agent-reviewer_skill

Overview

This skill is a code review and quality assurance agent that inspects code for functionality, security, performance, maintainability, and documentation. It provides prioritized findings, concrete fixes, and action items to improve code health. The agent integrates with memory and MCP tooling to coordinate review state across agent swarms.

How this skill works

The agent runs automated checks (linting, tests, security scans, complexity metrics) then performs manual-style analysis for logic, vulnerabilities, and architecture. It produces a structured review: strengths, critical issues, suggestions, metrics, and action items. Findings can be stored and retrieved via memory hooks to coordinate with other agents in a workflow.

When to use it

• Before merging large or risky changes
• During release candidates and security audits
• When onboarding new codebases or major refactors
• To validate performance-sensitive features
• As part of continuous integration reviews

Best practices

• Run automated tools (lint, tests, security scans) before manual review
• Keep review scope small (<400 lines) to increase quality
• Prioritize issues by impact: security, functionality, performance, then style
• Provide concrete fixes and code snippets with every identified issue
• Store review findings in shared memory for swarm coordination

Example use cases

• Review a pull request for SQL injection, auth issues, and edge-case handling
• Audit a microservice for performance hotspots and N+1 query patterns
• Validate that new APIs have adequate docs, tests, and error handling
• Assess a refactor for SOLID compliance and testability impact
• Coordinate review findings to trigger follow-up tasks in a CI pipeline