architecture

name	architecture
description	Run a quick architecture review via architect-reviewer agent

Architecture Skill

Invokes the architect-reviewer agent for design pattern and architecture review.

Information Needed

1. Target - What to review:
   • Proposed design or approach
   • Existing module or system
   • Integration pattern

Steps

1. Determine review scope:

   • Gather design documents or code context
   • Identify architectural concerns

2. Launch architect-reviewer agent via Task tool:

Task({
  subagent_type: "architect-reviewer",
  prompt: "Review the architecture of [target]. Evaluate:
    - Design patterns and SOLID principles
    - Scalability and performance implications
    - Technical debt and maintainability
    - Integration patterns
    - Trade-offs and alternatives

    Provide recommendations with rationale."
})

3. Review agent findings and present summary

Review Focus Areas

Area	What to Evaluate
Design Patterns	SOLID principles, appropriate abstractions
Scalability	Performance implications, bottlenecks
Technical Debt	Maintainability, complexity
Integration	Module boundaries, coupling
Trade-offs	Alternatives considered, decision rationale

Example: Review New Feature Design

User: "Review the architecture for real-time collaboration"

Launch agent:

Task({
  subagent_type: "architect-reviewer",
  prompt: "Review proposed architecture for real-time collaboration:

    Proposed approach:
    - Extend existing Loro CRDT sync
    - Add presence indicators via P2P
    - Cursor position sharing

    Evaluate:
    - Scalability with multiple peers
    - Integration with existing sync protocol
    - Performance implications
    - Alternative approaches (WebRTC, etc.)"
})

Example: Review Existing Module

User: "Review the sync module architecture"

Launch agent:

Task({
  subagent_type: "architect-reviewer",
  prompt: "Review the architecture of src/lib/sync/:

    Current structure:
    - P2P discovery via mDNS
    - TCP connections for sync
    - Loro CRDT for conflict resolution

    Evaluate:
    - Module organization and boundaries
    - Error handling patterns
    - Scalability concerns
    - Potential improvements"
})

Notes

• Always runs via the architect-reviewer agent (model: opus)
• Focuses on strategic evaluation, not implementation details
• Provides trade-off analysis and alternatives
• Advocates for evolutionary architecture