MCP Screenshot Website Fast

Community

Fast, efficient webpage screenshots optimized for AI vision workflows.

How to install ↓ Compare GitHub source repository ↗

Category: Dev Tools #258 of 438
Stars: ★ 110 Very popular
Transport: stdio (local process)
Runtime: Node.js 20+
Credentials: No credential needed
License: MIT
Last commit: 22 days ago
Tools: 3

50FMRS · D

This server is optimized for AI workflows, capturing screenshots in a format ideal for vision models. The automatic tiling is excellent for full-page captures. Installation is straightforward across multiple clients. However, there is a resolution cap, and it relies on an external browser, which may require additional setup in some environments. Overall, it's reliable and practical.

Strongest · Documentation 14/20 Weakest · Reliability 7/20

Reliability

7/20

Security and permissions

8/20

Maintenance

8/20

Documentation

14/20

Setup experience

13/20

Why each score

Reliability 7/20: Evidence is limited to the README and repo metadata: it describes take_screenshot, capture_selector, take_screencast and an auto-restart wrapper, but there is no source code, no tests, no CI workflow, and no package manifest to verify the MCP init/tool-list handshake; the server manifest is also not cached. Under static calibration the score must be conservative because execution evidence is absent; the happy path is plausible but error handling and edge cases are thinly evidenced, so points are deducted.
Security and permissions 8/20: The README shows no credential handling and no malware/real-secret red flags; however the server by design fetches arbitrary URLs, take_screencast accepts jsEvaluate for arbitrary JavaScript injection, and Puppeteer downloads a browser automatically without documented sandboxing, network-boundary controls, or confirmation for sensitive operations. The main risks are visible but permissions/confirmation/scoping are incomplete, hence 8.
Maintenance 8/20: The repo metadata shows an MIT license, non-archived status, and 0 open issues, and the README claims a GitHub Actions Release workflow; but there is no commit history, release timeline, dependency-update policy, or security-response channel. Stars are treated only as a discovery signal, not as points. Activity is suggested but governance/versioning evidence is thin, so maintenance gets 8.
Documentation 14/20: The README is well structured with multi-client install instructions, parameters and examples for the three tools, CLI options, dev commands, architecture, and troubleshooting; however it lacks security/network-boundary notes, cost or additional limitation details, and source files cannot be compared to verify the docs match reality. Documentation is usable but leaves hidden assumptions and troubleshooting gaps, so 14.
Setup experience 13/20: Setup steps are clear and include configuration examples for Claude Code, VS Code, Cursor, JetBrains, and raw JSON, with prerequisites stated explicitly (Node 20+, npm/npx, automatic Chromium download). However static calibration caps setup at 15 without verifiable execution evidence (real CI workflows plus committed tests covering key paths), and no package manifest or startup logs are available for verification, so setup gets 13.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessWrites / deletes local filesUses the networkControls a browser

Best for

Development workflows that need to provide screenshots to AI models
AI coding tools like Claude Code, Cursor that support MCP
Automated testing scenarios requiring full-page screenshots

Not for

Scenarios requiring high-resolution original screenshots (pixel limit)
Real-time applications where latency is critical (no caching but still dependent on load times)
Situations requiring complex browser interactions such as login or cookie management

Required permissions

Network access to load target web pages
Local file system access when saving screenshots to a directory
Execution of Puppeteer, which downloads Chrome/Chromium browser

Risks and side effects

Server crash may trigger auto-restart (with exponential backoff limit)
Screenshots may contain sensitive information; handle accordingly
Dependency on Puppeteer; network/firewall may block browser download

Setup

Before you start

Runtime：Node.js 20+

Other optional settings (1)

PUPPETEER_SKIP_CHROMIUM_DOWNLOAD optional Optional; set to true to skip Puppeteer's automatic Chromium download and supply a custom browser executable, useful when downloads are blocked.

For Claude Code: claude mcp add screenshot-website-fast -s user -- npx -y @just-every/mcp-screenshot-website-fast. For VS Code, use the code --add-mcp command or JSON config. For JetBrains IDEs, add via settings with JSON. Alternatively, add the JSON config to any MCP client's mcp.json.

claude_desktop_config.json

{
  "mcpServers": {
    "screenshot-website-fast": {
      "command": "npx",
      "args": [
        "-y",
        "@just-every/mcp-screenshot-website-fast"
      ]
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "screenshot-website-fast": {
      "command": "npx",
      "args": [
        "-y",
        "@just-every/mcp-screenshot-website-fast"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add screenshot-website-fast -- npx -y @just-every/mcp-screenshot-website-fast

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

After installing, take_screenshot, capture_selector, and take_screencast should appear in the client's tool list; calling take_screenshot(url="https://example.com") and receiving an image confirms the connection works.

Troubleshooting

If Puppeteer fails to download Chrome, set PUPPETEER_SKIP_CHROMIUM_DOWNLOAD=true and provide custom executable
If screenshot quality is poor, adjust viewport dimensions or wait strategies
For timeout errors, increase wait time or use networkidle strategy

Things to try

Once connected, you can ask your AI assistant things like:

Take a full-page screenshot of https://example.com
Capture just the #main element on https://example.com
Record a 10-second screencast of https://example.com
Save screenshots of https://example.com to the /path/to/screenshots directory

Tools 3

take_screenshot read-only

Captures a high-quality screenshot of a webpage, with options for viewport dimensions, fullPage tiling, wait conditions, and saving to a directory for file paths instead of base64.

capture_selector read-only

Captures a screenshot of a specific DOM element matched by a CSS selector, with support for wait strategies and timeout.

take_screencast read-only

Captures a series of screenshots over time to create a screencast, with configurable duration, interval, optional JavaScript execution, and saving as animated WebP.

Use cases

Taking web page screenshots for AI vision workflows

Capturing specific elements for automated testing

Recording screencasts of dynamic page changes

Supported clients

Claude Desktop

VS Code

Cursor

JetBrains IDEs

Listed from the project's documentation, not tested by this site.

Overview

This MCP server uses Puppeteer headless browser to quickly capture web page screenshots, automatically tiling full pages into 1072x1072 pixel chunks for optimal processing by AI models like Claude Vision API. It supports configurable viewports, wait strategies, selector capture, and screencast recording with optional animated WebP output. Designed for CLI coding tools like Claude Code, VS Code, Cursor, it integrates seamlessly into AI workflows.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 63.3k · Tools 2 Compare with this →

NPM Sentinel MCP Server 79 · B

AI-powered NPM package analysis MCP server

★ 18 · Tools 19 Compare with this →

MartinLoop MCP Server 79 · B

An independent control layer that gives AI coding agents budgets, brakes, and receipts

★ 413 Compare with this →

GitHub MCP Server 76 · B

GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language

★ 33.7k · Tools 15 Compare with this →

Compare with other MCP servers

Source revision e23fbe0afcc4 Data synced 2026-10-19 Read the FMRS scoring method