Run an AI-powered browser agent.
This endpoint launches a browser session where an AI agent will execute your directive. You can watch the agent work in real-time via the history_url or poll our API for results.
The agent can navigate websites, interact with page elements like a human, and complete multi-step workflows.
Credits are consumed based on the number of steps taken and the VLM model used.
API key for authentication. Get your API key at https://app.altrina.com/settings
The instruction for the browser agent to execute
Starting URL for the browser session
Chrome DevTools Protocol URL if you want to run Altrina on your own browser. When provided, the API will connect to your existing browser instead of provisioning a new one
Live view URL for your browser session (only used when providing your own cdp_url)
VLM model to use for action selection. Options: claude-sonnet-4-20250514 (5 credits/step), gemini/gemini-2.5-flash (1 credit/step), o3 (3 credits/step), claude-opus-4-1-20250805 (25 credits/step), gpt-4o (5 credits/step)
Enable advanced shadow DOM functionality with closed shadow root support. Feature is in beta. Exercise caution in critical applications.
Browser configuration options. NOTE: These options are only applied when the API provisions the browser (no user-provided CDP URL). If you are connecting to your own browser, provide cdp_url (required) and optionally live_url if you want to view it in the Altrina platform.
Browser agent job started successfully
Response payload when starting a browser agent job.
Unique identifier for this job
Initial status of the job (running)
running, completed, failed, user_taken_over, paused Link to the Altrina platform where you can view live session, action history, detailed logs, and take over control
API endpoint to poll for job status updates
Direct link to watch the live browser session. You can observe the agent working in real-time
Chrome DevTools Protocol URL