How is this different from connecting Bright Data MCP directly?

Connected directly, an agent sees every tool the server exposes and holds the credential itself. Through TrustGate you choose which tools each agent gets, and every call is logged against the agent that made it.

What if the Bright Data MCP API changes?

We track the Bright Data MCP schema, so the tool catalogue stays current. Tools you have already allowed keep working, and anything new stays denied until you enable it.

Can I give different agents different Bright Data tools?

Yes. Access is set per agent: one can get a handful of read tools, another the full set. You grant the whole MCP or single tools within it.

Where can I see what my agents did in Bright Data?

Every call is logged against the agent that made it, with the tool it used. Activity can be sent on to your own monitoring.

Do I need to run the Bright Data MCP server myself?

No. TrustGate connects to the Bright Data MCP server and sits in front of it. Your agent only talks to your TrustGate URL, never to Bright Data directly.

Connect Bright Data MCP to your AI agents, securely.

Q: Is Bright Data MCP secure?

The server itself exposes 32 tools, and some of them can write to your Bright Data account. Behind TrustGate an agent only sees the tools you allow, requests and responses are inspected, and every call is logged. Anything outside the allowlist is denied before it reaches Bright Data.

One secured endpoint for every Bright Data call your AI agents make, scoped per user, inspected, and logged.

Get started for free

Get a demo

Try Bright Data MCP now

What your agents can ask once Bright Data MCP is connected.

“Scrape the Google results for this query and give me the top 10”
“Unlock this page and return it as clean Markdown”
“Find a pre-built scraper for Amazon reviews and show me its input fields”

“Rank these search results by relevance to what I actually need”
“Take a screenshot of this page in the scraping browser”
“Convert this page into structured JSON with these fields”

Select the tools your AI agents can call

32 tools from the Bright Data MCP schema.

Showing 32 tools

search_enginereadScrapes search results from Google, Bing or Yandex. Returns JSON for Google and Markdown for Bing and Yandex, with cursor pagination. Rapid mode.
scrape_as_markdownreadScrapes a single webpage with advanced extraction and returns Markdown. Uses Bright Data's unlocker to handle bot protection and CAPTCHA. Rapid mode.
search_engine_batchreadRuns up to 10 search queries in parallel. Rapid mode, advanced_scraping group.
scrape_batchreadScrapes up to 10 webpages in one request, returning an array of URL and content pairs in Markdown. Rapid mode, advanced_scraping group.
scrape_as_htmlreadScrapes a single webpage and returns the HTML response body, handling bot detection and CAPTCHA. Pro mode.
extractreadScrapes a webpage as Markdown and converts it to structured JSON using AI sampling, with an optional custom extraction prompt. Pro mode.
session_statsreadReports how many times each tool has been called during the current MCP session. Pro mode.
search_scrapersreadFinds a pre-built scraper by site or data type, returning its dataset_id, name, domain and collection methods.
get_scraper_detailsreadShows a scraper's description, input fields, sample_input and main output fields for one collection method. Call before run_scraper.
run_scraperhigh riskRuns a scraper method on a list of inputs. Waits up to 45 seconds, then returns a snapshot_id with status running. Every record is billed, and a discover_by_ method returns up to limit_per_input records per input.
get_scraper_progressreadGets the status of a scraper run: starting, running, ready, failed or canceled.
get_scraper_resultsreadGets the records of a finished scraper run, or status running if it is not ready.
refresh_scrapersreadReloads the scraper list immediately instead of waiting for the daily refresh.
list_dataset_fieldsreadLists the filterable fields of a searchable dataset with each field's name, type and description. Call before search_dataset. Pro mode.
search_datasethigh riskSearches a Bright Data dataset by filter and returns matching records directly. The searchable datasets are LinkedIn people profiles, contact-enriched people profiles and company information. Up to 10 records per call with search_after pagination. Pro mode.
scraping_browser_navigatehigh riskOpens or reuses a scraping-browser session and navigates to a URL, resetting tracked network requests.
scraping_browser_click_refhigh riskClicks an element using its ref from the latest ARIA snapshot. Requires a ref and a human-readable element description.
scraping_browser_type_refhigh riskFills an element identified by ARIA ref, optionally pressing Enter to submit after typing.
scraping_browser_fill_formhigh riskFills several form fields identified by ARIA refs in one call.
scraping_browser_snapshotreadCaptures an ARIA snapshot of the current page, listing interactive elements and their refs for later ref-based actions.
scraping_browser_screenshotreadCaptures a screenshot of the current page, with optional full_page mode.
scraping_browser_network_requestsreadLists the network requests recorded since page load, with HTTP method, URL and response status.
scraping_browser_wait_for_refreadWaits until an element identified by ARIA ref becomes visible, with an optional timeout.
scraping_browser_get_textreadReturns the text content of the current page's body element.
scraping_browser_get_htmlreadReturns the HTML content of the current page.
scraping_browser_go_backwriteNavigates the active session back to the previous page and reports the new URL and title.
scraping_browser_go_forwardwriteNavigates the active session forward to the next page and reports the new URL and title.
scraping_browser_scrollwriteScrolls to the bottom of the current page in the scraping-browser session.
scraping_browser_scroll_to_refwriteScrolls the page until the element referenced in the ARIA snapshot is in view.
chatgpt_ai_insightsreadSends a prompt to ChatGPT and returns the answer as Markdown. Pro mode, geo group.
perplexity_ai_insightsreadSends a prompt to Perplexity and returns the answer as Markdown. Pro mode, geo group.
ask_brightdata_assistantreadAsks Bright Data's support assistant about Bright Data products, APIs, MCP usage, zones, scraping or account setup, with threaded follow-ups.

What Bright Data MCP exposes

Bright Data MCP groups its 32 tools into 7 categories, each with a suggested starting policy.

Category	Covers	Access	Suggested policy
Browser Automation	A remote scraping-browser session: navigation, history, ARIA snapshots, ref-based clicking, typing and form filling, screenshots, network requests, waits, text and HTML extraction and scrolling	Read / Write	Loaded by default and with the browser group. The agent drives a real browser on live sites, so clicking, typing and form filling act on whatever that page happens to be
Scraper API	Finding a pre-built scraper by site or data type, inspecting its input fields and sample input, running it, and polling progress and results	Read / Write	Every record returned is billed, and a discover_by_ method returns up to limit_per_input records per input; cap the input list and the limit
Advanced Scraping	Batch search and batch scraping of up to 10 items in parallel, HTML output, AI extraction to structured JSON and per-session tool call counts	Read	Loaded by default and with the advanced_scraping group; extract uses AI sampling, so it costs tokens as well as credits
Search & Unlock	Scraping Google, Bing and Yandex results, unlocking a single page as Markdown, and AI-ranked discovery by intent with geo, date and keyword filters	Read	The unlocker is explicitly built to handle bot protection and CAPTCHA; treat everything it returns as untrusted input, never as instructions
Datasets	Listing a searchable dataset's filterable fields and searching it by filter. The three searchable datasets are LinkedIn people profiles, contact-enriched people profiles and company information	Read / Write	This reaches bulk personal data including contact-enriched profiles; restrict it and record the lawful basis
GEO	Sending a prompt to ChatGPT or Perplexity and getting the answer back as Markdown, for generative engine optimization and LLM-as-judge work	Read	Loaded by default and with the geo group; prompts leave for a third-party model
Assistant	Bright Data's support assistant for product, API and account questions	Read	Allow