Which integrations are available for this server?

Provides tools for fetching recent arXiv papers by topic and retrieving paper abstracts by arXiv ID. Enables Google Gemini LLM to call MCP tools through an interactive agent interface.

How do I use arXiv MCP Server?

1. Click on "Install Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@arXiv MCP Server fetch recent papers on quantum computing" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

MCP Tutorial

Overview

This repository demonstrates a small MCP (Model Context Protocol) ecosystem built around FastMCP. It contains a sample MCP server exposing tools (arXiv fetch, abstract retrieval, save-markdown) and example clients: a simple FastMCP client and a Google Gemini-integrated agent. The code is intended as a practical tutorial and reference for running and testing an MCP server locally, integrating it with LLMs (Gemini), and using the MCP Inspector for interactive exploration.

Key points:

Server uses the fastmcp library to define and expose tools.
Client examples show how to call tools programmatically and how to attach an LLM (Gemini) to MCP tools.
Includes Docker-ready server and agent examples for containerized testing.

Quick start

Clone and install dependencies:

git clone https://github.com/kmkarakaya/mcp_tutorial.git; cd mcp_tutorial
pip install -r requirements.txt

Start the MCP server (local):
```
python 4_mcp_server.py
```
The server exposes tools and listens for client connections (default behavior provided by FastMCP).

Usage and examples

Inspect available tools with the official MCP Inspector (requires Node.js and npm or Corepack installed):
```
npx @modelcontextprotocol/inspector
```
Note: npx comes with npm (or use Corepack on newer Node versions). Install Node.js from https://nodejs.org/ if you don't have node/npm on your system.
Run the example client to call a tool programmatically:
```
python 5_mcp_client.py
```
Start an interactive Gemini-based agent that can call MCP tools (requires Google GenAI credentials):
```
python 6_mcp_gemini_agent.py
```
When using the Gemini agent, set your environment variables per google-genai docs (e.g., GOOGLE_API_KEY or equivalent) before running.

Docker

Dockerfiles/examples are included. Example build/run:

docker build -t mcp_server .
docker run -p 8000:8000 mcp_server

Tools implemented in the server

fetch_arxiv_papers(topic: str, number_of_papers: int = 3) — fetches recent arXiv papers for a topic.
get_arxiv_abstract(arxiv_id: str) — retrieves an arXiv paper abstract.
save_md_to_file(text: str, filename: str) — saves given markdown to ./reports.

Dependencies

See requirements.txt. Important packages:

fastmcp
google-genai (for Gemini integration)
uvicorn (if you run the server via ASGI)

Notes

The server file (4_mcp_server.py) prints helpful messages when functions are called and includes a mcp.run() entry point.
Use the MCP Inspector to verify tool metadata and try calls interactively.

A full video tutorial and repository walkthrough is available on the Murat Karakaya Akademi YouTube channel (live recording): https://youtube.com/live/c7yl0GS2mJQ?feature=share