ccg

Orchestrate parallel Codex and Gemini queries and synthesize outputs into a reconciled response.

Updated Mar 26, 2026
One-click install
npx skills add https://github.com/INNERJOINT/HarnessSkills --skill ccg-innerjoint
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ccg
Source: https://github.com/INNERJOINT/HarnessSkills/tree/main/skills/ccg
Command: npx skills add https://github.com/INNERJOINT/HarnessSkills --skill ccg-innerjoint

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Combines parallel external model perspectives (Codex and Gemini) and reconciles them via Claude to produce a single, consolidated recommendation, removing the need to manually run and compare multiple advisor outputs.

Core Features & Use Cases

  • Parallel advisor orchestration: Decomposes a request into Codex-focused and Gemini-focused prompts and runs both via the omc CLI.
  • Artifact-based synthesis: Collects advisor outputs from .omc/artifacts/ask and synthesizes agreed recommendations, conflicts, final direction, and an action checklist.
  • Fallback behavior: Continues with available providers when one or both external CLIs are unavailable and clearly notes missing perspectives.
  • Use Case: Run a fast two-perspective code review where Codex analyzes architecture and correctness while Gemini evaluates UX and documentation, then Claude produces the final unified verdict.

Quick Start

Run the CCG skill to synthesize Codex and Gemini perspectives for a task by executing the CCG invocation, for example: /oh-my-claudecode:ccg Review this PR - architecture/security via Codex and UX/readability via Gemini

Frequently Asked Questions about ccg

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I synthesize code review feedback from multiple AI models into one answer?

You can synthesize code review feedback from multiple models by orchestrating parallel queries to Codex and Gemini, then using Claude to reconcile their outputs into a single consolidated recommendation with agreed points, conflicts, and an action checklist.

Can I cross-validate architecture and UX analysis using Codex and Gemini simultaneously?

Yes, cross-validation of architecture and UX is possible by decomposing a request into Codex-focused architecture prompts and Gemini-focused UX prompts, running both in parallel via the omc CLI, and synthesizing the results via Claude.

What do I need to run multi-model orchestration for code review without manual team runtime setup?

To run multi-model orchestration without manual setup, you need the omc CLI and optional Codex and Gemini CLIs installed. The process reads advisor outputs from the .omc/artifacts/ask directory and uses Claude for final synthesis.

Does parallel model orchestration still work if the Codex or Gemini CLI is unavailable?

Parallel model orchestration includes graceful fallback behavior when one or both external CLIs are unavailable. It continues with available providers and clearly notes any missing perspectives in the final synthesized output.

What is the best way to compare architecture correctness and documentation readability across different models?

The best way to compare these perspectives is to run a two-perspective review where Codex analyzes architecture and correctness while Gemini evaluates readability, allowing Claude to produce a final unified verdict resolving any conflicts.