/architecture-review

Architecture Review Command

Principal-level architecture assessment against ADRs and design patterns.

Usage

/architecture-review

What It Does

1. **Context**: Establish review scope 2. **ADR Audit**: Check architecture decisions 3. **Interaction Mapping**: Diagram dependencies 4. **Principle Checks**: Validate design patterns 5. **Risk Assessment**: Document consequences

Scope

• Module boundaries
• Dependency graphs
• Data flow patterns
• Coupling analysis
• ADR compliance

War Room Checkpoint (Automatic)

**Purpose**: Assess whether architecture findings warrant expert deliberation before recommendation.

**Auto-triggers when** (moderate approach):

• ADR violations without clear remediation path, OR
• High coupling score (>0.7), OR
• Multiple module boundary violations (>2), OR
• Conflicting design principles detected

**Checkpoint invocation** (automatic, after Risk Assessment):

Skill(attune:war-room-checkpoint) with context:
  source_command: "architecture-review"
  decision_needed: "Architecture recommendation for [component/PR]"
  blocking_items: [
    {type: "adr", description: "Violates ADR-003 without justification"},
    {type: "coupling", description: "High coupling between auth and payment"}
  ]
  files_affected: [modules under review]
  profile: [from user settings, default: "default"]

**War Room Questions** (when escalated):

• "Should we recommend blocking or propose remediation?"
• "Does this coupling warrant refactoring or is it acceptable?"
• "Should a new ADR be created to document this deviation?"

**Response handling**:

| RS Score | Mode | Action | |----------|------|--------| | RS <= 0.40 | Express | Quick recommendation, continue | | RS 0.41-0.60 | Lightweight | Panel reviews trade-offs | | RS > 0.60 | Full Council | Full deliberation on architecture decision |

**Skip conditions**:

• All ADRs compliant
• Coupling within acceptable thresholds
• No boundary violations

Output

• Architecture assessment
• ADR compliance status
• Coupling violations
• Principle adherence
• Recommendations (Approve/Block)
• War Room session ID (if escalated)