# GetScreenshot > GetScreenshot gives AI agents eyes: a hosted screenshot API with an MCP server that returns images sized for vision models, from $5/month. One GET request renders a public web page in a hosted browser and returns JSON with the URL of a PNG, JPEG or PDF. It works from places that cannot launch a browser: cloud coding agents, sandboxes and CI runners. GetScreenshot is made by Rasterwise, LLC. Everything needed to make a correct API call is on this page. **Endpoint** ``` GET https://api.rasterwise.com/v1/get-screenshot?apikey=KEY&url=ENCODED_URL ``` **Authentication.** Pass the API key as the `apikey` query parameter. Header authentication is not supported on the REST API. A person gets a key by subscribing at https://www.getscreenshotapi.com/#pricing and the key arrives by email. Every plan is paid (the first month of the $5 plan is free with code FREE5) and an agent cannot sign up by itself. If you are an agent without a key, ask your user for one. Details: https://www.getscreenshotapi.com/auth.md **Example** ```bash curl "https://api.rasterwise.com/v1/get-screenshot?apikey=$GETSCREENSHOT_API_KEY&url=https%3A%2F%2Fexample.com&fullpage=true" ``` **Response.** Always JSON, never image bytes and never a redirect. Fetch `screenshotImage` to get the file. ```json { "status": "success", "originUrl": "https://example.com", "pageStatus": 200, "captureDate": "2026/11/18", "captureHour": "14:03:22", "aiAnalysis": null, "screenshotImage": "https://storage.googleapis.com/getscreenshot-storage/imageAbC123xYz9.png" } ``` **Five parameters cover most calls.** All parameters are lowercase query parameters. URL-encode every value. | Parameter | Default | What it does | |---|---|---| | `url` | required | Page to capture. `http` or `https`, URL-encoded. Must be reachable from the public internet. | | `fullpage` | `false` | `true` captures the full scrollable page instead of the viewport. | | `width`, `height` | `1280`, `800` | Viewport size in CSS pixels. | | `element` | none | CSS selector. Captures only that element. A selector that matches nothing causes a 500. | | `format` | `png` | `png` or `jpeg`. For a PDF pass `pdf=true`; the response then has `pdfFile` instead of `screenshotImage`. | Things to know before calling: - A capture takes about 9 seconds (median) and can take up to 90 seconds. Set a client timeout of 90 seconds or more. - Every request with a valid key uses 1 unit of the monthly quota. Failed captures count. - Images are stored for 30 days at a public URL. Anyone with the URL can fetch the file. - If a page is slow or the call fails with a 500 "Could not load the target page", retry with `waituntil=domcontentloaded`. - For a phone or tablet view pass `preset` (for example `preset=iphone12`). With a preset, use `noheight=true` for a full-page capture. - An unrecognised API key usually returns HTTP 500 with header `x-amzn-errortype: AuthorizerConfigurationException`, and can also return 403 `{"message":"Forbidden"}`. A missing key returns 401. A used-up quota returns 429 `{"message":"Limit Exceeded"}`. - The API cannot reach `localhost` or private networks, and it does single captures, not multi-step logged-in browsing sessions. **MCP server.** If your client supports the Model Context Protocol, use the MCP server instead of calling the REST API by hand. It returns the image inline, downsized and sliced to fit a vision model's input budget. Run it locally with `npx -y getscreenshot-mcp` and the environment variable `GETSCREENSHOT_API_KEY`, or connect to the hosted endpoint `https://mcp.getscreenshotapi.com/mcp` with the header `Authorization: Bearer `. Tools: `capture_screenshot`, `render_html`, `compare_screenshots`, `capture_pdf`, `get_usage`. It works with any API key on any plan. ## API reference - [Full reference in one file](https://www.getscreenshotapi.com/llms-full.txt): every parameter, the response and error tables with fixes, worked examples in curl, Python