Playwright MCP Browser Automation Server

Give an MCP-compatible AI agent a live Playwright browser with isolated sessions, domain controls, predictable limits, and ten focused tools.

Connect in one minute

Use the Actor Standby endpoint and include every domain the browser may navigate to:

https://produkdigitalali--playwright-mcp-browser-server.apify.actor/mcp?allowedDomains=example.com

Cursor and other remote MCP clients can use:

{
  "mcpServers": {
    "secure-playwright": {
      "url": "https://produkdigitalali--playwright-mcp-browser-server.apify.actor/mcp?allowedDomains=example.com"
    }
  }
}

For Claude Desktop, add the same URL as a custom remote MCP connector. In n8n, use it as the Streamable HTTP URL in an MCP Client Tool node. Never paste an Apify token into a prompt or issue.

Three useful examples

1. Extract a public table

   • URL: https://example.com
   • Allowlist: example.com
   • Prompt: "Open the page, extract its visible text and links, then tell me the current URL."

2. Fill a public form

   • Allowlist every expected form and redirect domain.
   • Prompt: "Open the contact form, fill the name and message fields, take a screenshot, and stop before submitting."

3. Capture evidence

   • Prompt: "Open this public page, wait for the main heading, extract the text, and save a full-page screenshot."

Ten tools

Tool	Purpose
navigate	Open an allowlisted public HTTP(S) URL and check redirects.
click	Click the first matching element.
fill	Replace a form value without logging that value.
press	Press a keyboard key on an element or page.
extract_text	Return bounded visible text.
extract_links	Return up to 200 links.
screenshot	Save a PNG in Apify key-value storage and return its key.
get_url	Return the current page URL.
wait	Wait for a visible element or up to 10 seconds.
close	Close the browser context and release session resources.

Version 1 deliberately excludes arbitrary JavaScript execution, cookies, downloads, raw HTML, PDF generation, and network capture.

Session controls

The endpoint accepts these query parameters:

Parameter	Default	Accepted
allowedDomains	required	Comma-separated exact domains; subdomains are included.
maxToolCalls	50	1-200
idleTimeoutMs	90000	30000-300000
hardTimeoutMs	900000	60000-1800000

Each MCP session gets its own Playwright browser context. The server blocks localhost, private and link-local networks, reserved and multicast addresses, URL credentials, non-HTTP(S) schemes, and ports other than 80/443. Navigation redirects are checked against the same domain allowlist. External public subresources may load, while any private-network subresource is blocked.

Result shape

Successful tools return compact JSON in both text and structured MCP content:

{
  "success": true,
  "tool": "extract_text",
  "currentUrl": "https://example.com/",
  "durationMs": 24,
  "data": {
    "text": "Example Domain ...",
    "truncated": false
  }
}

Screenshots return a storage reference:

{
  "storageKey": "screenshot_SESSION_HASH_TIMESTAMP_ID.png",
  "contentType": "image/png",
  "bytes": 18432
}

Pricing

The launch price is $0.012 per successful tool call ($12 per 1,000 successful calls). Failed calls are not charged as successful calls.

• 10 successful calls: $0.12
• 50 successful calls: $0.60
• 100 successful calls: $1.20

Apify Standby users also cover platform usage for the running container. Keep sessions short and call close when finished. Pricing will be reviewed only after enough external usage exists to measure typical session cost.

When to use this Actor

• Give Claude, Cursor, n8n, or another MCP agent an interactive browser.
• Restrict an agent to one public domain or a small approved domain set.
• Extract text and links one decision at a time.
• Fill forms with a human review before submission.
• Capture screenshots as referenced evidence.
• Run browser work where a fixed batch script cannot decide the next action in advance.

Choosing the right browser Actor

Tool	Choose it when
Puppeteer Browser Automation Runner	You want a simple deterministic Chrome workflow.
Selenium Browser Automation Runner	You use WebDriver locators or Selenium knowledge.
Chrome DevTools Automation Runner	You need low-level CDP commands or rendering diagnostics.
Playwright Browser Automation Runner	You want a deterministic batch workflow with Playwright actions.
Browser Network & HAR Capture Inspector	You need network, API, failed-request, or HAR diagnostics.
Browser Use AI Task Runner	You want to submit a natural-language goal as a batch task.
Playwright MCP Browser Automation Server	You need a live browser controlled one tool call at a time by an MCP agent.

Limits and responsible use

Use this Actor only on sites you are authorized to access. It does not bypass CAPTCHAs, access controls, robots restrictions, or rate limits. A target may return 403/429 or block cloud browsers. The Actor does not expose or persist cookies, API keys, authorization headers, or proxy credentials.

Support

Include the approximate time, tool name, target hostname, MCP client, and sanitized error in an issue. Never post API tokens, passwords, cookies, authorization headers, form values, or private URLs.