Runnable examples demonstrating each integration shipped with
@neo4j-labs/agent-memory. Each example is a self-contained Node.js
script you can clone, install, and run against the hosted Neo4j Agent
Memory Service.
- Node.js 20+
- A
MEMORY_API_KEYfrom memory.neo4jlabs.com
| Folder | What it shows |
|---|---|
vercel-ai/ |
Memory-augmented chat using the Vercel AI SDK middleware |
mcp/ |
Registering the 12 memory tools with an MCP server |
langchain/ |
Chat history + entity retriever shapes for LangChain JS |
mastra/ |
Wrapping the client as a Mastra-compatible memory provider |
strands/ |
AWS Strands agent with session persistence, three-tier context, and reasoning capture |
cd vercel-ai
cp .env.example .env # then edit .env with your MEMORY_API_KEY
npm install
npm startEach example's own README has the full step-by-step.
- These examples use
"@neo4j-labs/agent-memory": "file:../.."to depend on the in-tree SDK during development — so contributors can iterate on both without an npm publish round-trip. When using one of these examples as a template for your own project, replace thefile:path with a published version:"@neo4j-labs/agent-memory": "^0.3.0" - Each example is type-checked in CI by the
type-check-examplesmatrix in.github/workflows/ci-typescript.yml— drift between an example and the SDK's public API fails the PR that introduces it. Runtime execution (npm start) still needs API keys you provide locally.