and Node, the MCP server, the CLI and the agent skill. - [OpenAPI 3.1 description](https://www.getscreenshotapi.com/openapi.yaml): machine-readable contract for `/get-screenshot`, `/usage` and `/get-screenshot/validate-key`. - [API catalog](https://www.getscreenshotapi.com/.well-known/api-catalog): RFC 9727 linkset that points to the OpenAPI file and the docs. - [API reference (docs)](https://docs.rasterwise.com/docs/getscreenshot/api-reference-0/): the human-readable parameter reference. - [Getting started (docs)](https://docs.rasterwise.com/docs/getscreenshot/getting-started/): first request in several languages. - [API examples](https://www.getscreenshotapi.com/api/examples.html): copy-paste request examples. ## Authentication and pricing - [auth.md](https://www.getscreenshotapi.com/auth.md): how an agent gets and uses a key, where the key goes for REST, MCP and CLI, what each auth error looks like, and what to tell your user. - [pricing.md](https://www.getscreenshotapi.com/pricing.md): plans ($5, $10 and $20 per month), per-screenshot price, what counts as a request, rate limits and image retention. ## For AI agents - [AI agents hub](https://www.getscreenshotapi.com/agents/): overview of the MCP server, CLI and agent skill. - [MCP server](https://www.getscreenshotapi.com/agents/mcp-server.html): install, tools, output format and image sizing rules. - [Agent skill](https://www.getscreenshotapi.com/agents/agent-skill.html): the `getscreenshot` skill, installed with `npx skills add rasterwise/getscreenshot-mcp`. - [MCP server docs](https://docs.rasterwise.com/docs/agents/mcp-server/): reference for the MCP tools and their inputs. - [Agent skill docs](https://docs.rasterwise.com/docs/agents/agent-skill/): reference for the agent skill. - [LLM quickstart](https://docs.rasterwise.com/docs/agents/llm-quickstart/): a quickstart written for LLMs and agents. - [Claude Code setup](https://www.getscreenshotapi.com/agents/claude-code.html): add the MCP server to Claude Code. - [Cursor setup](https://www.getscreenshotapi.com/agents/cursor.html): add the MCP server to Cursor. - [Claude Desktop setup](https://www.getscreenshotapi.com/agents/claude-desktop.html): add the MCP server to Claude Desktop. - [Codex setup](https://www.getscreenshotapi.com/agents/codex.html): add the MCP server to Codex. - [VS Code and GitHub Copilot setup](https://www.getscreenshotapi.com/agents/vscode-copilot.html): add the MCP server to VS Code. - [Gemini CLI setup](https://www.getscreenshotapi.com/agents/gemini-cli.html): add the MCP server to Gemini CLI. - [Windsurf setup](https://www.getscreenshotapi.com/agents/windsurf.html): add the MCP server to Windsurf. - [Visual verification for coding agents](https://www.getscreenshotapi.com/use-cases/visual-verification-for-coding-agents.html): let a coding agent look at the page it just changed. - [GetScreenshot vs Playwright MCP](https://www.getscreenshotapi.com/compare/vs-playwright-mcp.html): when to use a hosted screenshot API and when to use a local browser tool. ## Features and integrations - [Features](https://www.getscreenshotapi.com/features/): full-page capture, element targeting, PDF generation, viewport control, custom CSS and JavaScript, cookie banner and chat widget removal, keyword highlighting, email and webhook delivery. - [Integrations](https://www.getscreenshotapi.com/integrations/): Zapier, Make, n8n and Google Sheets. Zapier, Make and n8n need the Hamilton plan or higher. - [Use cases](https://www.getscreenshotapi.com/use-cases/): website monitoring, SEO audits, archiving, reporting and more. ## Optional - [Comparisons](https://www.getscreenshotapi.com/compare/): GetScreenshot next to other screenshot APIs. - [Glossary](https://www.getscreenshotapi.com/glossary/): definitions of screenshot API terms. - [Homepage](https://www.getscreenshotapi.com/): product overview, live demo and checkout. - [Support](mailto:support@rasterwise.com): support@rasterwise.com