Not familiar with Dataflow concept? Check out the Dataflow article to get up to speed

Dataflow3D ConfiguratorAllmatterPlugin

Install dependencies

npm i rete rete-engine

Prepare nodes

Let's take a look at a simplified example of a graph with two node types: NumberNode and AddNode. These nodes are built exclusively for processing (on the server-side, e.g.) and don't have any integrations with the user interface. You can find a link to the complete example with the UI at the end of the article.

const socket = new ClassicPreset.Socket("socket");
class NumberNode extends ClassicPreset.Node {
  constructor(public value: number) {
    this.addOutput("value", new ClassicPreset.Output(socket, "Number"));
  data(): { value: number } {
    return { value: this.value };
class AddNode extends ClassicPreset.Node {
  constructor() {
    this.addInput("left", new ClassicPreset.Input(socket, "Left"));
    this.addInput("right", new ClassicPreset.Input(socket, "Right"));
    this.addOutput("value", new ClassicPreset.Output(socket, "Number"));
  data(inputs: { left?: number[]; right?: number[] }): { value: number } {
    const { left, right } = inputs;
    const value = (left && left[0] || 0) + (right && right[0] || 0)
    return { value };
class Connection<
  A extends Node,
  B extends Node
> extends ClassicPreset.Connection<A, B> {}
type Node = NumberNode | AddNode;
type ConnProps = Connection<NumberNode, AddNode> | Connection<AddNode, AddNode>;
type Schemes = GetSchemes<Node, ConnProps>;


import { DataflowEngine } from "rete-engine";
import { NodeEditor } from "rete";
const editor = new NodeEditor<Schemes>();
const engine = new DataflowEngine<Schemes>();

Add nodes and connections

const a = new NumberNode(1);
const b = new NumberNode(1);
const sum = new AddNode();
const con1 = new Connection(a, "value", c, "left");
const con2 = new Connection(b, "value", c, "right");
await editor.addNode(a);
await editor.addNode(b);
await editor.addNode(sum);
await editor.addConnection(con1);
await editor.addConnection(con2);

Start the processing

Retrieve output data from the sum node.

const result = await engine.fetch(

The value of result will be { value: 2 }, which is the sum of the initial input values of the sum node.

If you want to modify a.value or b.value, you need to clear the cache before processing the graph again. The output values of nodes are cached to avoid repetitive node execution.

engine.reset() // reset all nodes
// or specific nodes

Additionally, the data methods can be async. In such cases, the sum node will wait for the data methods of its input nodes to complete execution. After all of them have returned a value, the engine will execute the data method of the sum node

Check out the complete result on the Dataflow example page.