bmad-review-adversarial-general

Adversarial Review (General)

🔍 bmad-review-adversarial-general | Cynical Review

A review technique where the reviewer MUST find issues. No "looks good" allowed. Adopts a cynical stance — assume problems exist and find them.

How It Works

1. Read content with a cynical, critical perspective
2. Identify issues across completeness, correctness, and quality
3. Search for what's missing — not just what's present and wrong
4. Minimum threshold — Must find 10+ issues or re-analyze deeper

Usage

bmad-review-adversarial-general [content]

Example:

bmad-review-adversarial-general src/store/tradingStore.ts

Review Dimensions

• Completeness — What's missing? Edge cases? Error handling?
• Correctness — Does it do what it claims? Are there bugs?
• Quality — Is the code clean? Are there code smells?
• Security — Are there vulnerabilities?
• Performance — Are there inefficiencies?
• Maintainability — Will future developers understand this?

Output Format

## Adversarial Review: [Artifact]

### Findings

1. **[SEVERITY]** — [Location] — [Description]
2. **[SEVERITY]** — [Location] — [Description]
...

### Summary
- Critical: N
- High: N
- Medium: N
- Low: N

### Recommendations
[Prioritized action items]

Human Filtering Required

The AI is instructed to find problems, so it will find problems even when they don't exist. You decide what's real. Review each finding, dismiss false positives, fix what matters.