reviewing

# Reviewing Skill

Multi-model second opinions via OpenAI Codex CLI and Google Gemini CLI.

## When to Use

- Validating critical findings (security, correctness, soundness)
- Breaking ties when uncertain
- Cross-checking complex reasoning
- Architecture decisions with high stakes

**Don't use for**: Simple questions. Just think harder.

## Priority Order

| Priority | Tool | When |
|----------|------|------|
| **1st** | Codex | Primary second opinion (different architecture from Claude) |
| **2nd** | Gemini | Tie-breaker, or when Codex unavailable |
| **3rd** | `claude -p` | Fallback (fresh context, same architecture) |

## Codex (OpenAI)

OpenAI's CLI for code and reasoning tasks.

### Invocation

```bash
codex exec -s read-only -m gpt-5.2 -c reasoning=xhigh "<prompt>"
```

| Flag | Purpose |
|------|---------|
| `-s read-only` | **Required** - sandbox mode, no modifications |
| `-m gpt-5.2` | Best model for thorough review |
| `-c reasoning=xhigh` | Maximum reasoning effort |

### Models

| Model | Use Case |
|-------|----------|
| `gpt-5.2` | **Default** - thorough review, high-stakes |
| `o3` | Complex multi-step reasoning |
| `o4-mini` | Quick opinions, low-stakes |

### Reasoning Levels

**Choose the right level for the task.** Higher reasoning = more time.

| Level | Flag | Time | When to Use |
|-------|------|------|-------------|
| **xhigh** | `-c reasoning=xhigh` | 2-5 min | Security, correctness proofs, high-stakes architecture |
| **high** | `-c reasoning=high` | 30-90 sec | Standard code review, design tradeoffs, edge cases |
| **medium** | `-c reasoning=medium` | 10-30 sec | Straightforward questions, sanity checks |
| **(default)** | *(omit flag)* | 5-15 sec | Quick opinions, simple validations |

**Guidance:**
- Start with `high` for most review tasks—it's the sweet spot
- Escalate to `xhigh` only when correctness is critical or reasoning is complex
- Use `medium` or default for simple questions where speed matters
- If `xhigh` times out or you're w