🧠 MCP Research Assistant

A multi-agent, context-aware AI system built using the Model Context Protocol (MCP). This project is designed to explore and demonstrate agent orchestration, context tracking, and multi-step LLM workflows — all driven by modular and composable MCP definitions.

---

🚀 Project Goals

• Build an intelligent research assistant that can:

  • Interpret natural language queries
  • Find relevant academic papers (via arXiv/Semantic Scholar API)
  • Summarize each paper using an LLM
  • Combine results into a clear and final report

• Learn the architecture and hands-on implementation of:

  • ModelContext, ModelProtocol, ModelDefinition
  • Multi-agent workflows
  • Context propagation across agents
  • Modular AI system design using MCP

---

📦 Tech Stack

Component	Tool
LLM Backend	OpenAI / Claude / Local LLM
Agent Framework	LangChain, LangGraph (optional)
Protocol Design	Model Context Protocol (MCP)
Paper Search	Semantic Scholar API / arXiv API
Context Store	In-memory (or upgrade to Redis later)
Frontend (optional)	FastAPI / CLI

---

🧱 Architecture Overview

┌────────────────────────────┐
│        User Input          │
└────────────┬───────────────┘
             ▼
    ┌───────────────────┐
    │ Controller Agent  │ ◀── Creates context, splits tasks
    └────────┬──────────┘
             ▼
 ┌──────────────────────┐
 │   Research Agent     │ ◀── Calls paper search APIs
 └────────┬─────────────┘
          ▼
 ┌──────────────────────┐
 │   Summarizer Agent   │ ◀── Summarizes each paper
 └────────┬─────────────┘
          ▼
 ┌──────────────────────┐
 │  Synthesizer Agent   │ ◀── Aggregates & formats response
 └──────────────────────┘