domain-reviewer

name	domain-reviewer
description	Coordinate deep-dive reviews via specialized sub-agents as critics. Use for deep review of a specific module or domain (not general code review), when multiple specialized perspectives are needed (e.g. security plus performance plus correctness), or for pre-release review of critical system components.
license	MIT

Domain & Module Reviewer

Coordinate deep-dive reviews of specific system domains or modules by engaging specialized sub-agents as critics.

When to Use

Reach here when one module or domain needs several specialist lenses at depth (security + performance + correctness), coordinated via sub-agent critics. For a general triple-lens pass over an ordinary change before merge, the lighter sibling is audit-reviewer — don't spin up domain-reviewer's sub-agents for a routine diff.

Workflow

1. Scope: Define boundaries of the module/domain to review (e.g., "The Payment Gateway Service", "The Authentication flow").

2. Select Critics: Identify sub-agents for the review based on tech stack and concerns (e.g., Go code quality, security, SQL performance).

3. Formulate Prompts: Draft clear, scoped prompts for each sub-agent, asking them to evaluate specific aspects against architectural contracts.

4. Delegate: Execute sub-agent calls in parallel if possible.

5. Synthesize: Collect reports. Do not copy-paste output. Extract critical architectural flaws and document them in an ADR or action plan.

Completion

• All sub-agent reviews collected and synthesized
• Critical findings documented in ADR or action plan
• No copy-pasted sub-agent output — findings are curated

Anti-Patterns

• Dumping raw sub-agent output without synthesis — the value is in curation
• Overly broad scope — review one domain at a time, not the entire system
• Skipping parallel execution — sub-agents are independent, run them concurrently