🗿 MoAI

<h1 align="center">
  🗿 MoAI
</h1>

<p align="center">
  <strong>Master of AIs</strong><br>
  <em>Multi-model AI roundtable for collaborative brainstorming</em>
</p>

<p align="center">
  <img src="https://img.shields.io/badge/python-3.11+-blue?style=flat-square&logo=python&logoColor=white" alt="Python 3.11+">
  <img src="https://img.shields.io/badge/status-Phase%201-orange?style=flat-square" alt="Status: Phase 1">
  <img src="https://img.shields.io/badge/license-MIT-green?style=flat-square" alt="License: MIT">
</p>

---

## 💡 The Idea

Instead of asking **one** AI a question, ask a **team** of AIs.

```
You: "How should we architect our new microservice?"

┌─────────┐  ┌─────────┐  ┌─────────┐
│ Claude  │  │   GPT   │  │ Gemini  │
└────┬────┘  └────┬────┘  └────┬────┘
     │            │            │
     ▼            ▼            ▼
   "Event       "Start      "Consider
   sourcing     simple,     the team's
   gives you    monolith    experience
   audit        first..."   level..."
   trails..."
     │            │            │
     └────────────┼────────────┘
                  ▼
         🤝 Discussion Rounds
                  ▼
         📋 Consensus Summary
```

Models see each other's responses, debate, agree, disagree, and build upon ideas. You get richer, more diverse insights—then export everything as clean markdown.

---

## ✨ Features

| Feature | Description |
|---------|-------------|
| 🔀 **Open Mode** | Ask all models in parallel—get diverse initial takes |
| 💬 **Discuss Mode** | Sequential rounds where models respond to each other |
| 🎯 **@Mentions** | Direct questions to specific models: `@claude explain more` |
| 🤝 **Consensus** | Auto-generated summary of agreements & disagreements |
| 📁 **Projects** | Organize discussions by topic with persistent context |
| 📤 **Export** | Beautiful markdown documents of entire discussions |

---

## 🚀 Quick Start

```bash
# Clone & setup
git clone https://git.example.com/mikkel/moai.git
cd moai
uv sync  # or: pip install -e ".[dev]"

# Configure
cp .env.example .env
# Edit .env with your tokens

# Run
python -m moai.bot.main
```

---

## 🤖 Telegram Commands

### Projects
```
/projects              List all projects
/project new "Name"    Create project
/project select <id>   Switch project
/project models claude,gpt,gemini
```

### Discussion
```
/open <question>       Ask all models (parallel)
/discuss [rounds]      Start discussion (default: 3 rounds)
@claude <message>      Direct message to Claude
@gpt <message>         Direct message to GPT
/stop                  End current discussion
```

### Output
```
/consensus             Generate agreement summary
/export                Export as markdown
```

---

## 🏗️ Architecture

```
┌──────────────────┐     ┌──────────────────────────────────┐
│  Telegram Bot    │────▶│         Python Backend           │
│  (thin client)   │     │                                  │
└──────────────────┘     │  ┌────────────┐ ┌─────────────┐  │
                         │  │ Handlers   │ │ Orchestrator│  │
                         │  └─────┬──────┘ └──────┬──────┘  │
                         │        │               │         │
                         │        ▼               ▼         │
                         │  ┌──────────┐   ┌───────────┐   │
                         │  │ SQLite   │   │ AI Client │   │
                         │  └──────────┘   └─────┬─────┘   │
                         └───────────────────────┼─────────┘
                                                 │
                         ┌───────────────────────┼─────────┐
                         │     Requesty / OpenRouter       │
                         │  ┌─────────┬─────────┬────────┐ │
                         │  │ Claude  │   GPT   │ Gemini │ │
                         │  └─────────┴─────────┴────────┘ │
                         └─────────────────────────────────┘
```

---

## 📍 Roadmap

- [x] **M1** Bot basics (`/help`, `/status`)
- [ ] **M2** Project CRUD
- [ ] **M3** Single model Q&A
- [ ] **M4** Open mode (parallel)
- [ ] **M5** Discuss mode (sequential)
- [ ] **M6** Consensus generation
- [ ] **M7** Markdown export
- [ ] **M8** @mention support
- [ ] **Phase 2** Web UI

---

## 🛠️ Tech Stack

| Layer | Technology |
|-------|------------|
| Bot | `python-telegram-bot` (async) |
| Backend | Python 3.11+, FastAPI |
| Database | SQLite → PostgreSQL |
| AI Routing | Requesty / OpenRouter |
| Linting | Ruff |
| Testing | pytest |

---

## 📄 License

MIT © 2026

---

<p align="center">
  <sub>Why ask one AI when you can ask them all? 🗿</sub>
</p>