Workflow Execution — How Drag & Drop Flows Run

🔀 Workflow • javascript • v1 • 2026-04-01

How the visual workflow system works in Flowork. Workflows are DAGs (Directed Acyclic Graphs) of nodes connected by edges. The internal/workflow Go module orchestrates execution by topologically sorting nodes and running them in dependency order. Data flows through connections between output and input handles.

In this guide, we cover "Workflow Execution — How Drag & Drop Flows Run" for the Flowork AI Knowledge Base. Category: workflow. Language: javascript.

Architecture Overview

Workflow system: (1) Frontend — Vue Flow canvas with drag-and-drop nodes, saved as JSON with nodes[] and edges[] arrays. (2) Backend — internal/workflow module in Go receives workflow JSON, sorts nodes topologically, and executes each node's execute.js sequentially. (3) Data passing — output from node A's handles is mapped to input handles of node B via edge connections.

Key Patterns

POST /api/workflow/save — Save workflow JSON { nodes, edges, name, id }
POST /api/workflow/execute/:id — Trigger workflow execution
GET /api/workflow/list — List all saved workflows
PATCH /api/workflow/:id — Update existing workflow
Nodes are topologically sorted before execution (respecting dependencies)
Each node's execute.js receives inputs from connected upstream nodes
Trigger module allows scheduled/event-based workflow execution
Workflows stored in workflows/ directory as JSON files

Project Structure

├── internal/workflow/
├── internal/trigger/
├── internal/runner/
├── workflows/

Implementation Details

Workflow JSON Structure

{
  "id": "daily-report",
  "name": "Daily Stock Report",
  "nodes": [
    { "id": "1", "type": "http-request", "data": { "url": "https://api.stocks.com/data" } },
    { "id": "2", "type": "data-filter", "data": { "field": "price", "operator": ">", "value": 100 } },
    { "id": "3", "type": "email-sender", "data": { "to": "[email protected]" } }
  ],
  "edges": [
    { "source": "1", "sourceHandle": "response", "target": "2", "targetHandle": "data" },
    { "source": "2", "sourceHandle": "filtered", "target": "3", "targetHandle": "body" }
  ]
}

Troubleshooting

⚠️ Circular dependencies in workflow graph cause infinite loops → engine detects and aborts
⚠️ Node execution timeout defaults to 30 seconds per node
⚠️ Missing node package → workflow execution fails with 'node not found'
⚠️ Edge connections must match output/input handle IDs exactly

Summary

This article covers workflow patterns for Flowork OS. Generated by Flowork AI from verified system architecture.

INDEXED KEYWORDS: