PolyMind CLI

Multi-Model AI Council & Agentic Workflows in Your Terminal

PolyMind CLI is a next-generation terminal interface designed for developers who want to leverage the power of LLM Councils. Instead of relying on a single model, PolyMind orchestrates a debate between multiple AI personas (powered by Gemini 3.0, GPT-4o, Claude 3.5) to reach higher-quality conclusions through dialectic reasoning.

⚡ The Origin Story: 99% Vibecoded

The idea for PolyMind didn't come from a whiteboard session or a product roadmap. It surfaced in a flash of inspiration—a desire to see what would happen if you put the world's smartest AIs in a room (or a server) and forced them to agree.

This entire project was 99% vibecoded. It wasn't about rigid engineering specs; it was about capturing a feeling. The feeling of a futuristic council chamber, the matrix-like flow of data, and the raw power of dialectic reasoning in your terminal. We built this to look cool, feel premium, and actually work.

Built with Node.js, TypeScript, and Ink, it delivers a premium, interactive TUI (Terminal User Interface) experience.

---

🚀 Features

• 🤖 Multi-Model Council: Orchestrate debates between different AI personas to reduce hallucination and improve reasoning.
• ⚡ Real-time Streaming: Watch the debate unfold character-by-character with a matrix-style TUI.
• 🧠 Advanced Reasoning: Implements a 3-stage council workflow (Proposal -> Rebuttal -> Consensus).
• 🔌 Multi-Provider Support: First-class support for Gemini 3.0, GPT-4o, and Claude 3.5.
• 🎨 Interactive TUI: Beautiful, responsive terminal interface with animations, progress bars, and syntax highlighting.
• 🛠️ Extensible Config: Easy-to-use configuration for API keys, personas, and debate parameters.

---

📦 Installation

Ensure you have Node.js 18+ installed.

# Install globally via npm
npm install -g polymind-cli

# Or run directly with npx
npx polymind-cli debate "Is Rust better than C++?"

---

🛠️ Configuration

Before running your first council, initialize the configuration with your API keys.

polymind init

You can also configure providers manually:

# Set Gemini as the default provider
polymind config set provider gemini

# Add your API key
polymind config set apiKey AIzaSy...

Supported Environment Variables

Create a .env file or export these variables:

Variable	Description
GEMINI_API_KEY	Key for Google Gemini models
OPENAI_API_KEY	Key for OpenAI GPT models
CLAUDE_API_KEY	Key for Anthropic Claude models
POLYMIND_PROVIDER	Default provider (gemini, gpt, claude)

---

🧠 The LLM Council Engine

PolyMind uses a sophisticated Council Engine to elevate AI reasoning. Unlike standard chat, a Council Session involves multiple stages of deliberation.

Stage 1: First Opinions (Proposal)

Independent personas (e.g., "The Skeptic", "The Visionary", "The Engineer") analyze the user's query in parallel. They generate initial stances without being influenced by each other.

Stage 2: Peer Review (Rebuttal)

The engine shuffles the proposals and has each persona critique another's viewpoint. This "adversarial" phase exposes flaws, biases, and weak arguments.

Stage 3: Chairman Synthesis (Consensus)

A designated "Chairman" model (typically the most capable model, e.g., Gemini 3.0 Pro) synthesizes the proposals and rebuttals into a final, comprehensive answer.

---

💻 Usage

1. Start a Debate

The core feature of PolyMind. Ask a complex question and watch the council deliberate.

polymind debate "Should we rewrite our legacy monolith in Microservices?" \
  --rounds 3 \
  --personas "Architect" "ProductManager" "DevOps"

2. Interactive Chat

Chat one-on-one with a specific persona in a persistent session.

polymind chat --persona "Senior Engineer"

3. Interactive Mode (Slash Commands)

Enter a shell-like environment to run multiple commands.

polymind interactive
# Then type: /debate "AI Safety"

---

🖥️ TUI Overview

PolyMind features a rich Terminal User Interface built with React and Ink.

• Intuitive Commands: Access the LLM Council, live chat, debates, and configuration tools
• Rich Terminal Display: Beautiful gradient ASCII art and clear command documentation
• Multiple Modes: Council deliberation, live streaming, debates, and interactive sessions
• Status Indicators: Monitor system configuration and model availability in real-time

---

🤖 Supported Models

PolyMind is optimized for the latest generation of reasoning models.

Model	Type	Best For
Gemini 3.0 Pro	Multimodal	Chairman Role, Complex Synthesis
Gemini 3.0 Flash	Fast Inference	Initial Proposals, Quick Rebuttals
GPT-4o	Reasoning	Logic Checks, Code Verification
Claude 3.5 Sonnet	Creative	Nuanced Arguments, Ethical Review

---

🏗️ Architecture

PolyMind is built on a modular architecture designed for extensibility.

• CLI Entrypoint: Handles argument parsing and command dispatch (commander).
• Command Registry: Maps commands (debate, chat) to their handlers.
• Council Engine: (Simulation Mode in v1.0) Orchestrates the multi-stage debate flow.
• TUI Layer: Renders the state to the terminal using ink and react.
• Provider Abstraction: Unified interface for calling different LLM APIs.

---

🗺️ Roadmap

• v1.0.0: Initial Release with TUI and Simulated Council.
• v1.1.0: Real-time API integration for all providers.
• v1.2.0: Custom Persona definitions via YAML.
• v2.0.0: Local LLM support via Ollama integration.

---

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md for details.

1. Fork the repo
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

---

📄 License

Distributed under the MIT License. See LICENSE for more information.