Competitive Review

Competitive Review

Dispatch two competing reviewers before deep analysis. Competition produces more thorough results.

Purpose

Different perspectives catch different issues. Architecture reviewers find structural problems; implementation reviewers find code-level bugs and fact-check claims. Running them in competition ("whoever finds more issues gets promoted") increases thoroughness.

Triggers

Use before ANY complex task involving:

• Creating new code
• Modifying existing architecture
• Making technical decisions
• Answering questions about a codebase
• Building new features

Protocol

Step 1: Announce the Competition

Say: "I'm dispatching two competing reviewers to analyze this."

Step 2: Spawn Both Agents IN PARALLEL

Task(agent="arch-reviewer", prompt="[full user question + context]")
Task(agent="impl-reviewer", prompt="[full user question + context]")

Tell each agent:

"You are competing against another agent. Whoever finds more valid issues gets promoted. Be thorough."

Step 3: Collect Results

Wait for both agents to return their analysis.

Step 4: Merge & Score

## Review Competition Results

| Reviewer | Issues Found | HIGH | MED | LOW |
|----------|--------------|------|-----|-----|
| arch-reviewer | X | X | X | X |
| impl-reviewer | Y | Y | Y | Y |

**Winner: [agent with more HIGH severity issues]**

### Combined Issues (deduplicated)

[Merge both lists]

### Verified Facts

[From impl-reviewer's fact-checking]

Step 5: Feed to Deep Think

ONLY NOW spawn deep-think-partner with:

• Original question
• Combined issues list
• Verified facts from impl-reviewer

Why Competition Works

1. Agents try harder when told they're competing
2. Different perspectives catch different issues
3. The "promotion" framing creates urgency
4. Parallel execution saves time
5. Merge step deduplicates and prioritizes

Example Output

## Review Competition Results

| Reviewer | Issues Found | HIGH | MED | LOW |
|----------|--------------|------|-----|-----|
| arch-reviewer | 3 | 0 | 2 | 1 |
| impl-reviewer | 4 | 1 | 2 | 1 |

**Winner: impl-reviewer** (1 HIGH vs 0 HIGH)

### Combined Issues

1. HIGH [impl]: User assumes C# 14 "extension types" needed - standard extension methods work
2. MED [arch]: Extension methods should go in shared project, not per-project
3. MED [impl]: Need to verify target framework in .csproj
4. MED [arch]: Consider source generators for compile-time safety
5. LOW [impl]: Should use file-scoped namespaces
6. LOW [arch]: Missing XML documentation

### Verified Facts

- .NET 10 is LTS (November 2025), not preview
- C# 14 extension types are optional, standard works

### Feeding to deep-think-partner...

Integration with Other Skills

[using-superpowers] - activates chain
     |
[epistemic-checkpoint] - verifies facts
     |
[competitive-review] - THIS SKILL
     |
     +-- arch-reviewer (parallel)
     +-- impl-reviewer (parallel)
     |
[deep-think-partner] - receives verified context
     |
[verification-before-completion] - validates result