🚀 Neo4j + Graphiti Agent Setup

This guide walks you through setting up a local development environment for working with Neo4j, Graphiti MCP Server, and a custom agent script.

---

📦 Prerequisites

• Install Docker
• Python 3.10 or above (for agent)
• uv for running async Python scripts

---

🧠 Step 1: Set Up Neo4j with Docker

🔹 Option 1: Use Docker Compose

docker compose up --build

Ensure you have a valid docker-compose.yml in your project root.

---

🔹 Option 2: Run with Persistent Volume and Custom Password

docker run \
  --publish=7474:7474 --publish=7687:7687 \
  --env NEO4J_AUTH=neo4j/your_password \
  --volume=/path/to/your/data:/data \
  neo4j:ubi9

✅ Replace /path/to/your/data with your local directory.
🔐 Replace your_password with a secure password of your choice.

---

🌐 Access Neo4j

Visit http://localhost:7474 in your browser.

• Username: neo4j
• Password: The one you set in NEO4J_AUTH

---

🕸️ Step 2: Clone & Run Graphiti MCP Server

git clone https://github.com/getzep/graphiti.git
cd graphiti/mcp_server
cp .env.example .env
uv run graphiti_mcp_server.py --model gpt-4o-mini --transport sse --group-id my-project

---

🤖 Step 3: Clone & Run This Agent Project

git clone https://github.com/ParthSinha-smartSense/Agent-with-memory-GDG.git
cp .env.example .env
python run agent.py

---

✅ Final Checklist

• Docker is installed and running
• Neo4j container is up
• Password is securely set and accessible
• Graphiti MCP Server is running
• Agent script is running and connected to MCP

---

🧩 Notes

• Ensure all .env files are properly configured.
• If using volumes, make sure the data directory exists to avoid mount errors.
• Use docker ps to check running containers.

---

📬 Support

For issues related to Graphiti, visit Graphiti GitHub Issues.

For this project, open an issue in this repo.