Multi-AI Workflow (MAW) is a command-line interface (CLI) framework designed to orchestrate multiple AI agents, including Claude, Codex, and Gemini, for collaborative task execution. It addresses the challenge of leveraging diverse AI strengths—Claude for reasoning and planning, Codex for code execution, and Gemini for multimodal analysis—within a unified workflow. This framework targets developers, researchers, and power users seeking to enhance productivity and tackle complex projects by harmonizing AI capabilities, offering benefits such as parallel processing and synchronized session management.

How It Works

MAW operates as a central orchestrator, routing tasks to specialized AI agents based on content analysis or predefined workflow modes. Its core architecture features a TypeScript CLI core and Python bridges for AI interaction. A key advantage is its unified session control, ensuring a single session ID synchronizes context across all participating AI agents. The framework supports parallel execution of tasks and offers seven distinct workflow modes, ranging from simple "Lite" execution to complex, multi-phase collaborations like "Five-Phase" and iterative "Ralph Loop" development, allowing users to select the approach best suited to their task's complexity and desired outcome.

Quick Start & Requirements

Installation requires Node.js (>= 18.0.0) and Python (>= 3.9). Optional but recommended are the Claude Code CLI, Codex CLI, and Gemini CLI. The recommended installation involves cloning the repository (git clone https://github.com/haoyu-haoyu/Multi-AI-Workflow.git), running the ./install.sh script, and then sourcing your shell configuration (source ~/.zshrc or ~/.bashrc). A manual installation guide is also provided for detailed setup.

Highlighted Details

• Seven Workflow Modes: Offers Lite, Lite-Plan, Plan, TDD-Plan, Brainstorm, Five-Phase, and Ralph Loop for diverse task management.
• Multi-AI Delegation: Seamlessly assigns tasks to Claude (planning), Codex (execution), and Gemini (analysis).
• Smart Routing: Automatically detects the most suitable AI agent for a given task.
• Report Generation: Creates professional reports with auto-generated diagrams (Mermaid).
• Real-time Dashboard: Provides visual workflow management with live updates via WebSocket.
• Unified Sessions: Maintains consistent context across all AI agents with a single session ID.

Maintenance & Community

The project is developed by haoyu-haoyu. While the README encourages contributions and provides a GitHub link, it does not detail specific community channels (like Discord or Slack) or list other notable contributors or sponsorships.

Licensing & Compatibility

MAW is released under the MIT License, which permits broad usage, including commercial applications and integration with closed-source projects, with minimal restrictions beyond attribution.

Limitations & Caveats

The provided README does not explicitly detail known limitations, bugs, or deprecation plans. Full functionality for Codex and Gemini may depend on the separate installation and configuration of their respective CLIs and potentially API keys for Gemini. The framework includes configurable sandbox levels (read-only, workspace-write, full-access) for security, implying potential restrictions on AI actions depending on the chosen level.