# CCG (Claude-Codex-Gemini) Skill ## Overview CCG sends your task to Claude, Codex (GPT-5), and Gemini simultaneously, then uses Claude to synthesise the best response from all three. Different models have different strengths and blind spots — CCG captures the consensus. ## Model Roles | Model | Strength | |-------|---------| | Claude | Reasoning, nuance, safety | | Codex (GPT-5) | Code generation, pattern matching | | Gemini | Long-context, research synthesis | ## How to Invoke ``` Skill: ccg Task: <your question or task> ``` ## Synthesis Process 1. Task dispatched to all three models in parallel 2. Claude receives all three responses 3. Claude identifies agreements, disagreements, and unique insights from each 4. Synthesised answer highlights cross-model consensus and flags divergences 5. Final recommendation made with confidence level ## Best Use Cases - "Should I use approach A or B?" — get three independent votes - Security review — different models catch different vulnerability classes - Algorithm selection — compare performance characteristics across models - API design — validate against multiple perspectives