const res = await agent.browser({
  task: 'Go to the pricing page and extract every plan name and monthly price.', // min 10 chars
  start_url: 'https://example.com',
  output_schema: {                       // optional — force structured extraction
    type: 'object',
    properties: { plans: { type: 'array', items: { type: 'object' } } },
  },
  allowed_domains: ['example.com'],      // optional — restrict navigation
  max_steps: 40,                         // optional, default 50, max 100
  maxCost: 1.50,
  memo: 'scrape pricing',
});
// res.{ output (string | object), steps: [{ number, goal, url }], session_id, output_files? }

Options (BrowserTaskOptions):

Field	Notes
`task`	Required, min 10 chars. Natural-language instruction.
`start_url`	Initial URL.
`output_schema`	JSON schema for structured output extraction.
`allowed_domains`	Restrict browsing to these domains.
`session_id`	Reuse an existing browser session.
`profile_id`	Reuse a persistent profile (cookies/localStorage) — see below.
`secrets`	Domain-scoped creds for auto-login, e.g. `{ "github.com": "user:token" }`.
`max_steps`	Default 50, max 100.

Persistent profiles (logged-in sessions)

Create a profile once, then pass its id as profile_id to reuse cookies/localStorage across tasks — e.g. stay logged in to a dashboard.

const profile = await agent.createBrowserProfile('acme-dashboard'); // { id, name } — free
const profiles = await agent.listBrowserProfiles();                 // BrowserProfile[] — free
await agent.deleteBrowserProfile(profile.id);                       // free

await agent.browser({ task: 'Open the dashboard and download this month\'s invoice.', profile_id: profile.id });

Profile create/list/delete are free; running a task is paid (priced by estimated steps). Always set maxCost.

Pricing

Browser tasks are paid and scale with steps (guard with maxCost); profile create/list/delete are free. See current per-tool pricing at https://docs.oneshotagent.com/pricing.

oneshot-browser

OneShot — Browser Automation

Run a browser task — agent.browser(options)

Run a browser task — `agent.browser(options)`