treat the review as advice. Fix what you agree with and keep your own release and security gates in place.
What does the omc-review skill make the agent do?
The skill is for evaluating finished work before it ships. The source states that review never authors the change it is judging, so the agent reads and critiques code but does not write the fix.
The agent follows a fixed workflow: establish what changed and what it was meant to do, read the change against that intent, check correctness first, then risk, then simplification, verify each candidate finding, and report findings most-severe first.
- Correctness: logic defects, edge cases, error paths, concurrency
- Risk: security boundaries, destructive operations, data integrity
- Reuse: existing utilities or patterns the change should have used
- Simplification: code that could be deleted or collapsed
- Coverage: behavior that ships untested
How do you install the omc-review skill?
Download the skill ZIP from the page where you found this guide, then unzip it into one of two locations. Use ~/.claude/skills/ to make the skill available across your projects, or use a project's .claude/skills/ folder to limit it to one repository.
If you use the Claude apps instead of files on disk, you can upload the skill under Settings → Capabilities → Skills.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload
How do you trigger the review with a natural request?
You do not need a special command. Describe the review you want in plain language and name the change. Mentioning the intent helps, because the skill compares the diff against what the work was meant to do.
Good requests say what to review and why. For example, ask the agent to review the current branch against the ticket it implements, or to check a pull request for security problems before merging.
- Review the diff on this branch against the ticket description and tell me what's wrong.
- Do a review of this pull request before I merge it. Focus on correctness and error handling.
- Check this change against our CONTRIBUTING.md and CLAUDE.md standards.
What does a review look like for a large change?
For a substantial diff, the skill describes running two separate sub-agent passes. One pass checks the diff against the repository's documented standards. The other checks it against the task, spec or ticket acceptance criteria.
The two passes are reported separately and are not merged into one ranking. The source explains that a change can pass one axis and fail the other, for example being standards-conforming but wrong in behavior. A small diff stays a single pass.
- Standards axis: repo standards documents such as CLAUDE.md, AGENTS.md, docs/standards/, CONTRIBUTING.md or CODING_STANDARDS.md, where they exist
- Intent axis: the task, spec or ticket acceptance criteria, with each finding quoting its source line
- Each axis gets its own verdict
What output should you expect?
Each finding should include the file, the line and a concrete failure scenario. The report also lists what was checked and found clean, and anything the agent could not assess. If the work is sound, "no findings" is a valid result.
Findings are verified before they are reported, so the list is meant to avoid plausible-sounding guesses that do not reproduce.
- Findings, most-severe first, with file, line and concrete failure scenario
- What was checked and found clean
- Anything that could not be assessed
What do you need before you start, and what are the safety notes?
The skill needs an AI agent runtime that loads SKILL.md skills, and access to the repository and diff being reviewed. The source says the skill assumes sub-agent support for large diffs.
According to the source, the skill runs no scripts and uses no network. It is advisory by default: it informs the release decision and does not fail builds. The SKILL.md says hard gates for release, security or destructive operations stay separate and fail closed.
- Requires: an agent runtime that loads SKILL.md, repository access, the diff under review
- Sub-agent support is assumed for large diffs
- Runs scripts: no
- Uses network: no
- Advisory: it does not block a release
What people use it for
- Second look at a pull request or diff before merging
- Checking a change against CLAUDE.md, AGENTS.md or CONTRIBUTING.md standards
- Confirming a change meets the acceptance criteria in a ticket or spec
- Spotting untested behavior and error paths in a feature branch
- Finding code that could be deleted or reused from an existing utility
Frequently asked questions
How do I install the omc-review skill?
Download the skill ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
What do I need to run the omc-review skill?
You need an AI agent runtime that loads SKILL.md skills and access to the repository and diff you want reviewed. The skill assumes sub-agent support for large diffs. It runs no scripts and uses no network, according to the source.
Is the omc-review skill free?
The source lists the project's licence as MIT. It does not state pricing for the skill, so check the project page for licence terms before relying on it.
Can omc-review fix the code for me?
No. The skill is explicitly review-only and never authors the change. It reports findings so you or another tool can make the fix.
Will omc-review block a release or fail my build?
No, not by default. The skill is advisory and informs rather than gates. Hard release, security and destructive-operation gates are meant to stay separate from it.
How is omc-review different from a linter or a normal code review?
The skill skips anything tooling already enforces in its standards pass and focuses on judgement calls. It checks the diff against both repo standards and the stated intent, and each finding includes a file, line and concrete failure scenario. The source does not compare it with specific other tools.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use omc-review if you want a structured second look at a diff or pull request, especially when your repo has written standards or a clear ticket to check against. Pick something else if you need automated release or security gating, want the agent to write the fix, or are reviewing non-code documents with no diff or standards to compare against.
Download the skill