lxzl WORKFLOW ENGINE
Reference

Docs

Product guidance, workflow building patterns, engineering notes, and API reference.

Getting Started

lxzl is a visual workflow automation platform. Build pipelines by connecting nodes on a canvas, configure them with clicks, and run them instantly.

Quick Start

  1. Go to the Dashboard and click Create Workflow
  2. Choose a template or start blank
  3. On the canvas, add a start/source node, then use the compatibility-aware picker for the next node
  4. Click a node to use the right-side workbench: data coming in, settings/manipulation, then data coming out
  5. Use Run to here, Inspect, or Table View whenever you need to see the current shape
  6. Click Run, then read the Run Review trace at the bottom of the canvas

A First Useful Workflow

  1. Add a Trigger node and set it to manual while you build.
  2. Add an HTTP Request node, call a small JSON endpoint, and set a clear output key such as raw_customers.
  3. Add Filter or Transform to shape the array. Use the Field Picker instead of typing field paths by memory.
  4. Add Table View or Inspect so you can see the rows on the canvas after each run.
  5. Only after the workflow works manually, switch the trigger to webhook or cron and add notifications or downstream writes.

This order keeps the feedback loop tight: fetch data, inspect it, transform it, inspect again, then automate the schedule.

Data-First Building Loop

The fastest way to build is to keep one selected node and move down the right panel: confirm the incoming data, configure how this node changes it, then confirm the outgoing data. The Next actions in the output pane open a node picker scoped to the selected output field, so the next node is already filtered by what can actually consume that data.

Mental Model

A run is one growing data object. Nodes do not pass a private packet to the next node; they add named keys to shared run data. Downstream nodes can read any upstream key that reached them through the graph.

Node titleUsed to derive readable output keys when no explicit output field is set
Output fieldThe safest way to create stable references across edits and renamed nodes
{{field}}Template interpolation for prompts, URLs, messages, and action inputs
nodeName.fieldCross-node reference style exposed by NodeRefInput autocomplete

Core Concepts

Nodes

Nodes are the building blocks of a workflow. Each node performs a specific action: triggering, transforming data, making API calls, sending notifications, directory operations, AI agent loops, etc. There are 61 node types organized into 5 families.

Edges

Edges connect nodes and define the data flow. Data passes from left to right through connected nodes.

Execution

When you run a workflow, the engine walks the graph in topological order. Nodes at the same level run in parallel. Each node receives the accumulated data from all upstream nodes and can add or modify fields.