Skip to main content
stagewise includes a full Chromium-based browser engine — the same rendering that powers Chrome and Edge. It’s integrated into the IDE alongside your agent chat, not a separate window.

​
The browser as agent context

The browser is a live viewport the agent can see and interact with. This enables workflows impossible with file-only agents:
  • Preview your app — Run your dev server and see the result in stagewise
  • Select elements visually — Click on elements to attach them as agent context
  • Cross-reference — Browse docs, designs, or APIs in one tab while the agent works in another
  • Debug visually — The agent takes screenshots and inspects what users see

​
DOM context selection

The DOM context selector lets you point at anything in the browser instead of describing it. When you select an element, stagewise stores a debug file and attaches it to your chat message. That debug file gives the agent the details it needs to pinpoint the change:
  • The selected element and nearby DOM structure
  • Element attributes, text, and selector data
  • Relevant CSS and layout information
  • Position and viewport context
  • Page URL and browser state
Select multiple elements before sending. stagewise attaches one debug file per selection, so the agent can compare the exact elements you referenced.

​
Tabs and the agent

Browser tabs are first-class agent context:
  • @mention tabs in chat messages to attach them as context
  • Open tabs from the agent — the agent can open URLs you reference
  • Navigate programmatically — the agent browses to pages it needs to inspect

​
Browser settings

The built-in browser supports standard features:
  • Configurable search engines for the address bar
  • Browsing history with search
  • Site-specific permissions (camera, location, notifications)
  • Granular data clearing
Configure these in Settings → Browsing. See the Browsing reference for detailed guides.

​
Not just for web development

While the browser is essential for web development, stagewise works with any project type. The agent reads, writes, and modifies any file type. The browser adds visual feedback when your project has a web interface — it’s not required for the agent to function.

​
What’s next

DOM context selector

Step-by-step guide to selecting elements for agent context.

Browsing settings

Configure search engines, permissions, and history.