afrexai-code-reviewer

What This Skill Does

The afrexai-code-reviewer is an enterprise-grade automated code review agent designed to act as an immediate, high-fidelity security and quality engineer within your OpenClaw environment. Unlike static analysis tools that require complex CI/CD pipeline integration, this skill leverages advanced agentic intelligence to analyze code contextual patterns. It evaluates projects against the proprietary SPEAR framework, ensuring that code is not just functional, but secure, performant, and architecturally sound.

Installation

To integrate this agent into your workflow, run the following command in your terminal: clawhub install openclaw/skills/skills/1kalin/afrexai-code-reviewer Once installed, the agent is ready to process local files, directories, or remote repository data immediately.

Use Cases

1. Pre-PR Security Audit: Run the reviewer on a local branch before opening a pull request to catch critical vulnerabilities like hardcoded secrets or SQL injection vectors before they ever touch your main branch.
2. Legacy Code Refactoring: Identify architecture smells and performance bottlenecks in existing modules to facilitate cleaner, faster code migrations.
3. Educational Code Reviews: Use the agent to provide feedback on junior developer contributions, helping to enforce team coding standards and best practices through consistent, objective scoring.
4. Performance Tuning: Detect N+1 query patterns and inefficient memory usage in your application layer without needing to run expensive load tests.

Example Prompts

• "Review the changes in this PR and prioritize any critical security risks found using the SPEAR framework."
• "Analyze src/controllers/user-controller.js for performance bottlenecks and provide a list of suggested improvements."
• "Perform a comprehensive review of these staged changes, focusing specifically on error handling gaps and potential memory leaks."

Tips & Limitations

• Weighting Logic: Remember that the agent prioritizes security (3x) and performance (2x) in its scoring. A perfect score is rarely possible in complex systems; look for the delta in scores over time.
• Scope: The agent works best on localized changesets. For massive monorepos, target specific files or PRs rather than the entire root directory to maintain granular accuracy.
• Context: The agent does not execute your code; it performs deep semantic analysis. Ensure you have the necessary read permissions for any repository you ask the agent to scan.