> ## 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
> Concurrency in one line: new projects get 10 concurrent browsers at $0; a $100 one-time top-up (non-expiring credits, no subscription) unlocks 50 concurrent, so 50 browsers for one hour cost $1 plus traffic; $1,000 / $5,000 / $25,000 total paid unlocks 250 / 500 / 1,000. Details below and at https://docs.browser-use.com/cloud/guides/concurrency.md.
> 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; V3 trades accuracy for speed and lower cost than V4; V2 is legacy and only for existing integrations. 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. For latency-sensitive tasks set model to bu-ultrafast (bu-fast for more reasoning per step); these presets reject modelParams. See https://docs.browser-use.com/cloud/agent/ultrafast.md.
> 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 V4 agent browser stays open about 2 minutes after its session's last run activity, then stops automatically; a later follow-up gets a fresh browser (same settings and profile, no tabs). Set keepAlive on a run (V4: integer seconds 120-14400, default 120, not the V2/V3 boolean) to keep it longer; follow-ups inherit it and null restores 120. Free-tier and ZDR runs always use 120. Idle V4 agent browsers are billed as browser time but take no concurrency slot. A follow-up that reuses the open browser emits browser.reattached instead of browser.ready. Closing CDP does not stop a 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 1.0, result.is_done() means the run completed (status completed); result.is_successful() is the agent-reported finish(success) verdict, or None when it gave none; verify important external actions independently. Cloud timeout, API client timeout, model timeout, and task completion are separate concepts.
> Pages under /open-source/legacy/0.x/ document browser-use 0.x (pip install 'browser-use<1'). For new open-source code, use the 1.0 pages under /open-source/ outside legacy/.
> 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 V4 agent browser stays open about 2 minutes (or keepAlive seconds) 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.

# Telemetry

> What anonymized usage data browser-use sends, what it never sends, where it goes, and how to turn it off.

browser-use sends anonymized usage telemetry: model, steps, cost, timing, helper names, the URLs visited (without query strings), error messages and the final result. Never the task text, typed text, code or secrets. `ANONYMIZED_TELEMETRY=false` turns it off, for the library, the CLI and the MCP server:

```bash theme={"theme":"css-variables"}
export ANONYMIZED_TELEMETRY=false
```

Any value that does not start with `t`, `y` or `1` turns it off. Off means off: no event, no background thread, no file written.

The events go to PostHog so we can see which models, browsers and helpers work, where runs spend their time and why they fail. There are no feature flags or remote settings: telemetry only sends, it never changes what the library does.

| Event | When |
| - | - |
| `agent_event` | One per `Agent.run()`. |
| `cli_event` | One per `browser-use` CLI call, scrubbed (below). |
| `mcp_server_event` | One per MCP tool call (`browser-use --cli-mcp` and `--mcp`). |

## Never sent

* The task text, follow-up tasks and `extend_system_message`. Where a URL, error message or result quotes the whole task, it is sent as `<task>`.
* Text typed into pages.
* The model's messages and thinking, and partial results.
* Code the model writes, and tracebacks (their frames quote the code). Repeated code is detected by a hash that never leaves the process.
* `sensitive_data` values. Where one appears in a URL, error message or result, it is sent as `<secret:name>`.
* API keys and credentials: `sk-...`, `sk-ant-...`, `bu_...`, `AIza...` and similar keys, JWTs, `Bearer` / `Basic` tokens and `password=...`, `token=...`, `api_key=...` values are sent as `<credential>`. Email addresses, anywhere, are sent as `<email>`.
* Query strings and fragments, and a URL's user, password and port. Hostnames that alone grant access (tunnels such as `*.ngrok-free.app` and `*.trycloudflare.com`, cloud IDE ports, Lambda function URLs, preview deploys) are sent as the provider's pattern, not the name.
* Page content: titles, HTML, screenshots, file names, `js()` expressions, `cdp()` parameters. The final result is sent, so text the agent extracted into it is too.
* CDP endpoints. Only their kind (`browser-use-cloud`, `localhost`, `private`, `public-other`).
* Paths and endpoints: never your `base_url`, `user_data_dir` or executable path.
* Uncaught exceptions from your process. There is no exception autocapture.

