Multi-Capable Processing (MCP) Smart Agent

Multi-Capable Processing (MCP) Smart Agent It is a modular and extensible AI-driven agentic server system that connects specialized agents through a central REST API. These agents can analyze code repositories, fetch external data (like weather), generate text summaries, and remember past interactions using a persistent memory manager.

---

🚀 Key Features

• Multi-Agent Architecture: Modular design with specialized agents for code analysis, data lookup, and summarization.
• Tool-Integrated Agents: Each agent uses tools like GitHub API, weather services, or basic NLP techniques.
• Memory System: Keeps a persistent memory of prior tasks for contextual recall.
• RESTful Server: Easily integrate with frontends, CLI tools, or workflows via HTTP.
• Pythonic Structure: Fully testable and extensible project layout.
• Ready for Scaling: You can plug in OpenAI, LangGraph, Vector Databases, and more.

---

🗂️ Project Structure

mcp-smart-agent/
│
├── agents/                  # AI agents for specific task domains
│   ├── code_agent.py        # Analyzes GitHub repositories
│   ├── data_agent.py        # Fetches weather data
│   └── summary_agent.py     # Summarizes input text
│
├── tools/                   # External service integrations
│   ├── github_tool.py       # Simulates GitHub API access
│   └── weather_tool.py      # Simulates weather data fetch
│
├── memory/
│   └── memory_manager.py    # In-memory key-value storage (can be extended)
│
├── server/
│   └── mcp_server.py        # Flask API endpoints to interact with all agents
│
├── tests/
│   └── test_agents.py       # Unit tests for core functionality
│
├── main.py                  # Entry point to start the server
├── requirements.txt         # Python dependencies
└── README.md                # You're reading it!

---

🧠 How It Works

The system spins up a Flask server that exposes endpoints corresponding to different agents:

1. CodeAgent (analyze GitHub repo)

• Extracts data from a GitHub-like repository (mocked).
• Returns high-level analysis (e.g., number of files).
• Saves the result in memory.

2. DataAgent (get weather data)

• Accepts a location input.
• Returns mock weather data (can be connected to OpenWeatherMap, etc.).

3. SummaryAgent (text summarizer)

• Accepts long text and returns a basic summary.
• You can extend this to use GPT or HuggingFace models.

4. MemoryManager

• Saves outputs for reuse.
• Supports simple key-value memory (can be upgraded to Redis or vector DB).

---

🔌 API Endpoints

| Method | Endpoint | Description | |--------|----------------------|--------------------------------------| | POST | /analyze_repo | Analyze a GitHub repo | | POST | /get_weather | Get mock weather data | | POST | /summarize | Summarize a block of text | | POST | /retrieve_memory | Retrieve stored memory for a task |

🔧 Example Usage

curl -X POST http://localhost:5000/analyze_repo \
  -H "Content-Type: application/json" \
  -d '{"repo_url": "https://github.com/example/repo"}'

---

🧪 Testing

Run unit tests with:

python -m unittest discover tests

---

🛠 Installation & Run

Prerequisites

• Python 3.7+
• pip installed

Install dependencies

pip install -r requirements.txt

Start the server

python server/mcp_server.py

---

Ideas for Expansion

• Replace mock tools with real APIs (GitHub, OpenWeather, LangChain tools).
• Use vector databases like Pinecone or ChromaDB for persistent memory.
• Add LangGraph for long-running planning workflows.
• Replace summary agent with GPT-4 or HuggingFace Transformers.
• Add authentication, logging, and rate-limiting.

🙋‍♂ Author

Made by Adad — an open-source AI agent framework for rapid prototyping and experimentation.