ScreenshotOne MCP. Audit websites and generate reports from URLs.

Listen up. You gotta use ScreenshotOne when you need your AI agent to grab high-quality, programmatic screenshots or PDFs from any website URL without ever opening a browser tab. This whole server lets your agent handle visual auditing, whether you're pulling basic info, capturing specific little components, or demanding an image that runs the full length of a page.

It turns what used to be a massive headache—manually checking websites for QA or design review—into something simple, handled right in conversation.

It starts with gathering intel before it even takes a picture. If you just need structural data and don't want to waste time running a full visual capture, your agent can pull basic website info using get_page_metadata. This tool grabs site titles, descriptions, and page sizes from a URL instantly.

Next, if you need a final report—say for client sign-off or an internal audit—you’ll use the generate_pdf function. It takes all the content from a specified website URL and compiles it into one clean PDF file. You're done with messy screenshots; you've got a professional document.

Now, let's talk pictures. For full-page coverage, you can't just rely on what fits in a standard window. When you need to capture every single pixel from the top edge right down to the bottom, use take_full_page_screenshot. This tool gives you the complete view of the site design and content flow, making sure nothing gets cut off.

If that's overkill—if you just want to see what a user sees by default—you can run take_screenshot. That captures a standard image based on the visible viewport area.

But sometimes you need precision. You might only care about one specific widget or button on a page, and the whole screenshot is noise. In that case, use take_element_screenshot. This function lets your agent focus solely on capturing just one CSS element from a webpage. It’s perfect for QA checks where you're verifying a single component works right.

For testing different devices or layouts—think mobile vs. desktop—you need control over the view size. take_viewport_screenshot is what you use for that. You pass it an exact pixel size, so your agent generates a screenshot that matches a specific, custom screen resolution. This capability is essential when verifying how your site looks on niche devices or testing tricky responsive layouts.

Putting it all together, your agent can manage multiple visual tasks in sequence. It could first run get_page_metadata to confirm the page title, then use take_viewport_screenshot to capture a mobile view, and finish by running generate_pdf so you've got both an image record and a shareable document, all without you ever touching Chrome or Safari.

You don't need any human intervention for these checks; your agent handles the whole process using these tools.