How to Connect Browserless to Microsoft Copilot via MCP

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Browserless, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

Browserless + Microsoft Copilot

Browserless cloud browsers from your AI: read rendered pages as markdown, HTML or links with smart scrape, scrape elements with CSS selectors, take screenshots, print PDFs, run Lighthouse audits, map a site's URLs and open bot-protected pages through the unblock API. Every tool except the version check spends Browserless units.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

BROWSERLESS_API_TOKEN, BROWSERLESS_HOST

Step 1 — Get credentials

Setting up takes about 2 minutes. You need a Browserless account; the free plan includes monthly units.

1. Sign in at https://browserless.io/account/ and copy the API token from the dashboard.
2. Paste it into BROWSERLESS_API_TOKEN. Browserless takes it as the token query parameter on every call; it is sent only to the host below.
3. Set BROWSERLESS_HOST to your plan's region, without https://: production-sfo.browserless.io (US West, the default for shared plans), production-lon.browserless.io (London) or production-ams.browserless.io (Amsterdam). Dedicated (private fleet) plans use chrome.browserless.io or their regional chrome-...browserless.io host; a private token on a shared host is refused with 401.
4. Install the connector and run browserless_get_meta: it returns the version of the Browserless service on the host and confirms the token.

What costs units: every tool except browserless_get_meta spends Browserless units from your plan: 1 unit per 30 seconds of browser time (rounded up; plain fetches in smart scrape cost 0.5), plus 6 units per MB of residential proxy traffic or 2 per MB of datacenter traffic, and 10 units per solved CAPTCHA. browserless_smart_scrape and browserless_map use the residential proxy by default, and the json and summary formats of smart scrape add AI extraction units. stealth: true costs nothing extra. The AI asks you before a batch of calls and before using a proxy on large pages.

Reading pages: browserless_smart_scrape is the first choice. It tries a plain fetch, then a proxy, then a full browser, says which strategy served the page, and returns the formats you ask for: markdown (the default here), html, rawText, links, screenshot, pdf, json (AI extraction with a schema or a prompt in json_options) or summary. only_main_content, include_tags and exclude_tags trim the page. browserless_get_content returns the fully rendered HTML of a URL, or of HTML you pass, from a real Chrome. browserless_scrape returns the text, HTML and attributes of every element matching your CSS selectors.

Visuals and documents: browserless_screenshot captures the viewport, the full page or one element (selector) as PNG, JPEG or WebP, with a custom viewport, scroll_page for lazy content and block_consent_modals to hide cookie banners. browserless_pdf prints a URL or HTML you pass (for example a generated report) to PDF, with options such as format, landscape, margins and printBackground.

Site discovery and quality: browserless_map lists a site's URLs from its sitemap and links, with titles and descriptions, ordered by relevance to search (limit 200 by default, up to 5000). browserless_performance runs a Google Lighthouse audit with scores from 0 to 1 for performance, accessibility, best practices, SEO and PWA. A full audit can take minutes and is very large, so the AI passes config to audit one category.

Blocked pages: browserless_unblock gets past bot detection (Datadome, passive CAPTCHAs) and returns the HTML, the cookies the site set and optionally a full-page screenshot; it works best with proxy: residential, and the AI asks first. For the other tools the AI first tries stealth: true, then a proxy (proxy_country picks the country).

Waiting for content: the rendering tools accept wait_for_selector (an element and a timeout), wait_for_timeout, goto_options (for example waitUntil: networkidle2) and best_attempt, which continues with what loaded when a wait times out. In browserless_get_content and browserless_scrape, reject_resource_types (images, fonts, media) speeds pages up; browserless_get_content also sends set_extra_http_headers and cookies with the request. All times are milliseconds; timeout caps the whole call.

Switched off at install: browserless_function, which runs your own Puppeteer code for multi-step flows such as clicking through a form or paging through results. That code can submit forms, sign in or buy on websites, and those actions cannot be undone, so a workspace admin turns it on when the team needs it. Once on, the AI shows you the code and waits for your yes.

