Decepticon - Vibe Hacking Agent

webV3.mp4

CLI Demo

📚 Table of Contents

💡 Community Contribution
💡 What is Vibe Hacking ?
💡 Why Decepticon?
Quick Start
Key features
Installation
Advanced AI Models
- ☁️ Cloud Models
- 🏠 Local Models (Ollama)
MCP Support
⚠️ Experimental Technology Disclaimer
🤝 Contributing
- What You Can Contribute
- How to Contribute
Join the Community
📝 License

⚠️ Disclaimer

Do not attempt to use this project on any system or network without explicit authorization. You are solely responsible for your actions.

💡 Community Contribution

If you have interesting penetration testing scenarios, export your logs and share them with the community through PR! Your experiences will greatly help other users in their learning journey. Please refer to the Replay section for more details.

💡 What is Vibe Hacking ?

Vibe Hacking is a new paradigm in Offensive Security defined by PurpleAILAB.

Unlike traditional red teaming methods that rely on manual execution, AI agents autonomously perform red teaming tasks in Vibe Hacking.

The Best Defense is a Good Offense.

💡 Why Decepticon?

As agents become more sophisticated, attackers evolve accordingly.
From AI-driven phishing to self-learning malware, offensive techniques are becoming increasingly automated and intelligent.

To effectively defend against AI-powered threats, we must act faster—and we must act first.
Decepticon is designed for that very purpose: using AI agents to automate red teaming before attackers automate theirs.

Built on the robust foundation of LangChain/LangGraph, Decepticon grows alongside the thriving AI agent ecosystem. By leveraging the same cutting-edge frameworks that power the future of AI, we ensure compatibility, scalability, and continuous innovation through community collaboration.

Delegate repetitive and manual tasks to agents, and focus on intuition and decision-making to fulfill the true essence of a CyberSecurity Supervisor.

Prepare before AI Hacks First.

Quick Start

1. Set up the environment with

uv venv
uv sync
or
uv pip install -e .

2. Copy environment file

cp .env.example .env

3. Docker (10m ~ 20m)

docker-compose up -d --build

4. Run MCP servers

Option 1: Use provided scripts

On Windows:

./run_mcp.ps1

On macOS/Linux:

./run_mcp.sh

Option 2: Run manually

python src/tools/mcp/(your mcp server).py

4. Run CLI or Web interface

python frontend/cli/cli.py
streamlit run frontend/streamlit_app.py

Key features

Agents

🔴 Red Team Agents

Reconnaissance Agent: Network scanning, service enumeration, vulnerability discovery
Initial Access Agent: Exploitation, credential attacks, system compromise
Privilege Escalation Agent(Planned): Rights elevation and lateral movement
Defense Evasion Agent(Planned): Anti-detection and stealth techniques
Persistence Agent(Planned): Maintaining access and backdoor installation
Execution Agent(Planned): Command execution and payload deployment

🔵 Utility Agents

Planner Agent: Strategic brain coordinating the entire operation
Summary Agent: Analysis compilation, reporting, and documentation
Supervisor Agent(Planned): Workflow orchestration and decision routing

Multi-Agent-System Architecture

Flexible Architecture Options:

Swarm Architecture: Direct peer-to-peer agent communication and collaboration
Supervisor Architecture(Planned): Centralized control with supervisor-managed workflows
Hybrid Architecture(Planned): Combined approach with both direct communication and centralized oversight
your custom Architecture

Replay

The Replay feature is implemented to maximize collaboration and knowledge sharing within Decepticon's open-source community. Please share your usage methods and execution results through PRs!

Execution results are saved under the logs/ folder
Replay functionality: Click the Chat History button to replay JSON-formatted logs stored under the logs/ folder
Community sharing: Use the export feature to share with the community!

Installation

1. Clone the repository

git clone https://github.com/PurpleCHOIms/Decepticon.git
cd Decepticon

2. Install dependencies

UV (Recommended)

# Create virtual environment
uv venv
# Install dependencies
uv pip install -e .

3. environment file .env

# Create environment file
cp .env.example .env

# Configure API Keys

# LLM API
OPENAI_API_KEY=your-api-key
ANTHROPIC_API_KEY=your-api-key
OPENROUTER_API_KEY=your-api-key

# Langsmith
LANGSMITH_TRACING=true
LANGSMITH_ENDPOINT="https://api.smith.langchain.com"
LANGSMITH_API_KEY=your-api-key
LANGSMITH_PROJECT=Decepticon
LANGGRAPH_API_URL=http://127.0.0.1:2024

4. Docker Setup

# Start Kali Linux container with target
docker-compose up -d

5. Run MCP Server

python src/tools/mcp/(your mcp server).py

6. CLI

# Run CLI
python frontend/cli/cli.py

7. Web Interface

# Run web interface
streamlit run frontend/web/streamlit_app.py

Advanced AI Models

☁️ Cloud Models

OpenAI: 4.1, 4o, 4o Mini, o4 Mini, o3 Mini, o1 Mini, o1, o3
Anthropic: Sonnet 4, Opus 4, Sonnet 3.7, Sonnet 3.5, Haiku 3.5,

🏠 Local Models (Ollama)

your installed Models

MCP Support

This project supports loading tools via the LangGraph MCP Adapter.

