Use the ScreenshotOne MCP to give your agents screenshots of any web page, without ads or cookie banners, plus the page text as clean Markdown.
Works with every MCP client:
Stripe brand colorsPull a brand palette from a screenshot
Stripe's home page pairs a tight core palette with a bright ribbon gradient:
#061B31 for headlines and navigation#533AFD for buttons and accents#64748D for supporting headline text#B4BDFE #FE9A15 #FE5B66 #FE85E5 #8D60F0 in the hero:root { --stripe-navy: #061B31; --stripe-blurple: #533AFD; --stripe-slate: #64748D; --stripe-ribbon: linear-gradient( 120deg, #B4BDFE, #FE9A15, #FE5B66, #FE85E5, #8D60F0 );}The ribbon stops are sampled from the screenshot, so treat them as close approximations.
Found 3 leads in leads.csv. Capturing each homepage now.
Done. I added a notes column to leads.csv:
company,website,notesBoltAI,boltai.com,"Clear headline and CTAs; customer logos start below the fold"UserJot,userjot.com,"Right half of the hero is empty; product preview starts below the fold"Framekit,framekit.ai,"A news popup covers the headline and subheadline on first load"Strongest pitches:
With standard pricing, Stripe keeps $1.75 from a $50 domestic card payment, so you receive $48.25:
For an international card, add 1.5% and the fee becomes $2.50. If currency conversion is also needed, another 1% brings it to $3.00.
Linear's September 24 update has two parts:
For a small team, adaptive routing is the one to try: it saves tokens on routine issues without any setup. Team project labels and statuses are only on Business and Enterprise plans.
The release looks good. I reviewed the full page in three slices:
I can rerun this check after each deploy and attach the slices to the release notes.
A practical agent loop
The agent chooses what to capture. ScreenshotOne handles the browser work. The agent can then inspect the result with a vision model or include it in its output.
Provide a URL and the state the agent needs to observe, such as a full page or a specific viewport.
ScreenshotOne loads the page, applies capture options, and returns the final screenshot.
The agent uses the image as visual context or preserves it as evidence for the completed task.
Capture a full-page screenshot of screenshotone.com/pricing. Block ads and cookie banners, return a cached URL, and verify that the pricing table is visible.
{
"url": "https://screenshotone.com/pricing/",
"full_page": true,
"block_ads": true,
"block_cookie_banners": true,
"response_type": "json",
"cache": true
}
The pricing table is visible. Basic, Growth, and Scale render without overlaps.
cache_url: https://cache.screenshotone.com/…
Read how Uneed uses ScreenshotOne with AI agents →ScreenshotOne is so simple to use I didn't even realized my agents used it to take screenshots. I followed the journey of ScreenshotOne since the beginning, and it's among the few products I can put 100% of my trust in.
Agent integrations
Start with a hosted tool, install a reusable skill, or make screenshot capture part of your application code.
Connect a compatible agent through OAuth and render screenshots without running a local server or copying an API key into the client.
Install a reusable screenshot workflow for Codex and other agents that support skills.
Call ScreenshotOne directly when screenshots are a deterministic step inside your own agent product or workflow.
Workflows
Screenshots show how a page looks, and Markdown shows what it says. Together they let your agent check, compare, and report on any page.
Let an agent capture the rendered result after a release and include visual proof with its QA summary.
Combine a screenshot with clean Markdown when an agent needs to understand both how a page looks and what it says.
Use ScreenshotOne inside a scheduled agent or automation workflow to capture the same page over time.
Return temporary image URLs and full-page slices an agent can inspect or preserve with a report.
Have an agent read competitor pricing pages as Markdown and keep a screenshot of each plan table it compares.
Give an agent the rendered page and its copy so it can critique the layout, messaging, and calls to action.
We thought making a screenshot tool for our AI agents would be simple. But we faced countless unexpected problems.
So, we ditched the idea and chose ScreenshotOne API instead. It saved us lots of time and it just works!
Reliable browser execution
Real pages load asynchronously, show overlays, and change across devices. Capture controls make the visual input cleaner and more predictable.
Wait for browser events, a CSS selector, or an explicit delay before the screenshot is captured.
Block ads and cookie banners or hide selected elements before an image reaches the agent.
Choose a viewport, emulate a device, capture the full page, or isolate one element with a selector.
Return the image directly or use JSON and caching when an agent needs a reusable screenshot URL.
Quick answers to common questions about the ScreenshotOne MCP and giving your agents web page screenshots and text.
https://mcp.screenshotone.com that connects your ScreenshotOne account to any MCP-aware AI client. Your agent can render website screenshots, split long pages into full-page slices, and extract clean page content as Markdown.https://mcp.screenshotone.com, and you connect with OAuth in the browser. If you prefer a local setup, an open-source command-line server is also available on GitHub.Read
Practical guides to taking website screenshots with MCP and choosing the right tools for your agent.
Take screenshots with Vercel Labs' agent-browser. Capture a viewport, full page, or element; save PNG or JPEG files; and control image resolution.
Set up Playwright MCP, capture viewport, full-page, and element screenshots, choose image resolution, and find the saved files.
Take viewport, full-page, and element screenshots with Chrome DevTools MCP. Save PNG, JPEG, or WebP files and control image dimensions.
Add the hosted MCP server to your client, approve access in the browser, and let your agent capture live pages, extract clean Markdown, and verify what actually rendered.