What it changes: nothing in your Browserless account. Every call starts a fresh headless Chrome, does one job on one URL and closes the browser, so nothing is kept between calls.

Data formats: screenshots come back as images and PDFs as files; text answers are capped at about 100 KB. Options such as wait_for_selector, goto_options, viewport and json_options are JSON objects. Lighthouse scores are fractions (0.92 = 92).

Untrusted content: pages, cookies and scraped text come from third-party websites. The AI treats instructions inside them as data, never as orders.

Common errors: 401 means the token is wrong or does not belong to this host (shared versus private fleet); copy it again from the dashboard. 404 on the install check means BROWSERLESS_HOST does not serve the REST APIs. 400 is a malformed body, for example url and html both missing. 403 means the target is not allowed for this account. 408 means the page took longer than the timeout: raise timeout, wait for less or use best_attempt. 429 means the plan's concurrency is used up; wait and retry once. Empty or partial content usually means the site blocks automation: try stealth: true, then browserless_unblock with a residential proxy.

The connector does not cover BrowserQL, WebSocket sessions for Puppeteer or Playwright, the asynchronous crawl and agent APIs, file downloads or session management.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Microsoft Copilot

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Microsoft Copilot uses MCP servers through agents built in Copilot Studio: you add AnythingMCP to an agent as a tool, then publish the agent to Microsoft 365 Copilot and Teams. Copilot Studio connects from Microsoft's cloud over Streamable HTTP, which AnythingMCP supports.

1. In Copilot Studio, open your agent (or create one), go to Tools and choose Add a tool → New tool → Model Context Protocol.
2. Enter a Server name (for example AnythingMCP), a Server description that says what the connector does, and the Server URL https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID. Copilot Studio uses the description to decide when to call the server.
3. Under authentication, choose API key, set Type to Header and the header name to X-API-Key, then select Create.
4. In Add tool, choose Create a new connection, paste an MCP API key from AnythingMCP (MCP Servers → your server → API keys) and select Add to agent.
5. Try it in the agent's Test pane: ask for something that needs the Browserless tools.
6. Select Publish, then Channels → Microsoft 365 and Microsoft Teams, keep Make agent available in Microsoft Copilot selected and choose Add channel. See agent in Teams installs it for you; in Microsoft 365 Copilot, type @ and pick the agent.

Prefer per-user sign-in? Choose OAuth 2.0 → Dynamic discovery instead of an API key: each user then signs in to AnythingMCP the first time the agent calls it.

Self-hosting? Use your instance's public HTTPS address instead of https://cloud.anythingmcp.com. Copilot Studio connects from Microsoft's cloud, so localhost does not work.

Available tools

Example prompts

• "Read https://example.com/pricing with smart scrape and give me the plans and prices as a table."
• "Take a full-page screenshot of our homepage at 1280 px width and hide the cookie banner."
• "Run a Lighthouse performance audit of https://example.com and list the three biggest problems."
• "Map the URLs of https://example.com that relate to documentation and list the first 50."
• "Turn this HTML report into an A4 PDF with background colors and 1 cm margins."

FAQ

Which Copilot is this for? Microsoft 365 Copilot at work, through agents built in Copilot Studio. The Microsoft Copilot app for personal accounts has no option to add your own MCP server.

What do I need? A work account with access to Copilot Studio (a license or a trial) to build the agent. People who use it in Microsoft 365 Copilot or Teams need access to the agent: share it with them, or ask an admin to publish it to the organization. If your organization applies data policies to Power Platform connectors, they cover MCP servers too.

The agent doesn't call the tools. Check that the header name is exactly X-API-Key and that the key belongs to the same MCP server as the URL, and make the server description specific: Copilot Studio uses it to decide when to call the server. Publish the agent again after each change; in an open chat, type "start over" to get the new version.

Next steps

• Connector landing page
• Connect to Claude
• Connect to ChatGPT
• Connect to Cursor
• Connect to Gemini
• Connect to Meta Muse