How do I provide a plan for the skill to review?

You can save a plan in the .claude/plans/ directory or simply discuss the plan in your current Claude conversation.

Can I use this for small, single-file changes?

Yes, but for plans under 50 lines, the skill will suggest using a simpler code-review agent to avoid unnecessary complexity.

What specialized roles are included in the review process?

The skill utilizes a code architect, cybersecurity expert, UX designer, performance engineer, and a dedicated bug finder.

How does the design-review skill differ from a standard code review?

While a code review looks at existing code, this skill evaluates a 'plan' or architectural proposal before code is written, using five specialized agents to catch systemic issues early.

How are the final review findings organized?

Findings are synthesized into a unified summary and categorized by severity: Critical, High, Medium, and Low.

Multi-Agent Design Reviewer

Name: Multi-Agent Design Reviewer
Author: sjawhar

bysjawhar

•

Security & Testing

Analyzes architectural plans through five specialized AI lenses to identify security, performance, UX, and structural risks.

The Design Review skill automates comprehensive technical audits by deploying five specialized agents—Architect, Security Expert, UX Designer, Performance Engineer, and Bug Finder—in parallel. It evaluates proposed software designs from multiple perspectives simultaneously, synthesizing findings into a prioritized summary of critical blockers and suggested enhancements. This is particularly useful for validating complex system changes before implementation, ensuring high-quality code and robust system architecture without the overhead of manual multi-departmental reviews.

Key Features

01Automatic plan detection from project directories or conversation history

02Synthesized reconciliation of conflicting expert recommendations

03Intelligent triage that suggests simpler reviews for trivial code changes

04Parallel multi-agent analysis for architectural, security, and performance insights

052 GitHub stars

06Prioritized issue categorization from critical blockers to low-priority enhancements

Use Cases

01Validating complex system architecture and module decomposition before coding

02Performing pre-implementation security audits to identify data exposure risks

03Evaluating API ergonomics and developer experience for new backend services

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add sjawhar/dotfiles design-review

For use in Claude.ai and ChatGPT