Skip to main content
Glama
getscreenshot-mcp

getscreenshot-mcp

Official
by rasterwise

Capture screenshot

capture_screenshot
Read-only

Capture a public web page as a full-resolution image and inline JPEGs for vision models, so you can inspect live sites, verify deployed changes, or check mobile layouts.

Instructions

Capture a public web page as an image you can see. Renders the URL in a hosted headless Chromium and returns a link to the full-resolution image plus inline JPEG images sized for vision models. Use it to look at a live page, verify a deployed change, check a design on a phone preset, or read content that needs JavaScript to render. By default it captures the 1280x800 viewport (one image, about 1,300 image tokens). Pass element to capture one part of the page, or full_page: true for the whole page; tall captures come back as up to max_slices slices from the top. Costs 1 screenshot of the monthly quota per call, failed captures included. Takes about 9-25 seconds. Limits: the page must be reachable from the public internet (no localhost or private networks: use render_html for local HTML). One shot per call, no login flows or multi-step interaction. Sites with aggressive bot protection may refuse headless browsers.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "url": "<url>"
    },
    "name": "capture_screenshot"
  }
}

Input Schema

NameRequiredDescriptionDefault
urlYesAbsolute http(s) URL of the page, reachable from the public internet. localhost and private-network addresses cannot be captured; use render_html for local HTML.
langNoBrowser language / Accept-Language, for example "de-DE". API default "en-US".
clickNoCSS selector of an element to click once before the capture, for example to open a menu or dismiss a dialog. A selector that matches nothing makes the capture fail.
scaleNoDevice pixel ratio of the stored image: 1 (default), 2 (retina) or 3. Affects the file at screenshot_url; inline images are downscaled either way.
widthNoViewport width in CSS pixels. API default 1280. Ignored when device is set.
cookieNoOne cookie to set for the target host before loading, for example a feature flag or locale. Do not pass credentials or session tokens.
detailNoSize of the inline images. "standard": at most 1280 px wide, about 1,568 tokens per image. "high": at most 1920 px wide, about 4,784 tokens per image; use it only to read small text.standard
deviceNoDevice preset: sets the viewport and, for phones, a mobile user agent. Overrides width and height. iphone* and pixel* are phones, ipad* are tablets, the rest are fixed desktop sizes.
formatNoFormat of the stored file at screenshot_url. Inline images are always JPEG.png
heightNoViewport height in CSS pixels. API default 800. Ignored when device is set.
returnNo"image_and_url" attaches the images. "url_only" returns just the link and sizes, for when you need the file but do not need to look at it (uses no image tokens).image_and_url
elementNoCSS selector of one element to capture, for example "#pricing" or "main .hero". Cheapest way to look at one part of a page. A selector that matches nothing makes the capture fail (and still uses quota).
wait_msNoExtra wait in milliseconds after load, before the capture. API default 2000. Raise it for late animations or client-side rendering.
custom_jsNoJavaScript run in the page before capture.
full_pageNoCapture the whole scrollable page instead of just the viewport. Tall pages are returned as several image slices and cost more tokens.
highlightNoKeyword(s) to highlight on the page.
custom_cssNoCSS injected into the page before capture.
max_slicesNoMaximum number of inline images. A tall capture is cut into slices from the top; slices beyond this are left out and reported as truncated. Slices are also left out if the inline images would exceed about 1.5 MB in total (4 MB at high detail).
wait_untilNoWhen the page counts as loaded. API default "networkidle2". Use "domcontentloaded" for heavy or slow sites (fastest, most reliable); "networkidle0" waits for all network activity to stop.
hide_selectorsNoCSS selector(s) to hide (display:none) before capture, comma-separated, for example ".newsletter-popup, #ad-banner".
hide_chat_widgetsNoHide common chat widgets (Intercom, Drift and similar).
full_page_strategyNoFull-page capture strategy 0-3 (default 0). Try another value if a full_page capture renders badly (missing sections, repeated headers).
hide_cookie_bannersNoHide common cookie-consent banners. Set false to keep them in the picture.
transparent_backgroundNoKeep the background transparent where the page sets none (PNG only). Inline images are flattened onto white.

Output Schema

NameRequiredDescriptionDefault
formatYesFormat of the stored file at screenshot_url.
imagesYesOne entry per inline image, in the same order as the image blocks.
originalYesPixel size of the stored image. Null if it could not be read.
warningsYesThings worth knowing about this result, for example an error status from the page.
truncatedYesTrue when images were requested and some slices were left out, because of max_slices or the inline size limit. Always false with return: "url_only", where no image is attached by request.
origin_urlYes
captured_atYesISO 8601 UTC time of the capture.
page_statusYesHTTP status the page answered with. 400 or above usually means an error page was captured.
total_slicesYesHow many slices the whole image needs at this detail level, whether or not they are attached. With return: "url_only" this is how many a normal call would produce.
screenshot_urlYesFull-resolution image. Public to anyone who has the link.
expires_in_daysYesStored files are deleted after this many days.
estimated_image_tokensYesApproximate total vision tokens of the inline images.

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.0

TDQS

A4.8/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already cover safety (readOnlyHint=true, destructiveHint=false, openWorldHint=true), but the description adds substantial context beyond them: quota cost of 1 screenshot per call including failed captures, 9-25 second latency, the default 1280x800 viewport and its ~1,300 image token cost, slice truncation behavior, and the fact that aggressive bot protection may cause refusals. These are operational traits an agent cannot infer from the schema.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with purpose, then usage, cost, timing and limits in a logical order, with zero filler sentences. It is dense as a single paragraph for a 24-parameter tool; light structuring would make the limits and quota details easier to scan, but nothing is wasted.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Covers the full decision surface for a complex 24-param open-world tool: what it does, when to use it, cost, latency, reachability limits, and interaction limits. An output schema exists, yet the description still summarizes the return shape (link plus inline images) helpfully rather than redundantly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the schema carries the baseline. The description still adds value the schema lacks: element vs. full_page selection with max_slices slicing from the top, the token economics of detail levels, and phone presets as a design-check workflow.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ("Capture a public web page as an image you can see") and immediately explains the mechanism (hosted headless Chromium, returns link plus inline JPEGs). It also draws a boundary against a named sibling by directing local HTML to render_html.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives concrete positive cases (look at a live page, verify a deployed change, check a phone-preset design, read JS-rendered content) and explicit exclusions (no localhost/private networks, one shot per call, no login flows or multi-step interaction). It names the alternative tool for the excluded case, which is exactly the routing information an agent needs.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.