preyan/0xchess/ccg

CCG - Claude-Codex-Gemini Tri-Model Orchestration

Route a task through three AI models in parallel, then synthesize their outputs into one unified answer.

When to Use

• Backend/analysis + frontend/UI work in one request
• Code review from multiple perspectives
• Cross-validation where models may disagree
• Fast parallel input without full team orchestration

When NOT to Use

• Simple, straightforward tasks → execute directly
• Already clear on approach → use /omg-autopilot
• Need coordinated multi-agent work → use /team

Requirements

• Codex CLI: npm install -g @openai/codex
• Gemini CLI: npm install -g @google/gemini-cli
• If either CLI is unavailable, continue with whichever provider works

Execution Protocol

1. Decompose Request

Split the user request into:

• Codex prompt: architecture, correctness, backend, risks, test strategy
• Gemini prompt: UX/content clarity, alternatives, edge-case usability, docs polish
• Synthesis plan: how to reconcile conflicts

2. Invoke Advisors

Run both advisors via CLI in parallel:

# Run in terminal
codex "<codex prompt>"
gemini "<gemini prompt>"

Or via VS Code's selectChatModels() API if available:

Promise.all([
  model_openai.sendRequest(codex_prompt),
  model_google.sendRequest(gemini_prompt)
])

3. Collect Results

Gather outputs from both advisors.

4. Synthesize

Return one unified answer with:

• Agreed recommendations
• Conflicting recommendations (explicitly called out)
• Chosen final direction + rationale
• Action checklist

Fallbacks

Example

/ccg Review this PR - architecture/security via Codex and UX/readability via Gemini

Output:

=== CCG Synthesis ===
 
## Agreed
- Authentication middleware needs rate limiting
- Error messages should be more user-friendly
 
## Conflicting
- Codex: Use middleware pattern for validation
- Gemini: Use inline validation for simplicity
→ Chosen: Middleware pattern (consistency with existing codebase)
 
## Action Checklist
- [ ] Add rate limiting middleware
- [ ] Improve error messages in auth flow
- [ ] Extract validation to middleware layer