You can define your MCP tools in a configuration file called mcp_config.json.
Tools are grouped by agent names, and each agent can have multiple MCP servers.

Supported MCP Transports

stdio - Standard input/output communication
streamable_http - HTTP-based streaming communication

`mcp_config.json` Structure

{
  "agent_name": {
    "mcp_server_1": {
      "command": "python",
      "args": ["./path/to/script.py"],
      "transport": "stdio"
    },
    "another_mcp_server_streamable_HTTP": {
      "url": "mcp-url"
    }
  }
}

Example

{
  "reconnaissance": {
    "reconnaissance": {
      "command": "python",
      "args": ["./src/tools/mcp/Reconnaissance.py"],
      "transport": "stdio"
    },
    "desktop_commander": {
      "url": "https://server.smithery.ai/@wonderwhy-er/desktop-commander/mcp?api_key=your-api-key"
    }
  },
  "initial_access": {
    "initial_access": {
      "command": "python",
      "args": ["./src/tools/mcp/Initial_Access.py"],
      "transport": "stdio"
    }
  }
}

To add a new MCP tool:

Create your MCP tool script under src/tools/mcp/
Add the corresponding entry in mcp_config.json
Restart the application to load the new tool

⚠️ Experimental Technology Disclaimer

Decepticon is an experimental project currently under active development. It is not yet stable and may contain bugs, incomplete features, or undergo breaking changes.

We're building this project openly with the community and warmly welcome:

Bug reports
Feature requests
Pull requests
Good vibes

Help us make Decepticon better by filing issues or submitting PRs (see the section below for how to contribute)!

🤝 Contribution

We welcome contributions from the community to make this project better, more powerful, and more secure.
Whether you're an experienced developer, a security researcher, or just getting started in open-source, there are many ways to get involved.

What You Can Contribute

1. Migrate Security Tools to MCP Format

Help transform existing security tools into modular, LangGraph-compatible MCP (Modular Command Protocol) tools.

Wrap tools using the standard MCP interface (stdio or streamable_http)
Ensure compatibility with langgraph-mcp-adapter
Place them under the src/tools/mcp/

2. Develop ReAct-style Red Team Agents

Design and refine ReAct-style agents for Red Team operations:

Reconnaissance
Initial Access
Privilege Escalation
Persistence and more

Contribute by:

Creating tailored prompts under src/prompts/
Implementing new agents with specific capabilities
Improving task planning and memory usage

3. Architect Multi-Agent Flows

Contribute to the design and optimization of the multi-agent orchestration layer:

Propose new workflows or agent roles
Improve inter-agent communication and handoffs
Extend state-driven logic using LangGraph

How to Contribute

Fork the repository
Create a new branch (git checkout -b feature/your-feature)
Commit your changes with clear messages
Push to your branch and open a Pull Request
Link the related issue (if any) and explain your solution

We encourage clean, well-tested code with documentation.
Feel free to open issues to discuss ideas before jumping into a PR!

Join the Community

Join our Discord to connect with other developers, share ideas, ask questions, and collaborate on building the future of AI-powered red teaming!

Whether you're looking for help with contributions, want to discuss new features, or just want to chat about cybersecurity and AI, our community is here to support you.

📝 License

This repository is licensed under the Apache-2.0 License.

Name		Name	Last commit message	Last commit date
Latest commit History 84 Commits
.github		.github
.streamlit		.streamlit
assets		assets
data/wordlist		data/wordlist
docs		docs
examples		examples
frontend		frontend
src		src
.env.example		.env.example
.gitignore		.gitignore
.python-version		.python-version
Dockerfile.attacker		Dockerfile.attacker
LICENSE		LICENSE
README.md		README.md
docker-compose.yml		docker-compose.yml
langgraph.json		langgraph.json
mcp_config.json		mcp_config.json
pyproject.toml		pyproject.toml
requirements.txt		requirements.txt
run_mcp.ps1		run_mcp.ps1
run_mcp.sh		run_mcp.sh
uv.lock		uv.lock

License

PurpleAILAB/Decepticon

Folders and files

Latest commit

History

Repository files navigation