Guide · Agent skill
How to use the Code Review Skill agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The Code Review Skill is an agent skill from the agno-agi cookbook that gives an AI agent a fixed, checklist-driven process for reviewing code. This guide explains what an agent skill is, how to install this one, how to ask for a review, and where its limits are. By the end you will be able to run a first-pass review of Python code and know when a human or a full linter should take over.
Code Review Skillagno-agi/agno · Apache-2.0 Download the skill Quick start
- 1Download the ZIP. Download the code-review skill ZIP from this page and unzip it.
- 2Choose where to install. Place the code-review folder in ~/.claude/skills/ for personal use or in .claude/skills/ for a single project.
- 3Or upload it in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill.
- 4Check Python 3. Make sure Python 3 is available, since the style script needs it.
- 5Ask for a review. Paste your Python code and request a review, mentioning style, bugs or linting.
- 6Read the severity groups. Work through Critical items first, then Important, then Suggestion.
- 7Confirm with a linter. Run ruff or flake8 for complete linting and get a human review for security-sensitive code.
What is an agent skill and what does this one make the agent do?
An agent skill is a folder containing a SKILL.md file with instructions the agent can load when a task matches them. This skill tells the agent to follow a fixed review process rather than improvising.
The agent first consults a bundled Python style guide, then runs a small Python script that flags simple style problems. After that it looks for bugs, security issues and performance problems, and reports its findings at three severity levels.
- Critical: must fix before merge, such as security vulnerabilities or bugs that cause crashes.
- Important: should fix but not blocking, such as performance issues or code smells.
- Suggestion: nice-to-have improvements, such as naming, documentation or minor refactoring.
How do you install the Code Review skill?
Download the ZIP of the code-review skill from this page and unzip it. You can install it for yourself or for a single project.
For a personal install, place the unzipped code-review folder in ~/.claude/skills/. For a project install, place it in the project's .claude/skills/ folder. In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal: ~/.claude/skills/code-review/
- Project: .claude/skills/code-review/ inside your repository
- Claude apps: Settings → Capabilities → Skills
~/.claude/skills/code-review/What does the skill need to run?
The skill needs an agent runtime that supports SKILL.md skills and provides the get_skill_reference and get_skill_script tools. The style check script, scripts/check_style.py, needs Python 3 to run.
The skill declares that it runs scripts and does not use the network. The package contains three files: SKILL.md, references/style-guide.md and scripts/check_style.py.
- Python 3 to run scripts/check_style.py
- An agent runtime with get_skill_reference and get_skill_script tools
- Files: SKILL.md, references/style-guide.md, scripts/check_style.py
How do you trigger the skill with a natural request?
You do not need a special command. Ask the agent to review code in plain language, and mention that you want a review, linting or a style check. The skill's description covers code review assistance with linting, style checking and best practices, so requests like these should match it.
Paste the code or point the agent at the file, then say what you want checked. Naming the focus, such as style or security, helps the agent prioritise.
- Review this Python module for style and bugs before I open a pull request.
- Do a code review of this function and group the feedback by severity.
- Check this file against the team style guide for naming problems.
What does a review look like?
The agent reports findings grouped under Critical, Important and Suggestion. It also works through a checklist covering naming conventions, hardcoded secrets or credentials, error handling, function length (under 50 lines), obvious security vulnerabilities and whether tests are included for new functionality.
The style script catches simple things such as long lines, trailing whitespace, camelCase names and single-letter variables. Treat its output as hints, because the agent's own reading of the code does most of the work.
- Checklist items: naming, hardcoded secrets, error handling, function length, security, tests
- Style script checks: long lines, trailing whitespace, camelCase names, single-letter variables
What are the safety and privacy considerations?
The skill does not use the network, but the code you submit for review is sent to the model provider as part of the conversation, because the agent reads it. Do not use the skill on code you cannot share with an AI service.
The bundled check is a basic heuristic and does not run a security scanner. Do not rely on it for security-critical code that needs a formal sign-off.
- Code under review is sent to the model provider.
- The skill runs a local Python script but does not use the network.
- No security scanner is included.
What are the limits and tips for better reviews?
The style check is approximate. The camelCase check, for example, is not exact, and it only covers Python. For thorough linting, run a dedicated linter such as ruff or flake8 alongside the skill.
Review small units of code at a time, and ask for the severity grouping explicitly so the output is easy to act on.
- Use ruff or flake8 for full linting; the bundled check is a simple heuristic.
- Keep reviews to one module or function at a time.
- Always have a human reviewer look at security-sensitive changes.
What people use it for
- Getting a first-pass review of a Python module before a teammate looks at it.
- Checking new code against a shared style guide for naming and imports.
- Learning good practice by reviewing your own functions against the checklist.
- Spotting hardcoded secrets or missing error handling before a pull request.
Frequently asked questions
How do I install the Code Review skill?
Download the ZIP from this page and unzip it. Put the code-review folder in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder. In the Claude apps, you can upload it under Settings → Capabilities → Skills.
What does the Code Review skill require?
It needs Python 3 to run scripts/check_style.py and an agent runtime that supports SKILL.md skills with get_skill_reference and get_skill_script tools. The skill declares that it does not use the network.
Is the Code Review skill free?
The skill is licensed under Apache-2.0. The source does not state pricing for the agent runtime or model provider you use with it.
Does it replace ruff or flake8?
No. The bundled style check is a simple heuristic that covers only Python, and its camelCase check is approximate. Use ruff or flake8 for full linting.
Is it safe to use on private code?
Code you submit is sent to the model provider during review. Avoid using it on code you cannot share with an AI service.
Can it find security vulnerabilities reliably?
The agent looks for security issues as part of its review, but the skill does not run a security scanner and the agent can miss vulnerabilities. Do not treat it as a sign-off for security-critical 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
Use this skill if you want a consistent, checklist-driven first pass on Python code, or a shared style reference for a team. Pick a full linter or a dedicated security scanner instead if you need accurate style enforcement, non-Python coverage, or formal security sign-off.
Download the skill