A workflow is a visual sequence of connected nodes that perform browser-based tasks automatically. In Agentic WorkFlow, you build workflows right inside your browser extension.
Click the Agentic WorkFlow icon in your browser to open the extension.
Choose Create New Workflow or Start from Blank .
A blank workflow canvas is displayed.
Click Add first step… to insert your first node (e.g. “Get Selected Text”).
Configure the node by opening its settings panel (double-click it).
Continue adding nodes, configuring, and connecting them.
To test, make sure you’re on a relevant web page, then click Execute Workflow .
The output will appear in each node’s results panel.
You can automate tasks such as:
Extracting page content
Interacting with forms or elements
Processing data with AI
Branching logic or loops
Combining browser and external data
If you’re new, try the Quick Intro Tutorial first to experiment with basic browser nodes.
Good for testing or workflows without triggers:
Open the page you want to automate.
Open the extension and choose your workflow.
Click Execute Workflow .
Nodes will run in order using the current page state.
When workflows start with a Trigger node , they can run automatically when the trigger condition is met. (This is similar to how workflow triggering works in n8n, where you must activate workflows to run automatically.) ([n8n Docs][1])
Until then, manual execution gives you full control and helps with debugging.
Workflows run in the page you currently view, so they see real-time content and page structure.
Because they run inside the browser, workflows are subject to browser security rules (e.g. cross-origin constraints).
Very large workflows or heavy processing may hit browser performance limits (memory, CPU).
Workflows are saved locally in your extension — they persist across browser sessions. You can export/import workflows to share or backup them.
Add nodes gradually; test frequently instead of building many nodes at once.
Use If / conditional nodes to manage different paths.
In each node, inspect execution results (JSON or table view) to understand data shapes.
Use meaningful names or notes for nodes — this helps you and others read the flow.
Start with templates from the Marketplace to learn patterns and best practices.
Quick Intro Tutorial
A simple step‑by‑step example using browser extraction and basic processing.
Go →
Learning Path
See a full roadmap from beginner to advanced workflows.
Go →
Browser Nodes Reference
Discover all supported browser interaction nodes (click, scroll, forms, etc.).
Go →
AI & Local Models
Learn how to incorporate LLMs for text processing, summarization, reasoning, and more.
Go →