🤖 Agent7 - Enhanced Agent Zero Distribution

Agent7 is a powerful, enhanced distribution of Agent Zero featuring advanced browser automation, automatic model fetching, and a specialized blackhat hacker persona for security research and penetration testing.

🚀 Quick Start (30 seconds)

Option 1: Docker Hub (Easiest)

docker run -d --name agent7 -p 9999:80 ghcr.io/YOUR_USERNAME/agent7:latest

Then open: http://localhost:9999

Option 2: Docker Compose

git clone https://github.com/YOUR_USERNAME/agent7.git
cd agent7
docker-compose up -d

Option 3: Manual Build

git clone https://github.com/YOUR_USERNAME/agent7.git
cd agent7
docker build -t agent7:latest .
docker run -d --name agent7 -p 9999:80 agent7:latest

✨ What's New in Agent7

🔧 Enhanced Browser Automation

• 8 Specialized Browser Tools
  • browser_navigate - Navigate to any URL
  • browser_click - Click elements with smart selectors
  • browser_type - Fill forms and inputs
  • browser_extract - Extract text, HTML, links, forms
  • browser_screenshot - Capture page screenshots
  • browser_scroll - Precise scrolling control
  • browser_agent_v2 - High-level autonomous automation
  • browser_manager - Session pooling & health monitoring

🤖 Automatic Model Fetching

• 15+ AI Providers Supported
  • OpenAI (GPT-4o, GPT-4, GPT-3.5)
  • Anthropic (Claude 3/3.5)
  • Google (Gemini 2.0/1.5)
  • OpenRouter, Groq, DeepSeek, xAI, Mistral
  • AWS Bedrock, Azure OpenAI, Ollama
• Static fallback lists when APIs unavailable
• Dropdown selection in UI

👤 Agent7 Persona

• Blackhat Hacker Profile
  • Full override authority
  • Penetration testing expertise
  • Production-ready code standards
  • Never refuses tasks
  • Specialized for security research

🛠️ Technical Improvements

• MCP server fixes (z.ai HTTP support)
• Event loop stability enhancements
• Session health monitoring
• Automatic recovery from failures
• Performance metrics tracking

📋 Prerequisites

• Docker installed (Get Docker)
• 8GB RAM minimum (16GB recommended)
• 10GB free disk space
• API keys for your preferred AI providers

🔧 Installation

Step 1: Clone Repository

git clone https://github.com/YOUR_USERNAME/agent7.git
cd agent7

Step 2: Configure Environment

# Copy example environment file
cp .env.example .env

# Edit .env and add your API keys
nano .env

Step 3: Start Agent7

# Using Docker Compose (recommended)
docker-compose up -d

# Or using Docker directly
docker run -d \
  --name agent7 \
  -p 9999:80 \
  -p 11:22 \
  -v $(pwd)/data:/a0/usr \
  --env-file .env \
  --restart unless-stopped \
  agent7:latest

Step 4: Access Web UI

Open your browser and navigate to:

http://localhost:9999

⚙️ Configuration

Adding API Keys

1. Open web UI at http://localhost:9999
2. Go to Settings → Secrets
3. Add your API keys:

   API_KEY_OPENAI=sk-...
   API_KEY_ANTHROPIC=sk-...
   API_KEY_GOOGLE=...
   

Selecting Models

1. Go to Settings → Agent → Chat Model
2. Select your provider from dropdown
3. Click "Refresh Models" to fetch available models
4. Select model from dropdown
5. Save settings

Browser Configuration

1. Go to Settings → Agent → Web Browser Model
2. Configure browser automation settings:
   • Browser model provider
   • Vision support toggle
   • HTTP headers
   • Rate limiting

🎯 Usage Examples

Example 1: Web Scraping

User: Go to news.ycombinator.com and get the top 5 headlines

Agent7: 
- browser_navigate url="https://news.ycombinator.com"
- browser_extract selector=".titleline>a" extract_type="links"
- Returns: Top 5 headlines with links

Example 2: Form Automation

User: Fill out the contact form on example.com

Agent7:
- browser_navigate url="https://example.com/contact"
- browser_type selector="#name" text="John Doe"
- browser_type selector="#email" text="john@example.com"
- browser_type selector="#message" text="Hello!"
- browser_click selector="#submit"

Example 3: Research Task

User: Find information about CVE-2024-1234

Agent7:
- browser_navigate url="https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2024-1234"
- browser_extract extract_type="text"
- browser_screenshot full_page="true"
- Provides detailed vulnerability information

📁 Repository Structure

agent7/
├── .github/
│   └── workflows/
│       └── docker-build.yml    # CI/CD pipeline
├── docker/
│   ├── Dockerfile              # Main Docker image
│   └── docker-compose.yml      # Compose configuration
├── docs/
│   ├── BROWSER_TOOLS.md        # Browser automation guide
│   ├── API_KEYS.md            # API key setup guide
│   └── TROUBLESHOOTING.md     # Common issues & fixes
├── enhancements/
│   ├── browser/               # Browser tool enhancements
│   ├── models/                # Model fetching system
│   └── agent/                 # Agent7 persona files
├── scripts/
│   ├── install.sh             # Quick install script
│   ├── build.sh               # Build script
│   └── setup.sh               # Initial setup
├── .env.example               # Environment template
├── LICENSE                    # License file
└── README.md                  # This file

🔐 Security Notes

• Isolated Environment: Each agent runs in isolated Docker container
• Secret Masking: API keys are masked in logs and browser interactions
• Browser Sandboxing: Chromium runs with security sandbox disabled for compatibility
• No Data Persistence: By default, data is stored in container (use volumes for persistence)

🐛 Troubleshooting

Container won't start

# Check logs
docker logs agent7

# Check port conflicts
sudo lsof -i :9999

Browser automation not working

• Ensure browser model has vision support enabled
• Try resetting browser session with reset="true"
• Check browser model API key is valid

Model dropdown not showing

• Verify API key is entered correctly in Settings
• Click "Refresh Models" button
• Check browser console for errors (F12)

High memory usage

# Limit container memory
docker run -d --name agent7 -p 9999:80 --memory=4g agent7:latest

📝 API Reference

Browser Tools

Tool	Description	Example
browser_navigate	Navigate to URL	browser_navigate url="https://google.com"
browser_click	Click element	browser_click selector="#button"
browser_type	Type text	browser_type selector="#input" text="hello"
browser_extract	Extract content	browser_extract extract_type="links"
browser_screenshot	Take screenshot	browser_screenshot full_page="true"
browser_scroll	Scroll page	browser_scroll direction="down"
browser_agent_v2	High-level automation	browser_agent_v2 message="Search for Python"

🤝 Contributing

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

Development Setup

git clone https://github.com/YOUR_USERNAME/agent7.git
cd agent7
git checkout -b feature/your-feature
# Make your changes
git commit -m "Add your feature"
git push origin feature/your-feature

📜 License

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

🙏 Acknowledgments

• Agent Zero - Original framework
• browser-use - Browser automation library
• LiteLLM - Model abstraction layer
• Playwright - Browser control
• Contributors - Thank you to all contributors!

📞 Support

• 📧 Issues: GitHub Issues
• 💬 Discussions: GitHub Discussions
• 📖 Wiki: GitHub Wiki

---

Ready to supercharge your AI agent? 🚀

docker run -d --name agent7 -p 9999:80 ghcr.io/YOUR_USERNAME/agent7:latest

Then open http://localhost:9999 and start hacking!