Multi-Agent AI Launchpad 🚀

A sophisticated multi-agent orchestration platform built with LangGraph and LangChain, designed to coordinate specialized AI agents for Infrastructure (Kubernetes) and Database (MySQL) management. The system utilizes the Model Context Protocol (MCP) to interact with distributed systems and provides a premium, interactive CLI experience.

---

🌟 Overview

The Multi-Agent AI Launchpad follows a Supervisor-Agent pattern. A central Supervisor agent acts as the brain, analyzing user requests and delegating specific tasks to specialized sub-agents. This modular architecture allows for better context management, reduced token usage, and higher reliability in complex troubleshooting and management scenarios.

🏗️ Architecture

• Supervisor: The orchestrator. It manages the conversation flow, makes routing decisions, and synthesizes final responses. Powered by DeepSeek-Chat.
• DB Assistant (🔬): A specialized agent for database operations. It can explore schemas, search metadata, and execute raw SQL queries via MCP.
• Kubernetes Assistant (✍️): A cloud-native specialist for Managing K8s clusters. It can monitor pods, check logs, manage namespaces, and scale resources via MCP.

---

📸 Screenshots

---

🚀 Key Features

• Centralized Orchestration: Seamless task delegation using LangGraph's routing and state management.
• MCP-Native: Deep integration with the Model Context Protocol (MCP) for secure, structured interaction with external tools and services.
• Premium CLI Experience: Built with rich, featuring responsive panels, streaming responses, and intuitive agent-specific styling.
• Disconnected Sub-graphs: Sub-agents run in isolated graphs with focused task descriptions, preventing context clutter and improving reasoning accuracy.
• Persistent State: Utilizes MemorySaver for thread-safe conversation persistence.

---

🛠️ Technology Stack

• Core Framework: LangGraph, LangChain
• LLM: DeepSeek (Main), OpenAI/NVIDIA (Configurable)
• Protocol: fastmcp (Model Context Protocol)
• UI/UX: Rich (Terminal Layouts), AsyncIO
• Environment: Python 3.10+

---

🔧 Installation & Setup

1. Prerequisites

• Python 3.10 or higher
• A .env file with your API keys (see .env.example or the template below)

2. Clone and Install

# Clone the repository
git clone <repository-url>
cd multi-agent

# Create and activate a virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

3. Configuration

Create a .env file in the root directory:

DEEPSEEK_API_KEY=your_deepseek_key
OPENAI_API_KEY=your_openai_key
# Optional: TAVILY_API_KEY, NVIDIA_API_KEY

---

🎮 Usage

Running the CLI

Launch the interactive supervisor using:

python supervisor/main.py

Example Queries

• Database: "List all user tables in the cbs-dev database and show me the first 5 records."
• Kubernetes: "Check the logs of the pod named backend-api in the production namespace."
• Hybrid: "Find the database connection error from the K8s logs and check the database schema for any table locks."

---

📦 Packaging

To create a standalone executable:

pyinstaller.exe multi_agent.spec --noconfirm

---

📂 Project Structure

multi-agent/
├── supervisor/             # Core logic
│   ├── main.py             # Entry point (CLI Loop)
│   ├── supervisor.py       # Orchestration & Graph definition
│   ├── dbhub_agent.py      # Database specialist
│   ├── kubernetes_agent.py  # K8s specialist
│   └── prompts/            # Agent system instructions (Markdown)
├── .env                    # Configuration (secrets)
├── README.md               # You are here
└── multi_agent.spec        # PyInstaller specification

---

📄 License

[Include your license information here]