Code Review Agent Skill
Perform systematic code reviews covering best practices, security, performance, and maintainability.
The skill itself does not execute code or external operations, only provides review guidance, so it does not involve sensitive data handling or permission abuse. The repository has SECURITY.md describing security practices, but no specific authorization or data-flow descriptions for this skill; publisher identity unverified, and no rollback or recovery mechanisms stated. Therefore trust score is moderate-low.
The skill is purely a guide, with no executable scripts or tests, making it impossible to verify key-path reproducibility from static review. Abnormal input handling and failure feedback are undefined, so reliability is low.
The skill's goal is clear (code review), but it does not declare capability boundaries or non-fit scenarios; trigger conditions rely only on the description. Also no Chinese support or mention of overseas service dependencies, limiting fit for Chinese users.
The skill has clear title and description, but lacks installation notes, known limitations, versioning, changelog, and clear maintenance responsibility. Examples are simple, no FAQ or troubleshooting, so convention is moderate.
The skill provides a systematic checklist and format, but no actual outputs or validation results confirm effectiveness. Some content like examples could be misleading, marginal value limited.
No tests, CI, or third-party execution evidence; relies solely on prescriptive text, so verifiability is low.
- The skill is a static guide without actual tests; verify applicability before use.
- Publisher identity unverified; assess maintenance activity independently.
- Repository dependencies may require proxy access from mainland China, but the skill itself does not require network.
What does this skill do, and when should you use it?
This skill provides a structured framework for conducting comprehensive code reviews. It includes a checklist covering code quality, security analysis (including OWASP Top 10), performance considerations, best practices (DRY, SOLID), and maintainability. The output is formatted with Summary, Critical Issues, Major Concerns, Suggestions, and Positive Feedback. It is a pure prompt-based skill with no code or dependencies.
The skill guides the user through: 1) evaluating code readability, naming, structure, comments, and complexity; 2) checking for SQL injection, XSS, authentication flaws, insecure data handling, and input validation; 3) identifying bottlenecks, inefficient algorithms, unnecessary queries, and caching opportunities; 4) enforcing DRY, SOLID, error handling, testing, and dependency review; 5) assessing maintainability, abstractions, scalability, and technical debt; 6) producing a structured review with severity-based prioritization and example outputs.
- A developer performing a pre-merge self-review of a pull request, wanting to ensure all common issues are caught.
- A tech lead conducting a formal code review and needing a consistent format for reporting findings to the team.
- A security engineer auditing a codebase for OWASP Top 10 vulnerabilities, using the checklist as a guide.
- A new hire learning the team's review standards and how to prioritize issues by severity.
What are this skill's strengths and limitations?
- Structured and comprehensive coverage across five key dimensions.
- Includes actionable vulnerability lists and examples.
- Clear output format that is easy to read and act upon.
- Language-agnostic and simple to adopt.
- Prompt-only; no automated analysis — requires human/LLM judgment.
- No built-in integration with CI/CD or IDE.
- Lacks a test suite or sample code.
- Effectiveness depends on the reviewer's knowledge.
How do you install this skill?
The skill is located in the swarms repo at examples/single_agent/capabilities/skills/code-review/SKILL.md. Clone the repository and navigate to that directory, or copy the entire skills directory to your Agent Skills folder.
How do you use this skill?
When asked to review code, follow the checklist in SKILL.md: check code quality, security, performance, best practices, and maintainability. Use the prescribed output format: Summary, Critical Issues, Major Concerns, Suggestions, and Positive Feedback, and provide specific examples with severity ratings.
How does this skill compare with similar options?
This skill provides a prompt-based checklist without any external dependencies, similar to GitHub Copilot's code review feature but with a more structured output format.