A customized Agent Zero AI agent system running in Docker. Built on Debian Trixie 13.3 with specialized agent profiles, an intuitive web UI, and Web3-native token inference support.

🐧 Platform Note: Deby Lite runs on Debian Trixie 13.3 (stable/testing), whereas the base Agent Zero framework runs on a Kali Linux flavored Docker image. Deby is optimized for clean Debian environments prioritizing stability and production deployments.

Deby Lite is a Web3 compliant AI agent that supports decentralized token-based inference — compatible with the venice.ai ecosystem.

• Pay for LLM inference using A0T or Diem tokens
• Compatible with venice.ai decentralized inference marketplace
• Token-gated agent capabilities and compute tiers
• Wallet integration for seamless on-chain inference billing
• Works alongside traditional API key providers (OpenAI, Anthropic, etc.)

• 🐧 Debian Trixie 13.3 - Stable production-grade Linux base (vs Kali in base Agent Zero)
• 🧠 Multiple Agent Profiles - Default, Agent0, Developer, Researcher, Hacker, Deby
• 🛠️ Built-in Tools - Code execution, browser control, memory management, file operations, web search
• 🌐 Web UI - Full-featured chat interface with settings, scheduler, and project management
• 💡 Skills System - Extensible skill framework (AgentSkills.io standard)
• 📅 Task Scheduler - Cron-based, planned, and ad-hoc task scheduling
• 🔌 A2A Protocol - Agent-to-agent communication support
• 🔍 Vector Memory - FAISS-powered persistent memory with semantic search
• 🖥️ Multi-LLM Support - OpenAI, Anthropic, Google, Ollama, OpenRouter + more
• 🌐 Web3 Ready - A0T & Diem token inference payments

deby-lite/
├── agent.py              # Core agent logic
├── models.py             # LLM model definitions
├── run_ui.py             # Web UI server entry point
├── initialize.py         # System initialization
├── start_deby_lite.sh    # Docker startup script
├── requirements.txt      # Python dependencies
├── agents/               # Agent profile definitions
│   ├── agent0/           # Primary user-facing agent
│   ├── developer/        # Code-focused agent
│   ├── researcher/       # Research-focused agent
│   ├── hacker/           # Cybersecurity agent
│   ├── deby/             # Deby custom agent
│   └── _example/         # Template for custom agents
├── prompts/              # System prompt templates
├── conf/                 # Configuration files
├── python/               # Framework internals
│   ├── api/              # REST API endpoints
│   ├── tools/            # Agent tool implementations
│   ├── helpers/          # Utility modules
│   └── extensions/       # Hook-based extensions
├── webui/                # Frontend web interface
├── skills/               # Installed skills
│   └── create-skill/     # Skill creation wizard
├── knowledge/            # Knowledge base (gitkeep)
└── lib/                  # Browser automation scripts

docker run -p 8080:8080 \
  -e A0_ROOT=/deby \
  -v ./usr:/deby/usr \
  deby-lite

# Install dependencies
pip install -r requirements.txt

# Start the web UI
bash start_deby_lite.sh

Then open http://localhost:8080 in your browser.

Copy and edit the environment file:

cp .env.example usr/.env
# Add your API keys (OpenAI, Anthropic, etc.) or configure Web3 wallet for venice.ai token inference

Skills extend agent capabilities. Install from AgentSkills.io or create your own using the built-in create-skill wizard.

Based on Agent Zero by frdel.

Deby Lite is a fork and customization of the Agent Zero project.

All core framework code, architecture, tooling, Web UI, memory systems, and agent infrastructure are the original work of frdel and the Agent Zero contributors.

Deby Lite builds upon Agent Zero by providing:

• A Debian Trixie 13.3 base image instead of Kali Linux
• Extended agent profiles tailored for production and web development
• Custom skill integrations for VPS and Joomla management
• Web3 token inference compatibility (A0T + Diem)

If you find Agent Zero useful, please ⭐ star the original repo: https://github.com/frdel/agent-zero