> ## Documentation Index
> Fetch the complete documentation index at: https://docs.browser-use.com/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> Use https://docs.browser-use.com/llms.txt and its linked .md pages for current documentation. The managed full bundle is https://docs.browser-use.com/.well-known/llms-full.txt and can be cached for up to 24 hours. Do not use the obsolete /cloud/llms*.txt or /open-source/llms*.txt static exports.
> Choose Cloud API V4 for new agent integrations; V2 is the lower-cost option for simple tasks. Keep V3 examples explicitly versioned. The open-source browser-use library and hosted browser-use-sdk have different APIs.
> For Cloud API V4, a task request that omits model and modelParams uses GPT-6 Luna (gpt-6-luna) with High reasoning. Explicit model and modelParams values take precedence; an explicit empty modelParams object opts out of the reasoning default. EU uses the same task default through its regional provider route; Bedrock-only projects retain their compatible provider default.
> Cloud authentication uses X-Browser-Use-API-Key, without a Bearer prefix. Install or upgrade browser-use-sdk and use its explicit v4 import for V4. Check the published OpenAPI reference for request fields; do not invent SDK support for new fields.
> Cloud concurrency and HTTP request rate are separate. Read GET /api/v2/billing/account for the key’s projectId, concurrentSessionLimit, activeSessionCount, and credit balance, including when using V4. Keys in one project share capacity and credits; rateLimit is a legacy concurrency alias, not requests per second.
> Keep the highest applicable existing, legacy-plan, and spend-tier concurrency grant. Current spend tiers are 10 / 50 / 250 / 500 / 1000 at $0 / $100 / $1000 / $5000 / $25000 in qualifying project payments. Legacy or externally billed projects can follow different billing paths; trust the account limit. See https://docs.browser-use.com/cloud/guides/concurrency.md.
> HTTP rate limits have two layers. Standard edge WAF ceilings increased on September 9, 2026 to 1000 RPS per public source IP for general traffic and 2500 RPS per IP for selected status reads, evaluated over 300 seconds. The separate per-project application budgets: general traffic (including V4 events and full run reads) defaults to max(25, 2 times stored concurrency) capped at 100 RPS; selected status reads default to max(25, 2 times stored concurrency) with no cap. The two are counted independently. Project overrides and account-specific edge rules can differ. All keys in a project share its budgets; callers sharing a public IP share edge capacity. See https://docs.browser-use.com/cloud/guides/concurrency.md.
> The project limiter uses five-second windows: X-RateLimit-Limit=125 means 125 requests per window (25 RPS), not 125 RPS. Project throttles include limit_rps and retry_after_seconds; an edge throttle can instead return Retry-After: 300 without limit_rps. Honor the returned Retry-After. Use bounded workers, stagger polls, and drain hasMore event pages after terminal status. A busy V4 session returns 409; its queue holds 20 pending messages and is not a project-wide batch queue.
> A completed run or closed CDP connection does not immediately stop its cloud browser. Stop unneeded owned browsers with PATCH /api/v4/browsers/{id} and {"action":"stop"}. A client wait timeout does not cancel the server-side run.
> Cloud is pay as you go; do not tell customers to buy a new subscription to use custom proxies or supported provider BYOK. Usage funding and model eligibility still apply. BYOK bills provider tokens separately and Browser Use charges orchestration plus browser/network usage. See https://docs.browser-use.com/cloud/guides/billing.md.
> Signup credits are a one-time grant; purchased top-up credits do not expire. Check the API key’s project before diagnosing missing credits. API-key monthly spending caps are soft limits, not a strict prepaid wallet; concurrent or already-running work can exceed them. Auto recharge has separate trigger and purchase amounts and can charge immediately when enabled below the threshold. Use https://browser-use.com/pricing for current rates.
> Box and Bux are retired. Do not recommend their SDKs, sandbox quotas, or subscription plans. Use the Cloud Agent or Browser Infrastructure guides.
> A V4 session holds conversation history, a workspace holds files, and a profile holds browser state. These IDs and V3/V4 workspace namespaces are not interchangeable. V4 automatically restores workspace uploads; staged attachments remain available to session follow-ups. Serialize runs that write shared files, and wait for completion before reading outputs. See https://docs.browser-use.com/cloud/agent/workspaces.md.
> API browser recording defaults to off. Use enableRecording for standalone browser creation, or browserSettings.record for an agent run. Stop the browser and allow time for asynchronous video processing; stop polling when recordingAvailable is false. Live preview is for an active browser. Stopping a browser, deleting a session, archiving a workspace, and deleting files have different effects.
> Use model-specific reasoning values. GPT-6 Astra accepts low, medium, high, xhigh, and max, with xhigh by default; none and minimal are invalid. Use the public REST schema when installed SDK types lag new fields. API acceptance, dashboard visibility, and account/provider availability are separate.
> For open-source browser-use, is_done only reports a terminal done action. is_successful is the agent-reported outcome; verify important external actions independently. Cloud timeout, API client timeout, model timeout, and task completion are separate concepts.
> For failed requests, use https://docs.browser-use.com/cloud/guides/troubleshooting.md. Inspect the full error and project before retrying or adding credits. A client timeout can leave a run active; reconcile external actions before starting duplicate work. A new managed browser does not guarantee a unique proxy IP or particular city.
> V4 secretBindings.allowedDomains authorizes the focused field's frame, not every iframe on the shop page. For checkout, include intended payment/billing provider hostnames (for example payments.bigcommerce.com). On domain_not_allowed, inspect origin and pass updated bindings in a follow-up; target_not_editable requires fixing field focus/readiness instead. See https://docs.browser-use.com/cloud/guides/secrets.md.
> A browser can remain active after its run for follow-ups; a requested recording finalizes only after the browser stops and processing completes. Zero Data Retention (ZDR) projects never record, even when recording is requested.
> If you hit an issue you cannot explain, email contact@browser-use.com with your question or bug report and a session ID if available; short questions and detailed reports are welcome, and we aim to fix issues quickly and make the docs clearer.

