ScreenshotInk MCP Server

![npm version](https://www.npmjs.com/package/@screenshotink/mcp) ![MCP](https://modelcontextprotocol.io) ![license](LICENSE)

Give your AI agent eyes. Five tools to capture, batch, compare and audit any web page — through ScreenshotInk's hosted Chromium fleet. No local browser, no Puppeteer maintenance; works in Claude Desktop, Claude Code, Cursor, Windsurf, CI runners and cloud sandboxes.

Every result returns the screenshot inline in the same turn, so the agent can reason about pixels immediately.

You:    Compare our pricing page on staging vs production — did the new cards ship?
Agent:  ▸ compare_screenshots { url_a: "https://acme.com/pricing", url_b: "https://staging.acme.dev/pricing" }
        4.7% of pixels differ, all inside the plan grid — the new Scale card is present. Deploy looks correct. ✓

Quick start

You need a ScreenshotInk API key — free, 100 captures/month, no card: screenshotink.com/signup

Claude Code

claude mcp add screenshotink -- npx -y @screenshotink/mcp --key sk_live_YOUR_KEY

or connect to the hosted remote server (nothing runs locally):

claude mcp add --transport http screenshotink https://mcp.screenshotink.com/mcp \
  --header "Authorization: Bearer sk_live_YOUR_KEY"

Claude Desktop

claude_desktop_config.json:

{
  "mcpServers": {
    "screenshotink": {
      "command": "npx",
      "args": ["-y", "@screenshotink/mcp"],
      "env": { "SCREENSHOTINK_API_KEY": "sk_live_YOUR_KEY" }
    }
  }
}

Cursor / Windsurf

{
  "mcpServers": {
    "screenshotink": {
      "command": "npx",
      "args": ["-y", "@screenshotink/mcp", "--key", "sk_live_YOUR_KEY"]
    }
  }
}

Remote server (cloud agents, hosted runtimes, custom connectors)

Anything that speaks Streamable HTTP can connect directly — nothing to install. In Claude or ChatGPT, add it as a custom connector with the URL below and sign in over OAuth; for scripts/CI, pass your key as a bearer token:

URL:   https://mcp.screenshotink.com/mcp
Auth:  OAuth 2.1 (sign in on first use) — or Authorization: Bearer sk_live_YOUR_KEY

Tools

| Tool | What it does | |---|---| | take_screenshot | Capture any URL — viewport or full page, png/jpeg/pdf, dark mode, ad & cookie-banner blocking, lazy-load handling | | bulk_screenshots | Up to 20 URLs in one call — audit a funnel or a competitor set at once | | compare_screenshots | Pixel-diff two URLs (staging vs production, before vs after) — returns changed-% and a highlighted diff image | | run_lighthouse | Google Lighthouse scores (performance, accessibility, best practices, SEO) + core web vitals | | capture_sitemap | Fetch a sitemap.xml and screenshot every page it lists |

Same parameters and defaults as the REST API. Captures over MCP count against the same quota as the REST API — one key, one meter. Identical requests within 24h are served from cache for free.

Why hosted?

CI pipelines — no headless Chrome install, no flaky browser binaries in runners.
Cloud agents — hosted runtimes can't launch browsers; the remote endpoint gives them eyes with zero dependencies.
Zero maintenance — we patch Chromium, handle lazy-load, ads and cookie walls. Your agent just asks for pixels.

screenshotink-mcp