Discover and explore top open-source AI tools and projects—updated daily.
fengshao1227AI-powered multi-model development workflow system
New!
Top 41.7% on SourcePulse
CCG (Claude Code + Codex + Gemini) is a multi-model collaborative development toolset designed to enhance coding workflows. It targets developers using Claude Code CLI, enabling them to leverage specialized AI models like Codex for backend tasks and Gemini for frontend tasks, orchestrated by Claude Code. This system aims to improve efficiency and code quality through intelligent task routing, parallel analysis, and maintaining code sovereignty.
How It Works
CCG integrates Claude Code CLI as the orchestrator, delegating specific code generation and analysis tasks to specialized models. Gemini CLI is utilized for frontend development (UI, CSS, components), while Codex CLI handles backend logic, algorithms, and debugging. The system features intelligent routing to automatically select the appropriate model for each task. It employs multi-model parallelism, allowing Codex and Gemini to analyze concurrently for cross-validation, reducing errors. A key architectural choice is the "zero-write" permission model, where external models return only patches, ensuring Claude Code maintains ultimate control over the codebase. Token usage is optimized through dynamic injection of expert prompts (ROLE_FILE) without consuming main session tokens.
Quick Start & Requirements
Installation is performed via npx ccg-workflow, which automatically installs all 15 commands.
npx ccg-workflow@latest.Highlighted Details
/ccg:workflow) covering research, ideation, planning, execution, optimization, and review./ccg:commit) and interactive rollback (/ccg:rollback).Maintenance & Community
The project is developed by the "CCG Community". For support and issue reporting, GitHub Issues can be used. Community discussions are hosted on linux.do. Updates are managed through npm, with the latest version available via the npx ccg-workflow@latest command.
Licensing & Compatibility
The project is released under the MIT License. This license permits commercial use, modification, distribution, and patent use, making it compatible with most development environments and closed-source projects.
Limitations & Caveats
Without the optional Codex CLI or Gemini CLI installed, the system degrades to a pure Claude workflow, losing benefits like smart routing, multi-model parallelism, and the expert role system. The functionality related to code retrieval and prompt enhancement via the MCP tool requires separate configuration using either official services or community relays. Post-installation, users may need to refresh their shell environment (e.g., source ~/.zshrc) for commands to be recognized.
1 day ago
Inactive