Guide · Agent skill
How to use the Requesting Code Review agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The requesting-code-review skill teaches an AI coding agent to hand its finished work to a separate reviewer helper, called a subagent, before the work is merged. It is written for developers who run agents through multi-step coding tasks and want a second check. After this guide you will know what the skill does, how to install it, how to phrase a request that triggers it, and what to watch for.
Requesting Code Reviewobra/superpowers · MIT Download the skill Quick start
- 1Download the ZIP. Download the requesting-code-review ZIP from our page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use, or into your project's .claude/skills/ folder.
- 3Confirm git history. Make sure your project is a git repository with commits, since the skill relies on git rev-parse and git merge-base.
- 4Finish a task. Let the agent complete a task or feature in your repository.
- 5Ask for a review. Use a natural request such as asking for a code review before merging.
- 6Act on the feedback. Fix Critical and Important items, note Minor items, and push back with reasoning if the reviewer is wrong.
What the skill makes the agent do
When the skill is installed, the agent can ask a reviewer subagent to check its completed work against the requirements. The reviewer receives a short description of the change, the requirements or plan, and a git commit range. It does not receive your full chat history, so the review stays focused on the code.
The agent then sorts the reviewer's feedback into Critical, Important and Minor items. According to the skill, Critical issues are fixed immediately, Important issues are fixed before the agent moves on, and Minor issues are noted for later. If the agent believes the reviewer is wrong, it is directed to push back with technical reasoning.
Install the requesting-code-review skill
Download the ZIP file from our page and unzip it. You can install the skill in one of two places. For a personal setup, unzip it into ~/.claude/skills/. For a single project, unzip it into the project's .claude/skills/ folder.
If you use the Claude apps instead, open Settings, go to Capabilities, and upload the skill under Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills
Check that the requirements are met
The skill depends on two things. First, your project must be a git repository with commit history, because the skill uses git rev-parse and git merge-base to find the commit range. Second, your agent environment must be able to dispatch subagents. Claude Code is the example the source gives.
- A git repository with commit history
- An agent environment that supports dispatching subagents (for example Claude Code)
Trigger the skill with natural requests
You do not need a special command. The skill's description tells the agent to use it when completing tasks, implementing major features, or before merging. A plain request is usually enough. The skill also covers optional moments such as when the agent is stuck, before a refactor, or after fixing a complex bug.
- Review what you just built before we go on to the next task.
- Request a code review of the changes since main before I merge.
- I'm stuck on this bug. Get a fresh review of the current changes.
- Before you refactor this module, check the current code against the plan.
Example of a review request
In the skill's example, the agent has just finished a task, finds the commit where the previous task ended, and records the current commit. It then dispatches the reviewer with a description and the requirements. The reviewer returns strengths, a list of issues by severity, and an assessment of whether the work is ready to proceed.
BASE_SHA=$(git log --oneline | grep "Task 1" | head -1 | cut -d' ' -f1)
HEAD_SHA=$(git rev-parse HEAD)
[Dispatch code reviewer subagent]
DESCRIPTION: Added verifyIndex() and repairIndex() with 4 issue types
PLAN_OR_REQUIREMENTS: Task 2 from docs/superpowers/plans/deployment-plan.md
BASE_SHA: a7981ec
HEAD_SHA: 3df7661Understand the reviewer template
The skill includes a reviewer prompt template, code-reviewer.md. The agent fills in placeholders for the description, the requirements, the base commit and the head commit. The template asks the reviewer to run git diff --stat and git diff over that range. The template also tells the reviewer that the spec is a vision document and does not list every input or condition, so the reviewer should not treat every unlisted case as a defect.
- {DESCRIPTION}: brief summary of what was built
- {PLAN_OR_REQUIREMENTS}: what it should do
- {BASE_SHA}: starting commit
- {HEAD_SHA}: ending commit
Safety notes
The source states that this skill does not run scripts and does not use the network. Its work is done through git commands and a subagent. Because the reviewer is an AI, its output is an opinion and can miss issues or be wrong, so do not treat a clean review as proof that code is safe to ship.
- Runs scripts: no
- Uses network: no
- Reads git history and diffs in your repository
Tips for better reviews
Write a clear description and requirements. The reviewer only knows what you give it, so vague input produces vague feedback. Keep commit ranges tight, so the diff covers only the work under review.
- Name the plan file or requirements explicitly
- Review after each task in subagent-driven development, not only at the end
- Read the Critical and Important items yourself before accepting a fix
What people use it for
- Checking each task in a multi-step feature before the agent starts the next one
- Reviewing all changes on a feature branch against requirements before merging to main
- Getting a fresh look at code when the agent is stuck on a bug
- Checking the current code against a plan before a refactor
Frequently asked questions
Is the requesting-code-review skill free?
The skill is published under the MIT licence, as stated on this page. Check the repository's licence file for the full terms before you reuse it.
What do I need to use the requesting-code-review skill?
You need a git repository with commit history, because the skill uses git rev-parse and git merge-base. You also need an agent environment that can dispatch subagents, such as Claude Code.
Where do I install the requesting-code-review skill?
Unzip the downloaded ZIP into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps, upload it under Settings → Capabilities → Skills.
Why does the skill need git commit SHAs?
The reviewer needs a base commit and a head commit to see the change. Without them the reviewer cannot see the change, so the skill does not work in a project without git.
Does the skill run scripts or use the internet?
According to the source, it runs no scripts and uses no network. It works through git commands and dispatching a reviewer subagent.
Can I use this instead of a human code review?
No. The reviewer's output is an AI opinion and can miss issues or be wrong, so it should not replace human review on high-stakes code.
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
This skill suits developers who use AI agents for multi-step coding in a git repository and want a second check before merging or between tasks. It is a poor fit for projects without git, non-coding documents, or high-stakes code that needs human review as the final check.
Download the skill