Agents, MCPs & LangGraph - A Tutorial

This repository serves as the official companion for the agentic systems tutorial presented at GCPR 2025. The content focuses on two key frameworks for building robust, multi-step agents: LangGraph and Model Context Protocol(MCP).

All code and materials are provided to facilitate a hands-on learning experience.

Quick start

All the sessions can be run in Google Colab. Click on the icons below to open the respective notebooks in Colab and get started.

Session	Notebook/Folder name	Code along	Full solution
Intro to langgraph	01_basic_langgraph_chatbot	-
Langgraph fundamentals	02_chat_with_pdfs_and_youtube
MCP	hello-mcp/		-
Orchestration	03_deepresearch	-

Note

If using Colab, make sure to run the first cell in each notebook to install the required packages, since every Colab notebook runs in a fresh environment.

Remember to add your API keys in the .env file created in the first cell.

Local setup

For those feeling more adventurous, follow the instructions below to set up the repo locally on your system

1. Clone the repo

       git clone https://github.com/TuebingenAICenter/agent-tutorial-dev.git
       cd agent-tutorial

2. Create and activate a virtual environment using a python manager of your choice. Install the required packages. (Instructions for uv and venv are listed below)
3. Create a copy of the env.example file and rename it to .env. Add your API keys to this file.

venv

    # We do recommend using python 3.12, but any version >=3.10 should work
    python3.12 -m venv .venv
    source .venv/bin/activate
    pip install -r requirements.txt

uv [Recommended]

    uv venv -p 3.12
    source .venv/bin/activate
    uv pip install -r requirements.txt

Authors and collaborators

Peter Gehler Prof. Tech Transfer @Tübingen AI Center	Matthias Kümmerer Postdoc @bethgelab	Linus A. Schneider Master's student @Uni Tuebingen
Jaisidh Singh Master's student @Uni Tuebingen	Robin Ruff SE @Tübingen AI Center	Mamen SE @Tübingen AI Center