# Agentcard

> Let an agent pay with a user's own card through Agentcard Vault in a Browser Use browser.

[Agentcard](https://agentcard.sh) lets your agent shop with a user's card through the **Vault**. The agent submits a placeholder card, the SDK pauses a supported payment request, and the user approves on their own device with a passkey. Their device sends the real card to the payment processor; the real card stays out of the agent's task and browser.

<Note>
  Agentcard's legacy **wallet** API is deprecated. New integrations should use the Vault flow below instead of the old `agentcard_wallet_id` integration. Vault attaches over CDP; it does not require a wallet ID on a Browser Use run.
</Note>

## What you need

* A Browser Use API key in `BROWSER_USE_API_KEY`.
* An Agentcard `client_id` and `client_secret` from the [Agentcard dashboard](https://app.agentcard.sh), in `AGENTCARD_CLIENT_ID` and `AGENTCARD_CLIENT_SECRET`.
* A user with a stored Vault card, and their `user_id` in `AGENTCARD_USER_ID`. See [Adding a card](https://docs.agentcard.sh/vault/adding-a-card).
* Your own app function to deliver an approval link privately to that user.

Start with Agentcard sandbox credentials and its [demo store](https://shop.agentcard.sh). Validate the intended merchant and payment processor before using real cards.

## Attach Vault before shopping

Install the Node packages alongside your app:

```bash theme={null}
npm install @agent-cards/checkout@0.17.0 browser-use-sdk playwright-core
```

