Multi-AI Code Review Skill
When the user wants a comprehensive code review using multiple AI perspectives, delegate to the review-coordinator subagent.
When to Use
Use this skill when the user:
- Asks for a "multi-AI" or "multi-perspective" review
- Wants to compare what different AI tools think about their code
- Asks for "consensus" or "cross-checking"
- Mentions Claude + Gemini + Codex reviewing together
- Wants a "comprehensive" or "full project" review
- Asks what different AIs think about their code
Default Behavior
- CLIs: All three (claude, gemini, codex)
- Focus: Comprehensive (all categories)
- Timeout: 10 minutes per CLI
What the Subagent Does
The review-coordinator subagent will:
-
Validate Prerequisites
- Check which CLIs are installed
- Report any missing CLIs with install instructions
- Continue with available CLIs
-
Execute Parallel Reviews
- Run all available CLIs simultaneously
- Monitor for completion or timeout
- Handle failures gracefully
-
Aggregate Results
- Parse findings from each CLI
- Match similar issues using fuzzy matching
- Categorize by agreement level
-
Present Comparison Report
- Consensus findings (all agree) - highest confidence
- Majority findings (2+ agree) - likely valid
- Unique findings (1 found) - needs human judgment
Example User Requests
- "Run a multi-AI review of this codebase"
- "What do Claude, Gemini, and Codex think about my code?"
- "Get a consensus review of my project"
- "Compare AI opinions on this code"
- "Do a comprehensive code review"
- "Cross-check my code with multiple AIs"
Use the review-coordinator subagent to handle the review workflow.