getscreenshot-mcp
OfficialServer Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| PORT | No | Port the hosted server listens on. | 8080 |
| TRIAL_API_KEY | No | Turns on keyless preview mode. Off by default; not for production yet. | |
| IMAGE_QUEUE_MAX | No | Number of image jobs that may wait for a slot. | 24 |
| PUBLIC_BASE_URL | No | Public origin of the MCP endpoint. Must be set in production. | http://127.0.0.1:<port> |
| IMAGE_CONCURRENCY | No | Number of image jobs that run at once. | 3 |
| SHARP_CONCURRENCY | No | libvips threads per image job. | 1 |
| COMPARE_MAX_HEIGHT | No | Maximum rows compare_screenshots compares. | 2400 |
| TRUSTED_PROXY_HOPS | No | Proxies between Google's front end and the container that append to X-Forwarded-For. | 0 |
| IMAGE_QUEUE_WAIT_MS | No | Longest a job waits for a slot before it is refused. | 20000 |
| HTML_PUBLIC_BASE_URL | No | Public origin used in render_html links. | value of PUBLIC_BASE_URL |
| GETSCREENSHOT_API_KEY | Yes | Your GetScreenshot API key. Required for the server's tools to work. | |
| INLINE_MAX_BYTES_HIGH | No | Cap on the total inline JPEG bytes of one result at detail high. | 4000000 |
| RATE_LIMIT_PER_MINUTE | No | Tool calls per caller per minute. | 60 |
| GETSCREENSHOT_API_BASE | No | REST API base URL. For tests. | https://api.rasterwise.com |
| GETSCREENSHOT_MCP_BASE | No | Service that hosts HTML for render_html in stdio mode. | https://mcp.getscreenshotapi.com |
| HTML_MAX_DOCS_PER_CALLER | No | Live render_html documents per caller. | 5 |
| TRIAL_DAILY_LIMIT_GLOBAL | No | Preview captures per UTC day in total. | 200 |
| TRIAL_DAILY_LIMIT_PER_IP | No | Preview captures per client address per UTC day. | 5 |
| INLINE_MAX_BYTES_STANDARD | No | Cap on the total inline JPEG bytes of one result at detail standard. | 1500000 |
| RATE_LIMIT_PER_IP_PER_MINUTE | No | Tool calls per client address per minute across all keys. | 120 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
GetScreenshot captures public web pages with a hosted headless browser and returns images sized for vision models. Use capture_screenshot to look at a live URL; pass element to look at one part of a page cheaply, and full_page only when you need the whole page. Use render_html to see HTML you have as a string (a component or page you just wrote); this is also the way to check local work, because the service cannot reach localhost or private networks. Use compare_screenshots to verify a visual change (before/after, staging vs production), capture_pdf for an A4 PDF link, and get_usage to check the remaining quota. Every capture uses one screenshot of the monthly quota, failed captures included, and takes about 9-25 seconds, so fix the input instead of retrying a failure unchanged. If a page fails to load, retry once with wait_until: "domcontentloaded". Captures are single shots of a page: there is no session, login flow or multi-step interaction.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| capture_screenshotA | 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 |
| render_htmlA | Render an HTML string in a real browser and return a screenshot you can see. Use it to check a component, email or page you just wrote without deploying it, and to look at local work, since the capture service cannot reach localhost. The HTML is hosted at an unguessable temporary URL, captured like any page, deleted as soon as the capture returns, and the result has the same shape as capture_screenshot. Assets must be inlined or use absolute public URLs: relative paths and localhost URLs inside the HTML will not load unless the HTML sets to a public origin. The page runs in a sandbox without cookies or localStorage, so code that depends on them may render blank. Cookie-banner hiding is off for this tool, so the page renders exactly as written. Costs 1 screenshot of the monthly quota per call. Takes about 9-25 seconds. Do not put secrets in the HTML: the resulting image link is public to anyone who has it. |
| compare_screenshotsA | Visual diff of two pages or two existing screenshots. Use it to verify a change did what you intended and nothing else moved: before/after a deploy, staging vs production, or one page at two points in time. Each side is either |
| capture_pdfA | Save a public web page as an A4 PDF and return a link to the file. Use it when the user wants a document to keep, share or archive. The PDF is not shown to you; to look at a page use capture_screenshot. Costs 1 screenshot of the monthly quota per call, failed captures included. Takes about 9-25 seconds. The link is public to anyone who has it and expires in 30 days. Limits: the page must be reachable from the public internet (no localhost or private networks). |
| get_usageA | Report how many screenshots the configured API key has used this month, the plan's monthly limit, how many are left and when the quota resets. Use it before a large batch of captures or when a capture fails with a quota error. This check itself costs 1 screenshot of the quota, so do not call it routinely. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 5 tools
Each tool has a clearly distinct purpose: public-page screenshot, local HTML rendering, visual comparison, PDF export, and quota reporting. capture_screenshot and render_html overlap in output type, but the descriptions sharply separate public URLs from local HTML strings.
All five tools follow a consistent verb_noun snake_case pattern: capture_screenshot, render_html, compare_screenshots, capture_pdf, get_usage. There are no mixed naming conventions or vague standalone verbs.
Five tools is well-scoped for a screenshot service, and each tool earns its place by covering a distinct workflow. The set avoids both a thin single-tool surface and an unnecessarily large catalog.
The set covers the core lifecycle: capture public pages, render local HTML, compare versions, export PDFs, and check quota. A minor gap is that PDF export is only available for public URLs, not local HTML rendered via render_html, though this is a workaroundable limitation.