MCP for agents New 

Let your agent take screenshots.
================================

screeenly runs an MCP server. Add it to Claude, Cursor or any other MCP client, and your agent can capture pages, create PDFs and read rendered HTML without running a browser of its own.

 [ Start for free → ](https://3.screeenly.com/app/register) [ Try the playground ](https://3.screeenly.com/playground) 

Screenshot our pricing page at 390px and attach it to the PR.

→ capture_screenshot(url, window_width: 390, full_page: true)

← shot_url · expires in 24 hours

Done. Attached pricing-390.png.

  01 · Create a token Sign up and create an API token in your dashboard. The MCP server uses the same tokens as the API.

 02 · Add the server Point your MCP client to `https://3.screeenly.com/mcp` and send the token as a bearer token.

 03 · Ask your agent Your agent calls the tool that fits your prompt. Every capture counts toward your monthly usage, like an API request. Cached results are free.

    API requests handled 21,732,869   Launched as open source July 2014 

Every request runs in real Chrome. Revenue, customers and usage are public too.

 [See the open startup numbers →](https://3.screeenly.com/open) 

  [Tools](#tools) [Setup](#setup) [Authentication](#auth) [Use cases](#use-cases) [Questions](#faq) 

Four tools
----------

The server exposes these tools to your agent. `capture_screenshot` takes many of the options of the [screenshot API](https://3.screeenly.com/features/screenshots).

 ToolWhat it does  `capture_screenshot`Takes a screenshot (JPG, PNG, WebP) or creates a PDF of a URL or of the given HTML. Returns a download URL that is valid for 24 hours by default. Identical requests return the cached result for free. Other captures count toward the monthly usage. `get_rendered_html`Loads a URL or the given HTML in a browser, runs its JavaScript and returns the resulting HTML. Counts toward the monthly usage. `get_usage`Returns the plan, the requests used this month, the monthly limit and the credit balance. `list_recent_shots`Lists the most recent screenshots, PDFs and rendered HTML requests of the account, newest first.  

Add it to your client
---------------------

The MCP server is available over HTTP at `https://3.screeenly.com/mcp`. Replace `YOUR_API_TOKEN` with a token from your dashboard.

 Claude Code 

```
claude mcp add --transport http screeenly https://3.screeenly.com/mcp \
  --header "Authorization: Bearer YOUR_API_TOKEN"
```

 Cursor · .cursor/mcp.json 

```
{
  "mcpServers": {
    "screeenly": {
      "url": "https://3.screeenly.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_TOKEN"
      }
    }
  }
}
```

Other clients work the same way: add a remote HTTP server with that URL and an `Authorization` header.

Authentication
--------------

The MCP server accepts the same API tokens as the REST API. Requests count toward the same monthly limit and follow the same rate limit. Revoke a token in your dashboard and the agent loses access. A token with only the `shot:read` ability can read the usage and list captures, but cannot create new ones.

 Tip

Give the agent its own token
----------------------------

Create a separate token for each agent or tool. You can see in the request log which captures came from MCP.

What your agent can do with it
------------------------------

Pull request evidence

Attach a screenshot of the changed page to a pull request or a ticket.

Visual checks

Capture a page at several widths and look at the result before shipping.

Reading web apps

Read a JavaScript-heavy page through `get_rendered_html` instead of an empty app shell.

Reports as PDFs

Turn a page or generated HTML into a PDF and hand back the link.

screeenly is not a web scraper

screeenly loads pages the way a regular browser does. It cannot get past bot protection such as Cloudflare challenges, CAPTCHAs or other checks that decide a visitor is not human, and it does not try to. Use it on your own pages and on pages you are allowed to load.

Questions
---------

   Does the MCP server cost extra?  No. It is part of every plan. Captures through MCP count toward your monthly usage like API requests.

   How long are files available?  24 hours, the same as for API requests. The agent gets the expiry time with every capture.

   What happens when I reach my limit?  The tools return an error that explains which limit was reached.

   Can the agent capture a single element or a whole page?  Yes. `capture_screenshot` accepts `selector` for one element and `full_page` for the whole scrollable page.

   More from screeenly [     Screenshots ](https://3.screeenly.com/features/screenshots) [     PDFs ](https://3.screeenly.com/features/pdf) [     Your own HTML ](https://3.screeenly.com/features/your-own-html) [     Rendered HTML ](https://3.screeenly.com/features/rendered-html) [     OG image templates Soon ](https://3.screeenly.com/features/og-image-templates) [     No-code integrations Soon ](https://3.screeenly.com/features/no-code-integrations)