consulting-peer-llms

# Consulting Peer LLMs

## Overview

Get feedback from other LLM CLI tools (Gemini, Codex) on your current work. This skill executes multiple LLM reviews in parallel and synthesizes their responses into one actionable report.

**Core principle**: Use peer LLMs for external validation and diverse perspectives on implementation quality.

## When to Use

Use this skill when the user explicitly requests:
- "Validate this with codex"
- "Get feedback from gemini"
- "I want a review from other LLMs"
- "Do a peer review"

**Do NOT use this skill:**
- Automatically without user request
- For every piece of code (it's heavyweight)
- When quick internal review is sufficient

## Prerequisites

**Supported CLI tools:**
- `gemini` - Google Gemini CLI tool
- `codex` - OpenAI Codex CLI tool
- `claude` - Anthropic Claude CLI tool
- (Add more as needed)

**Detection and usage:**
- With arguments: Use only specified CLIs (e.g., `/review gemini codex`)
- Without arguments: Auto-detect all installed CLIs and use them

## Quick Start

**Minimal invocation:**
```
User: "Review this code with gemini and codex"

You:
1. Collect context (conversation + git changes)
2. Generate review prompt
3. Execute both CLIs in parallel
4. Show raw Gemini response
5. Show raw Codex response
6. Synthesize final assessment
```

## Workflow

### Step 0: Determine Target CLIs

# Arguments passed to /review command
REQUESTED_CLIS=("$@")

if [ ${#REQUESTED_CLIS[@]} -eq 0 ]; then
  # No arguments - default to standard set
  TARGET_CLIS=("gemini" "codex" "claude")
  echo "Using default CLIs: ${TARGET_CLIS[*]}"
else
  # Arguments provided - use specified CLIs
  TARGET_CLIS=("${REQUESTED_CLIS[@]}")
  echo "Using specified CLIs: ${TARGET_CLIS[*]}"
fi

**Key behaviors:**
- No arguments → Use all installed CLIs
- With arguments → Use only specified CLIs (skip if not installed)
- Always synthesize report (even for single CLI)

### Step 1: Collect Context

Gather comprehensive context from current session:

**Work Summa