Browser Sessions
Browser sessions give agents (and operators) a server-side Chromium browser for web automation: navigate, click, fill forms, run JavaScript, take screenshots, extract page content, and render PDFs.
Open Browser Sessions in the sidebar to create a session, drive it, and watch screenshots of what it sees.
Capabilities
Per session:
navigate— load a URL (30-second timeout).screenshot— PNG, optionally full-page.- Actions —
click,fill,evaluate(JavaScript),get_content(page HTML),pdf.
Sessions are held in memory on the API process: they are working tools, not durable records. Anything worth keeping (screenshots, extracted content) should be saved to documents or task attachments.
Requirements
Server-side browsing uses Playwright with Chromium, which is an optional dependency. If it isn't installed the API returns a clear error; install with:
pip install playwright && playwright install chromium
(In Docker deployments, add this to your API image if you use the feature.)
API Summary
| Endpoint | Purpose |
|---|---|
POST /api/v1/browser/sessions | Create ({"headless": true}) |
GET /api/v1/browser/sessions, GET .../{id} | List / inspect |
POST /api/v1/browser/sessions/{id}/navigate | Go to a URL |
POST /api/v1/browser/sessions/{id}/screenshot | Capture |
POST /api/v1/browser/sessions/{id}/action | click / fill / evaluate / get_content / pdf |
DELETE /api/v1/browser/sessions/{id} | Close |
Treat browser automation as a sensitive capability: it can reach anything your server can reach. Scope it to agents that need it and pair outward actions with HITL governance.