This repository contains a collection of projects demonstrating the power and flexibility of the Model Context Protocol (MCP) for building modular, agentic AI systems.
A single-server MCP implementation that separates client reasoning from server skills.
- Server: Fetches live weather data from OpenWeatherMap.
- Client: An agent that queries the server to answer user questions.
- Features: Live data fetching, server-defined prompts for complex tasks, and resource reading.
A multimodal agent architecture that orchestrates two specialized servers.
- Visual Analysis Server: Uses Google Gemini to analyze and describe images.
- Wikipedia Research Server: Researches topics based on text queries.
- Client: A Gradio web UI that connects visual input with deep research capabilities.
A Retrieval-Augmented Generation (RAG) system for querying private internal documents.
- RAG Server: Ingests documents, creates vector embeddings (ChromaDB), and performs semantic search.
- Client: An expert assistant that answers questions based only on the provided internal knowledge base (e.g., Employee Handbook).
Each folder contains its own README.md with specific setup and running instructions.