Getting started

Choose a language and a backend before installing an example. Each linked tutorial supplies its complete dependency setup, files, run commands and expected results.

Neo4j Labs project

This project is part of Neo4j Labs and is actively maintained, but not officially supported. There are no SLAs or guarantees around backwards compatibility and deprecation. For questions and support, please use the Neo4j Community Forum.

Python examples use the published neo4j-agent-memory==0.7.0 package and complete scripts shown on each page. TypeScript tutorials retain their source-build setup; Python and npm versions are independent. Follow Python installation or TypeScript setup.

Hosted memory with NAMS

Use a workspace-bound API key from the NAMS dashboard — sign up for a workspace, no local database to run. Keep the key on the server and select the intended workspace before writing memory. See authentication and key categories.

Python: store and retrieve hosted memory.
TypeScript: connect to hosted memory. The current TypeScript package requires Node.js 22 or later; check example-specific engine requirements too.

Store the conversation ID returned by the service when you want to resume it. A matching user identifier alone does not resume a conversation. Preferences, facts and client-created relationships are not hosted REST operations; see the capability reference.

Python with Neo4j Aura

Use a dedicated AuraDB instance with the bolt backend when you need client-side extraction, preferences, facts, geospatial queries or custom write Cypher — you provide and connect to the Neo4j instance yourself. Start with the Aura connection setup.

Build your first memory context provides one complete setup path.
Build a conversation application continues into model use.
Use Anthropic and local embeddings teaches a different provider setup in its own lesson.

Provider dependencies are explicit extras. The Python [all] bundle does not include every local model package; install the extra named by the selected lesson rather than relying on packages already in your environment. The Python extras table lists each extra and what it enables.

Configure and integrate

Configuration and environment variables describe the actual settings, aliases and precedence.
Python SDK and TypeScript SDK describe installation and public entry points.
Python framework integrations and TypeScript integrations help connect an existing agent.
Python MCP tutorial and TypeScript MCP tutorial expose memory through supported tools.

Copy and run an example

The Python tutorials show every required local file alongside the published package install. Create the lesson folder, copy each complete program or helper into its labelled filename, and run the displayed commands. No code download is required. Start with the Aura lesson or the NAMS lesson. TypeScript source examples require building the SDK before installing their local file: dependency; follow the TypeScript setup.

Verify your first result

Finish the selected tutorial’s storage/readback check before adding a framework or changing providers. When extraction is asynchronous, wait for the documented readiness condition before inspecting entities. If a step fails, keep its full error output and use the FAQ and the relevant SDK troubleshooting guide.