---
title: MCP Research
emoji: 🚀
colorFrom: red
colorTo: red
sdk: docker
app_port: 8501
tags:
- streamlit
pinned: false
short_description: Streamlit template space
---

# AI-Powered Scientific Research Companion

This repository implements an AI-driven research assistant using the Model Context Protocol (MCP). It enables end-to-end capabilities:

- 🔍 **Discovery & Ingestion**: Search journals, preprint servers, and web via MCP servers (e.g. `pskill9/web-search`, `metatool-ai/metatool-app`).
- 📓 **Reproducible Notebook**: Retrieve and run code snippets for core experiments via a Python-run MCP server.
- 🔗 **Knowledge Graph**: Build and visualize relationships (methods, datasets, concepts) via a vector-store MCP (e.g. Chroma).
- 🔧 **Provenance Tracking**: Store metadata and run history in a SQL database via MCP.

## Getting Started
1. Clone this repo:
   ```bash
   git clone https://huggingface.co/spaces/your-username/MCP_Research
   cd MCP_Research