Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/neo4j-labs/create-context-graph/llms.txt

Use this file to discover all available pages before exploring further.

1

Check prerequisites

You need the following before running the CLI:
  • Python 3.11+ — the CLI is a Python package. uv is recommended for running it without a permanent install.
  • Node.js 18+ — required for the generated Next.js frontend.
  • Neo4j 5+ — see step 3 for your options (cloud, Docker, or local).
  • An LLM API key — most frameworks use Anthropic. Get one at console.anthropic.com.
If you choose the OpenAI Agents SDK framework, you need OPENAI_API_KEY. For Google ADK (Gemini), you need GOOGLE_API_KEY.
2

Run the CLI

No installation needed. Run directly with uvx (Python) or npx (Node.js):
uvx create-context-graph
The interactive wizard guides you through selecting a domain, agent framework, and Neo4j connection. If you prefer to skip it, pass all required flags directly:
uvx create-context-graph my-app \
  --domain healthcare \
  --framework pydanticai \
  --demo-data
Omit the project name and the CLI auto-generates one from your domain and framework (e.g., healthcare-pydanticai-app).
Other non-interactive examples:
# Custom domain from a plain-English description
uvx create-context-graph my-app \
  --custom-domain "veterinary clinic management" \
  --framework pydanticai \
  --anthropic-api-key $ANTHROPIC_API_KEY \
  --demo-data

# Import real data from SaaS services instead of demo data
uvx create-context-graph my-app \
  --domain personal-knowledge \
  --framework pydanticai \
  --connector github \
  --connector slack

# Scaffold, reset the database, and ingest data in one step
uvx create-context-graph my-app \
  --domain financial-services \
  --framework langgraph \
  --demo \
  --neo4j-uri neo4j://localhost:7687
When the CLI finishes, you’ll see a summary of what was generated and the exact commands to run next.
3

Set up Neo4j

The generated app connects to any Neo4j 5+ instance. Choose the option that fits your workflow:
The easiest option. Neo4j Aura provides a free cloud-hosted instance with no local setup required.
  1. Go to console.neo4j.io and create a free instance.
  2. Download the .env credentials file from the Aura console.
  3. Either pass it at scaffold time:
uvx create-context-graph my-app \
  --domain healthcare \
  --framework pydanticai \
  --neo4j-aura-env path/to/Neo4j-credentials.env \
  --demo-data
Or copy the credentials into your project’s .env file manually after scaffolding.
No make neo4j-start or make docker-up needed — Aura is already running in the cloud.
4

Configure the environment

The generated project includes a .env.example with all required variables. Copy it and fill in your credentials:
cd my-app
cp .env.example .env
Then open .env and set your values:
# Neo4j connection
NEO4J_URI=neo4j://localhost:7687
NEO4J_USERNAME=neo4j
NEO4J_PASSWORD=password

# LLM API key (use the key for your chosen framework)
ANTHROPIC_API_KEY=sk-ant-...
# OPENAI_API_KEY=sk-...       # Required for openai-agents
# GOOGLE_API_KEY=AIza...      # Required for google-adk
If you passed --neo4j-aura-env during scaffolding, your .env already has the Aura connection details filled in.
5

Install dependencies and seed data

Install the backend and frontend dependencies, then load the domain data into Neo4j:
cd my-app
make install   # Installs backend (uv) + frontend (npm) dependencies
make seed      # Seeds entities, relationships, documents, and decision traces
make install is equivalent to running:
# Backend
cd backend && uv venv && uv pip install -e .

# Frontend
cd frontend && npm install
make seed loads all fixture data: entities and relationships, professional documents, and multi-step decision traces.
If you used --demo (instead of --demo-data) during scaffolding, data was already ingested into Neo4j at generation time. You can still run make seed to re-seed.
6

Start the app

Start the backend and frontend with a single command:
make start
This starts:
  • Backend (FastAPI) on port 8000
  • Frontend (Next.js) on port 3000
Or start them separately in two terminals:
# Terminal 1 — backend
cd backend && make dev

# Terminal 2 — frontend
cd frontend && npm run dev
7

Explore your app

Once both services are running, open these URLs:

Chat interface

http://localhost:3000 — Chat with your AI agent and explore the knowledge graph. Try asking about entities in your domain, or double-click a node in the graph to expand its neighbors.

Backend API docs

http://localhost:8000/docs — FastAPI auto-generated interactive API documentation. Useful for testing endpoints directly.

Neo4j Browser

http://localhost:7474 — Query the graph directly with Cypher. Try MATCH (n) RETURN n LIMIT 25 to see the seeded data.

Health check

http://localhost:8000/health — Confirms the backend and Neo4j connection are healthy.
Things to try in the chat interface:
  • Ask a domain-specific question: “Show me all patients with a diabetes diagnosis” or “What are the highest-risk accounts?”
  • Watch the tool call timeline on the right as the agent executes Cypher queries in real time
  • Click any node in the graph to see its properties, then click Ask about [entity] to query the agent directly
  • Open the Documents panel to browse the seeded professional documents
  • Open the Decision Traces panel to see the pre-generated multi-step reasoning examples

​
Next steps

Neo4j setup options

Detailed guide to Aura, Docker, and neo4j-local setup

Custom domains

Generate a full ontology from a plain-English description

SaaS connectors

Import real data from GitHub, Slack, Jira, Notion, and more

CLI reference

Every flag and option, with defaults and examples

Build docs developers (and LLMs) love

Get started for freeTalk to us