🗣️ DebatingAgents v0.2.0

An Intelligent Multi-Model AI Debate Arena & Consensus Engine

---

📖 Overview

DebatingAgents is a state-of-the-art multi-agent AI debate platform designed to tackle complex math proofs, logic puzzles, coding challenges, and open-ended analytical queries. By orchestrating parallel adversarial debates between competing top-tier AI models (OpenAI, Gemini, Anthropic Claude, Groq/DeepSeek, and local Ollama models), agents engage in line-by-line cross-examination to expose calculation mistakes, logical fallacies, and unstated assumptions.

When deadlocked without consensus, an elected AI Chief Justice / Moderator reviews the full transcript, evaluates argument strengths, and issues a final reasoned ruling.

---

🎯 Key Features

🤖 Multi-Model Arena Debate across OpenAI, Google Gemini, Anthropic Claude, Groq / DeepSeek, and Ollama (Local) simultaneously.	⚔️ Adversarial Cross-Examination Models are forced to audit peer reasoning line-by-line, pointing out logical fallacies and calculation errors in opposing arguments.
⚖️ AI Chief Justice Synthesis An elected Moderator AI analyzes deadlocked debates, pinpoints flaws in losing logic, and delivers a definitive winning verdict.	⚡ Parallel Async Performance Engineered with Python asyncio and httpx for concurrent, high-speed model responses and low latency.
🎨 Rich Visual Terminal UX Color-coded model badges, real-time live execution panels, confidence meters (0-100%), and callout banners powered by rich.	📊 Transcripts & Telemetry Automatically exports full debate logs as structured Markdown (.md) reports and raw JSON (.json) data into ./debates/.

---

🛠️ Supported AI Providers & Models

Provider	Supported Models	Required Environment Key
OpenAI	gpt-4o, gpt-4o-mini, o3-mini	OPENAI_API_KEY
Google Gemini	gemini-2.5-flash, gemini-2.5-pro, gemini-1.5-flash	GEMINI_API_KEY
Anthropic Claude	claude-3-5-sonnet, claude-3-5-haiku	ANTHROPIC_API_KEY
Groq	llama-3.3-70b, deepseek-r1-groq	GROQ_API_KEY
Ollama (Local)	ollama-llama3, ollama-deepseek	OLLAMA_BASE_URL

---

⚡ Quickstart

1. Clone the Repository

git clone https://github.com/smaranjitghose/DebatingAgents.git
cd DebatingAgents

2. Configure API Keys

Copy .env.example to .env and add your active API keys:

cp .env.example .env

Example .env configuration:

OPENAI_API_KEY=sk-proj-...
GEMINI_API_KEY=AIzaSy...
ANTHROPIC_API_KEY=sk-ant-...
GROQ_API_KEY=gsk_...

3. Install Dependencies & Launch

Using uv (recommended ultra-fast Python manager):

uv sync
uv run main.py

---

💻 Usage & CLI Examples

🔹 Interactive Mode

Run without arguments to launch the interactive prompt:

uv run main.py

🔹 Command-Line Problem Input

uv run main.py "How many prime numbers are less than 100?"

🔹 Custom Model Lineup & Round Limits

Specify exact debate participants and round caps:

uv run main.py "What is the sum of angles in an 8-sided regular polygon?" \
  --models gpt-4o-mini,gemini-1.5-flash \
  --rounds 3

🔹 High-Stakes Multi-Model Arena with AI Judge

Debate across OpenAI, Gemini, and Groq/DeepSeek with an explicit Judge model:

uv run main.py "In a room of 23 people, what is the exact probability that at least two share a birthday?" \
  --models gpt-4o-mini,gemini-1.5-flash,llama-3.3-70b \
  --judge gpt-4o-mini \
  --rounds 5 \
  --export-dir ./my_debates

---

🏗️ Architecture & Project Structure

graph TD
    A[User Question] --> B[DebateEngine Core]
    B -->|Async Parallel Query| C1[OpenAI Provider]
    B -->|Async Parallel Query| C2[Gemini Provider]
    B -->|Async Parallel Query| C3[Claude / Groq / Ollama]
    C1 --> D{Consensus Check}
    C2 --> D
    C3 --> D
    D -->|Agreed| E[Consensus Answer Output]
    D -->|Disagreed| F[Adversarial Cross-Examination Round]
    F -->|Deadlock after Max Rounds| G[AI Chief Justice Verdict]
    E --> H[Rich Terminal UI & Report Exporter]
    G --> H

Loading

DebatingAgents/
├── debating_agents/
│   ├── __init__.py           # Package version & metadata
│   ├── config.py             # Model registry & auto-key detection
│   ├── models.py             # Pydantic schemas (AgentResponse, JudgeVerdict, DebateTranscript)
│   ├── debate_engine.py      # Async multi-round debate runner & AI Judge synthesizer
│   ├── ui.py                 # Rich terminal panels, tables, and color badges
│   ├── cli.py                # Command-line interface & argument parser
│   └── providers/            # Async model adapters
│       ├── base.py           # Provider interface & JSON payload extraction
│       ├── openai_provider.py
│       ├── gemini_provider.py
│       ├── anthropic_provider.py
│       ├── groq_provider.py
│       └── ollama_provider.py
├── debates/                  # Exported Markdown & JSON report directory
├── main.py                   # Main CLI entrypoint
├── pyproject.toml            # Dependencies & Hatchling build target
├── .env.example              # API Key configuration template
└── README.md                 # Documentation

---

🤝 Connect & Contribute

Contributions, issues, and feature requests are welcome! Feel free to check out the issues page.

---

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.