## What is sent

`agent_event` holds:

* **Install and environment:** browser-use, Python, Pydantic AI and browser-harness versions; OS; whether it runs in Docker, CI or a notebook.
* **Configuration:** model id and provider, reasoning, mode, `max_steps`, timeout, whether structured output, custom tools (their names), secrets (a count) or a custom workspace are used, and the task's length.
* **Browser:** the mode (cloud, chromium, chrome, cdp), keep-alive, cloud proxy country and timeout, domain restrictions (counts, never the domains) and start time.
* **Outcome:** status, success, how the result was delivered, the exception class and the phase it happened in, the error messages and the final result (below).
* **Time:** where the run spent it: browser start, model requests, cells, helpers, waits, compaction, shutdown.
* **Model:** steps, requests, tokens, cost, calls per tool, retries by cause.
* **REPL:** cells, cell errors by class, helper and CDP method counts and timings, imported module names, screenshots.
* **Browsing:** the URLs visited (below), hostnames the agent navigated to, navigation and download counts, Chrome net error codes, CAPTCHA solver events, domain-restriction blocks.
* **Per step:** for the first 50 and last 150 steps, timings, token counts, tool names, the helpers used and the hostname the step ended on.

The text fields, each scrubbed as in [Never sent](#never-sent) before it is cut:

| Field | What | Limit |
| - | - | - |
| `urls_visited` | Pages the agent navigated to, in order, each once, as `scheme://host/path` (http, https, ws, wss). | 50 URLs, 200 characters each |
| `error_messages` | The run's error (`result.error`), then each failed cell's exception line (`ValueError: message`), each once. | 20 messages, 300 characters each |
| `final_result` | The result text, or a structured result as JSON. | 500 characters |

The other fields are numbers, bools, enums, hostnames, or names chosen by browser-use, the developer, Python or an installed package (model id, helper, tool, module, exception class, CDP method).

A run whose process was killed (out of memory, `SIGKILL`, a notebook kernel restart) is reported the next time browser-use runs on that machine, with status `crashed` and only its configuration, step count and time.

`cli_event` keeps the command, exit code, duration, helper names and counts, browser kind, client, and `error_message`: the exception line of the error, scrubbed, at most 300 characters. browser-harness's own event would include the piped code, its output and the full error text; the `browser-use` CLI removes those before sending. `mcp_server_event` keeps the tool name, duration, whether it failed, the exception class and `error_message`: the exception's message, scrubbed, at most 300 characters. Neither ever includes the code, its output or arguments.

## How it is sent

* A run never waits on the network. Recording is an in-memory counter; a background thread builds the event, gzips it and posts it to `https://eu.i.posthog.com/batch/` with Python's standard library.
* At process exit, events still queued are handed to a short-lived background process, so exit does not wait either.
* Files, all under `~/.config/browseruse` (`BROWSER_USE_CONFIG_DIR` moves it), and only while telemetry is on: `device_id`; `telemetry/inflight/` for a run in progress (deleted when its event is sent); `telemetry/outbox/` for events not sent yet (at most 20 files and 512 KB, deleted after 7 days).
* The `browser-use` CLI sends `cli_event` to the same place under the same id, and creates no browser-harness install id.

## Identity

One anonymous id per machine: `BROWSER_USE_DEVICE_ID` if set, else a random id stored in `~/.config/browseruse/device_id`. If that file can't be written, a hash of the machine's MAC address and hostname. PostHog derives a location from the request IP.

`BROWSER_USE_TELEMETRY_INTERNAL=1` marks events as internal (`is_internal`). We set it in our own tests and evals, so they don't count as usage.


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