Start here
MemoryNode adds durable memory to AI apps: store facts and preferences once, retrieve them on every turn.
Goal: first stored memory and a successful search in under 5 minutes.
Pick your path
| Path | Best for | Time |
|---|---|---|
| Console quickstart | Fastest — no local setup | ~2 min |
| SDK quickstart | Node/TypeScript apps | ~3 min |
| Node HTTP quickstart | curl-style, any language | ~4 min |
| Python quickstart | Python backends | ~4 min |
| MCP | Cursor, Claude Desktop, agents | ~5 min |
One env var (SDK)
export MEMORYNODE_API_KEY="mn_live_..." # console.memorynode.ai → Settings → API keys
import { MemoryNode } from "@memorynodeai/sdk";
const memory = new MemoryNode(process.env.MEMORYNODE_API_KEY!);
await memory.remember("User prefers bullet points.");
const hits = await memory.recall("how should I format answers?", { explain: true });
Next reads
- Integration — the two REST calls every product uses
- Common use cases — companions, SaaS copilots, support bots
- Debugging — why a memory was retrieved, deduped, or superseded
- Troubleshooting — auth, 402, empty search
- TypeScript SDK — full client surface