G4 Engine – Visual Sequential Workflow for VS Code

Design, run, and record automations visually — right inside VS Code.

Overview

G4 Engine turns Visual Studio Code into a full automation and orchestration IDE.

G4 is a plugin-based automation engine that runs controlled workflows across web browsers, desktop applications, APIs, and internal tools — and those workflows can be driven by people, CI/CD schedules, or AI agents. This extension brings the Sequential Workflow Designer straight into your editor: you build automations by dragging and dropping building blocks onto a canvas, connect them, and run them — no code required. When you do want code, it's right there too, next to Git, terminals, and your favorite AI assistants.

What you can automate: fill in and submit web forms, drive line-of-business desktop apps, call APIs, move data between systems — or record a person clicking through a task once and replay it on demand.

Get Started

Option 1 — Deploy the G4 sandbox (recommended)

One command installs everything you need — the engine, the tools, browser drivers, and a preconfigured copy of VS Code:

Windows (PowerShell):

irm https://raw.githubusercontent.com/g4-api/g4-sandbox/main/install-g4-sandbox.ps1 | iex

Linux:

curl -fsSL https://raw.githubusercontent.com/g4-api/g4-sandbox/main/install-g4-sandbox.sh | bash

Option 2 — Install the extension only

Already have VS Code and a G4 backend? Install G4 Engine Client from the VS Code Marketplace, or from a packaged .vsix:

code --install-extension g4-engine-client.vsix

👉 New to G4? The Quick Start walks you from zero to your first running automation, click by click.

Documentation

Two guided learning paths take you from zero to productive — each a self-contained folder of numbered modules:

• 📗 Quick Start — deploy the sandbox, install the extension, and build and record your first automation, click by click.
• 📘 Configure manifest.json — every runtime setting explained: what it does, when to change it, and copy-paste recipes.

Key Capabilities

• Visual Sequential Workflow Designer — full drag-and-drop workflow editing on a live canvas inside VS Code, with real-time rendering. No external browser or standalone app.
• Workspace-native drag & drop — drag automation files straight from the VS Code Explorer onto the canvas; workflows stay tightly coupled to your project files.
• Record and replay — capture a user clicking through a desktop or browser session and turn it into a workflow you can run again.
• Native VS Code integration — Command Palette commands, a dedicated Activity Bar tools panel, and a status-bar indicator for the live G4 connection.
• Extensible & AI-ready — works alongside Copilot and other AI extensions, and turns MCP servers' tools into G4 plugins automatically.

Why VS Code

By running inside VS Code, G4 Engine inherits an ecosystem you already trust:

• AI assistance — Copilot, Chat, and custom AI extensions
• Git and source control
• CI/CD tooling, terminals, and debuggers
• Familiar shortcuts, theming, and layouts

G4 doesn't replace your editor — it turns it into an automation IDE.

How It Works

The workflow canvas is powered by the Sequential Workflow Designer and embedded in VS Code using native webview APIs. The extension talks to the G4 engine over HTTP:

• Tree providers surface G4 assets from your workspace.
• Webview views host the workflow editor and recorder UI.
• Commands manage project creation, workflow loading, and execution.
• Status-bar integration reflects the live engine connection.

The result is performance, stability, and a first-class VS Code experience.

Project Structure

Running G4: Create a New G4 Project scaffolds the layout below and opens the src/ folder as your workspace:

<project-root>/
├── build/                              # build outputs
├── docs/
│   ├── examples/
│   └── G4 Manifest Configuration Guide.md
├── scripts/                            # custom scripts
└── src/                                # opened as the VS Code workspace
    ├── .agents/
    ├── .claude/
    ├── .github/
    ├── .vscode/
    │   └── mcp.json                    # MCP endpoint for Copilot / AI tools
    ├── base.bots/
    │   ├── chrome-automation-base.json
    │   └── uia-automation-base.json
    ├── base.templates/
    │   └── template-base.json
    ├── bots/                           # your automations
    ├── environments/
    ├── resources/
    ├── templates/
    └── manifest.json                   # project runtime configuration

📘 Configuration: manifest.json is the project's runtime configuration — connectivity, drivers, run behavior, recorders, and more. See the Configure manifest.json learning path for every setting.

Requirements

• Visual Studio Code ^1.105.0
• A compatible G4 Engine backend — local (via the sandbox) or remote
• Node.js — only for extension development/build