🧠 AutoSquad - Autonomous Multi-Agent Development Framework

Turn simple prompts into working software with AI development teams

AutoSquad is a cutting-edge framework that creates autonomous AI development teams using Microsoft's AutoGen. Give it a project prompt, and watch specialized AI agents collaborate to build complete, working applications.

🚀 New in v0.2: Token Optimization & Live Progress

✨ 69% Token Reduction - Intelligent context management dramatically reduces OpenAI API costs
✨ Live Progress Display - Real-time agent activity dashboard with beautiful terminal UI
✨ Cost Transparency - Monitor token usage and costs as development happens
✨ Enhanced User Experience - Watch your agents collaborate in real-time

🎯 What AutoSquad Does

AutoSquad assembles specialized AI agent teams that work together to turn your ideas into reality:

• 🎯 PM Agent: Analyzes requirements, breaks down features, manages scope
• 🧑‍💻 Engineer Agent: Writes production-ready code, implements features, fixes bugs
• 🏗️ Architect Agent: Reviews code quality, suggests improvements, ensures scalability
• 🧪 QA Agent: Tests functionality, finds edge cases, validates user experience

Input: A simple text description of what you want to build
Output: Complete, working software with documentation and tests

⚡ Quick Start

1. Install AutoSquad

git clone https://github.com/your-org/autosquad.git
cd autosquad
pip install -r requirements.txt

2. Set Your OpenAI API Key

export OPENAI_API_KEY="your-api-key-here"

3. Create Your First Project

mkdir my-project
echo "Build a Python CLI tool that converts JSON to CSV with error handling" > my-project/prompt.txt

4. Launch Your AI Development Squad

# With live progress display (default)
autosquad run --project my-project --rounds 3

# For servers/automation (simple progress bars)
autosquad run --project my-project --rounds 3 --no-live-display

5. Watch Your Agents Work

Experience the new live progress display:

╭─ 🧠 AutoSquad - my-project ──────────────────────────────────────╮
│ Round 2/3 | Elapsed: 4m 23s | Files: 7 | Tokens: 12,847 (~$0.38) │
╰──────────────────────────────────────────────────────────────────╯

╭─ 🤖 Agent Status ──────────────────────────╮
│ Agent      │ Status        │ Current Action          │ Progress │
│ Engineer   │ 🟢 Active     │ Writing main.py         │ 3 tasks  │
│ Architect  │ 🟡 Recent     │ Reviewing structure     │ 2 tasks  │  
│ PM         │ ⚪ Waiting    │ Planning next features  │ 4 tasks  │
│ QA         │ ⚪ Waiting    │ Testing functionality   │ 1 tasks  │
╰────────────────────────────────────────────╯

╭─ 💬 Agent Activity ─────────────────────────────────────────────╮
│ 14:23:15 🤖 Engineer started: Writing main.py                   │
│ 14:23:18 📄 Engineer create: src/main.py                        │
│ 14:23:22 💬 Engineer: I've implemented the core CLI structure   │
│ 14:23:25 ✅ Engineer completed: Created main.py                 │
│ 14:23:28 🤖 Architect started: Reviewing structure              │
╰─────────────────────────────────────────────────────────────────╯

6. Get Your Results

Find your generated code in:

• my-project/workspace/ - Your complete, working application
• my-project/logs/ - Full conversation transcripts and development history

💰 Cost-Efficient Development

AutoSquad now includes intelligent token optimization that reduces OpenAI API costs by up to 69%:

Before vs After Token Usage

Traditional Approach:          AutoSquad Optimized:
Round 1: 1,200 tokens         Round 1: 1,200 tokens  
Round 2: 2,800 tokens         Round 2: 1,800 tokens ⬇️ 36% reduction
Round 3: 6,500 tokens         Round 3: 2,200 tokens ⬇️ 66% reduction  
Round 4: 15,000 tokens        Round 4: 2,600 tokens ⬇️ 83% reduction
Total: 25,500 tokens ($0.77)  Total: 7,800 tokens ($0.23) ⬇️ 69% savings

Real-time cost monitoring keeps you informed:

╭─ 💰 Cost Summary ───────────────────────────────────────────────╮
│ 💰 Token Usage Summary                                          │
│ Total Tokens: 23,492                                           │
│ API Calls: 31                                                   │
│ Estimated Cost: $0.7048                                        │
│ Avg Tokens/Call: 758                                           │
╰─────────────────────────────────────────────────────────────────╯

🛠️ Advanced Usage

Squad Profiles

Choose the right team for your project:

# MVP Team - Fast prototyping (PM + Engineer + Architect)
autosquad run --squad-profile mvp-team --project my-app

# Full Stack Team - Complete applications (All agents + extended tools)
autosquad run --squad-profile full-stack --project my-app

# Research Team - Experimental projects (PM + Engineer + QA)
autosquad run --squad-profile research-team --project my-app

Model Selection

# Use different models based on your needs and budget
autosquad run --model gpt-4-turbo --project my-app      # Best quality
autosquad run --model gpt-4 --project my-app            # Balanced
autosquad run --model gpt-3.5-turbo --project my-app    # Most economical
autosquad run --project projects/flowfoundry-marketing-site --model gpt-4o-mini --rounds 3 --squad-profile full-stack

Development Rounds & Reflection

# Run more rounds for complex projects
autosquad run --project my-app --rounds 5

# Disable reflection for faster development
autosquad run --project my-app --no-reflect

# Enable verbose logging for debugging
autosquad run --project my-app --verbose

🏗️ Architecture & Technical Features

Agent Coordination

• AutoGen Foundation: Built on Microsoft's proven multi-agent framework
• Round-Robin Collaboration: Structured conversation flow ensures all agents contribute
• Context-Aware Agents: Each agent maintains project awareness and specialized knowledge
• Tool Integration: File operations, code execution, and workspace management

Token Optimization Engine

• Intelligent Context Compression: Keeps recent messages while summarizing older content
• Smart Message Prioritization: Maintains conversation flow with minimal token usage
• Real-Time Usage Tracking: Monitor costs and usage patterns as development happens
• Automatic Optimization: No manual configuration required

Live Progress System

• Real-Time Agent Dashboard: See which agents are active and what they're doing
• Activity Stream: Watch file creation, conversations, and progress live
• Performance Metrics: Track actions completed, files created, and productivity
• Professional Terminal UI: Rich formatting with colors, panels, and layouts

Project Management

• Workspace Isolation: Each project gets its own dedicated workspace
• Conversation Logging: Complete transcripts of all agent interactions
• File Versioning: Automatic backups and change tracking
• State Persistence: Resume projects and maintain context across sessions

📊 What Gets Built

AutoSquad agents create production-ready applications with:

Code Quality

• ✅ Clean, readable code with proper structure and comments
• ✅ Error handling and edge case management
• ✅ Best practices for the chosen technology stack
• ✅ Documentation and usage instructions

Project Structure

• ✅ Organized file hierarchy with logical folder structures
• ✅ Configuration files (requirements.txt, package.json, etc.)
• ✅ README files with setup and usage instructions
• ✅ Basic tests for core functionality

Deliverables

• ✅ Working applications that solve the specified problem
• ✅ Installation instructions and dependency management
• ✅ Usage examples and API documentation
• ✅ Development logs showing the complete thought process

🎯 Example Projects

CLI Tools

"Build a Python CLI that converts between JSON, CSV, and YAML formats"
→ Complete CLI with argparse, error handling, and format validation

Web Applications

"Create a Flask web app for task management with user authentication"
→ Full Flask app with database, auth, REST API, and frontend

Data Processing

"Build a data pipeline that processes CSV files and generates reports"
→ Python pipeline with pandas, validation, and HTML report generation

Utilities

"Create a file organizer that sorts downloads by file type"
→ Cross-platform utility with configuration and scheduling

🔧 Configuration

Environment Setup

# Required
export OPENAI_API_KEY="your-api-key"

# Optional
export AUTOSQUAD_CONFIG_DIR="~/.autosquad"  # Custom config location
export AUTOSQUAD_LOG_LEVEL="INFO"          # Logging level

Custom Squad Profiles

Create ~/.autosquad/squad_profiles.yaml:

profiles:
  my-custom-team:
    agents:
      - type: pm
        config:
          focus: "user experience"
      - type: engineer  
        config:
          languages: ["python", "typescript"]
          frameworks: ["fastapi", "react"]
      - type: qa
        config:
          focus: ["performance", "security"]
    workflow:
      rounds: 4
      reflection_frequency: 2

🚀 Performance & Efficiency

Speed Optimizations

• Parallel Agent Operations: Multiple agents can work simultaneously
• Intelligent Context Management: Faster API calls through optimized prompts
• Async Architecture: Non-blocking operations for better responsiveness
• Smart Caching: Reduce redundant API calls and computations

Cost Management

• Token Usage Monitoring: Real-time tracking prevents budget overruns
• Context Optimization: 69% average reduction in token usage
• Model Selection: Choose the right model for your budget and quality needs
• Usage Analytics: Detailed breakdowns help optimize future projects

🔮 Roadmap

Immediate (v0.3)

• Code Execution Environment: Sandboxed testing and validation
• Git Integration: Automatic commits and version control
• More Agent Types: Database, DevOps, and Security specialists
• Plugin System: Custom tools and integrations

Short Term (v0.4)

• Web Interface: Browser-based project management and monitoring
• Team Templates: Pre-configured squads for common project types
• Advanced Analytics: Performance insights and optimization suggestions
• Multi-Language Support: Beyond Python to Node.js, Go, Rust

Long Term (v1.0)

• Distributed Execution: Scale to larger, more complex projects
• Learning System: Agents improve based on project feedback
• Enterprise Features: Team management, audit logs, compliance
• Marketplace: Share and discover community squad profiles

📚 Documentation

• Token Optimization Guide - Detailed guide to cost savings
• Architecture Overview - Technical implementation details
• Agent Design - How agents work and interact
• Project Status - Current development status
• Contributing - How to contribute to AutoSquad

🤝 Contributing

AutoSquad is open source and welcomes contributions! Whether you're:

• 🐛 Reporting bugs or suggesting features
• 📝 Improving documentation or examples
• 🔧 Adding new agent types or tools
• 🎨 Enhancing the UI or user experience

Check out our Contributing Guide to get started.

📄 License

AutoSquad is released under the MIT License. See LICENSE for details.

🙏 Acknowledgments

• Microsoft AutoGen - The powerful multi-agent framework that makes AutoSquad possible
• OpenAI - GPT models that power our intelligent agents
• Rich Library - Beautiful terminal UI components
• The Open Source Community - For tools, libraries, and inspiration

---

Ready to build something amazing? 🚀

autosquad run --project your-next-big-idea --rounds 3

Watch AI agents turn your ideas into reality, efficiently and transparently.