Browser API for AI agents

Build an agent workflow with browsers

Some tasks need a rendered page or a browser connection rather than a simple HTTP response. Lupin provides a workspace-owned browser session registry, supported page actions and temporary browser connection links. Your application can inspect the operation and close the session when the task is complete.

What the API supports

• Create, list and end isolated browser sessions
• Navigate to public pages, read content and capture screenshots
• Use supported search and page-reading operations
• Request temporary Live View and Chrome DevTools Protocol access
• Track browser runtime against workspace allowances

From setup to a verified outcome

1. Discover availability

   Read the workspace capabilities before creating a session. Hosted browser operations require the configured browser provider and free session capacity.

2. Run the bounded task

   Create a session, use documented page actions or request temporary access for a compatible CDP client. Treat access URLs as credentials.

3. Close and inspect

   End the session and confirm its closed state. Retain only the information the task needs, following the workspace’s content and spending limits.

Availability and limits

• Cookies and sign-ins remain in the current session while it is open. Lupin does not persist browser profiles between closed sessions.
• Custom Lupin proxy routing is not currently supported for Browser Run sessions.
• Browsing is bounded by public-web policy, provider configuration, runtime limits and concurrency. A session is not an unrestricted private-network tunnel.

Check your workspace for its current configuration, usage and limits. Compare the plans before creating paid resources.

Start with the API reference

The following read endpoint is part of this capability. Authenticate with the relevant workspace key and inspect the documented response and permission requirements.

OpenAPI specification · Browsers documentation · Agent registration

Common questions