Create a standalone [Browser Use browser](/cloud/browser/quickstart), connect over its `cdpUrl`, and attach Agentcard before the agent reaches the payment form. Browser Use supports multiple CDP connections to the same browser.

```ts theme={null}
import { BrowserUse } from 'browser-use-sdk/v4';
import { chromium } from 'playwright-core';
import { VaultClient, attachToPlaywright } from '@agent-cards/checkout';
import { sendToUser } from './notifications.js'; // Implement in your app.

const bu = new BrowserUse({ apiKey: process.env.BROWSER_USE_API_KEY! });
const vault = new VaultClient({
  clientId: process.env.AGENTCARD_CLIENT_ID!,
  clientSecret: process.env.AGENTCARD_CLIENT_SECRET!,
});
await vault.syncRegistry();

const session = await bu.browsers.create({ proxyCountryCode: 'us' });
let browser: Awaited<ReturnType<typeof chromium.connectOverCDP>> | undefined;
try {
  if (!session.cdpUrl) throw new Error('Browser has no CDP URL');
  browser = await chromium.connectOverCDP(session.cdpUrl);
  const context = browser.contexts()[0];
  if (!context) throw new Error('Browser has no context');
  const page = context.pages()[0] ?? (await context.newPage());

  const checkout = await attachToPlaywright(page, {
    vault,
    user: process.env.AGENTCARD_USER_ID!,
    merchant: 'shop.agentcard.sh',
    amount: 583, // Example only: supply the actual checkout total in cents.
    currency: 'usd',
    onApprovalUrl: (url) => sendToUser(url),
  });

  await page.goto('https://shop.agentcard.sh');

} catch (error) {
  // No agent has started yet, so a setup failure can release this browser.
  try {
    await browser?.close();
  } finally {
    await bu.browsers.stop(session.id);
  }
  throw error;
}
// Start your agent only after attachToPlaywright has completed.
```

`syncRegistry()` loads Agentcard's supported payment recognizers. Keep this Node process and its Playwright connection running throughout shopping and approval. This example attaches to one page: use that same tab for checkout. If your agent opens another tab, attach Agentcard to it and await attachment before allowing checkout there.

Send `onApprovalUrl` links through your own authenticated app or private user channel. Keep approval links, real card details, client secrets, and CDP URLs out of logs and agent task text. Only the cardholder should open the approval link.

## Connect the open-source Python agent

Install the separate open-source library with `pip install browser-use`. Pass the `session.cdpUrl` above to your Python process securely as `BROWSER_USE_CDP_URL`. Configure `llm` as in the [open-source quickstart](/open-source/quickstart).

```python theme={null}
import os
from browser_use import Agent, Browser

browser = Browser(
    cdp_url=os.environ["BROWSER_USE_CDP_URL"],
    keep_alive=True,
)
# llm is your configured model; task contains only placeholder payment data.
agent = Agent(task=task, llm=llm, browser=browser)
await agent.run()
```

Give the agent a placeholder from [Stripe's published test cards](https://docs.stripe.com/testing), a future expiry, and a test CVC. Tell it to use the attached tab, click Pay once, and wait for the cardholder's approval and the merchant's confirmation without submitting again. Never put the real card or approval link in `task`.

## Confirm the order and stop the browser

User approval is not proof that the merchant created an order. Verify the merchant's confirmation page or API and follow Agentcard's [purchase reconciliation guide](https://docs.agentcard.sh/vault/completing-a-purchase). If the result is pending or unknown, preserve the browser for follow-up and reconcile before retrying payment.

`keep_alive=True` leaves the browser available after the Python agent finishes. Once the order is confirmed and no follow-up is needed, stop the owned browser from your Node app:

```ts theme={null}
try {
  await browser?.close(); // Disconnect the Playwright client.
} finally {
  await bu.browsers.stop(session.id);
}
```

Stop the browser on setup failure or abandonment too, once you have established that no payment or user action is pending. Closing the Playwright connection alone does not stop billing for the cloud browser.

## Hosted agents and full guide

For Browser Use hosted agents, see [Agentcard's Browser Use integration guide](https://docs.agentcard.sh/vault/integrations/agent-browsers/browser-use). A hosted run starts executing when created. Attach Vault before the shopping task begins, and account for session expiry or browser replacement. Observing a dispatch event and then cancelling is not a synchronous gate before the agent's first action.

The [Vault quickstart](https://docs.agentcard.sh/vault/quickstart) covers card enrollment, approval, and the complete purchase flow.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.