MCP Connector Privacy Policy
Last updated: November 5, 2026
Scope
This page explains what data the Browserless MCP connector collects when you use it from ChatGPT, Claude or another AI assistant, and what we do with it. The connector is the read-only service at https://mcp.browserless.io/mcp/connector. It offers five tools: browserless_read_page, browserless_search, browserless_map, browserless_export and browserless_performance.
This page adds to the Browserless Privacy Policy, which covers your Browserless account, billing and our website. Where the two differ for connector use, this page applies.
Data we collect
We collect only what each tool needs to do its job: the inputs your AI assistant sends, the results we return, and basic account and request data.
Tool inputs and results
| Tool | What it receives | What it returns |
|---|---|---|
browserless_read_page | A web address, a format (text, HTML, links or elements) and an optional page selector | The page's visible text, HTML, links or elements. A fresh browser is opened for each call and closed when it ends. |
browserless_search | A search query and optional filters (language, country, location, time range, source, category) | Result titles, addresses and short snippets |
browserless_map | A website address, an optional search phrase and a result limit (up to 500) | A list of page addresses found on that site |
browserless_export | A web address and loading options | The page's source as text. For files such as PDFs or images, a short note instead of the file. |
browserless_performance | A web address, audit categories and device type | Lighthouse scores and speed metrics. The full report is not stored. |
The pages you ask about may contain personal information published by others. We process it only to return the result to your assistant.
Account data. When you connect with OAuth, we receive your Browserless account identity and use it to look up your Browserless API key on our servers. The key is never included in tool results.
Usage and technical data. For each tool call, our service metrics record the tool name, whether it succeeded, how long it took and the type of any error, without your account identity. When the connector refuses a request, for example one that reuses another account's session, we also log the IP address and user agent it came from, to detect abuse.
How we use it. Only to run the tools you call, check your plan's limits and count usage, and keep the service reliable and secure. The connector doesn't collect usage analytics. We do not sell connector data, use it to train AI models, or build profiles from it.
Who receives it
We share connector data only with the parties below, and only for the purposes above.
| Recipient | What they receive | Why |
|---|---|---|
| The websites you ask about | A normal browser request from Browserless servers | To load the page |
| Third-party web search engines | Your search query and filters, sent from Browserless servers | To return search results |
| Our cloud hosting providers | All connector traffic and the data described above | To run the service |
| Our sign-in and account database provider | Your sign-in and your Browserless account details | To sign you in and look up your account |
| Our logging and monitoring provider | Service logs, including the IP address and user agent of refused requests | To keep the service reliable and secure |
| Your AI assistant (OpenAI or Anthropic) | The tool results you asked for | To answer you |
A current list of our subprocessors is available on request from security@browserless.io. We may also disclose data when the law requires it, as described in the main privacy policy.
How long we keep it
The connector does not store page content or tool results after returning them. Everything else is kept for the periods below.
| Data | Kept for |
|---|---|
| Page content, search results, exports and audit results | Not stored |
| Browser used for a call | Closed when the call ends |
| OAuth authorization codes | 5 minutes, single use |
| OAuth access tokens | 1 hour |
| OAuth refresh tokens | Single use, until the connection ends |
| OAuth connection (grant) | Until you log out or delete your account |
| Request and error logs | Up to 30 days |
| Backups | 7 days |
| Account and billing records | See the main privacy policy |
How we protect it
Connector traffic is encrypted over HTTPS. Datastores that hold customer data are encrypted at rest, and backups are encrypted. Our SOC 2 Type II report is available on request from support@browserless.io. For more on our security practices, see the Trust Center.
Your choices
- Disconnect: remove the Browserless connector in ChatGPT or Claude settings. It stops working immediately. If you connected with an API key instead of signing in, you can also rotate that key in your Browserless account.
- Access or delete your data: email support@browserless.io. We respond within 30 days.
- Delete your account: follow the steps in the main privacy policy. Connector data linked to your account is deleted with it.
Data we do not collect
The connector only reads public web pages. It does not sign in to websites, fill in forms or click buttons.
- No tool asks for passwords, API keys, one-time or MFA codes, payment card numbers, government IDs or health information. Please don't include them in your requests.
- Tool results never include your Browserless API key. A page you ask the connector to read is returned as it appears, so anything visible on that page, including any passwords or tokens it shows, can appear in the result.
- The connector does not read your ChatGPT or Claude conversation history or memory. It sees only the inputs your assistant sends for each tool call.
- We do not use connector data to track you across services or to profile you.
ChatGPT, Claude and other assistants
Once a tool result reaches your AI assistant, the assistant provider's privacy policy governs it, not this one. See the OpenAI Privacy Policy and the Anthropic Privacy Policy for how they handle your conversations.
Legal
- Your rights: depending on where you live, laws such as the GDPR, UK GDPR, CCPA and LGPD let you access, correct, delete or export your data, or object to its use. Email support@browserless.io and we respond within 30 days.
- International transfers: we process connector data in the United States, which may be outside your country. On self-service plans, our Terms of Service and this policy govern that processing. A Data Processing Addendum that includes the EU Standard Contractual Clauses is available on Enterprise plans.
- Children: you must be at least the age of majority where you live to use Browserless, as our Terms of Service require. The connector is not directed to children. If we learn we hold a child's data, we delete it.
Changes and contact
We will update this page when the connector's tools or data practices change, and change the date at the top. Questions or requests: support@browserless.io.