Url To Pdf

What the url-to-pdf MCP server does

The url-to-pdf MCP server accepts public HTTP or HTTPS URLs and produces four types of output: PDF documents, full-page PNG or JPEG screenshots, cleaned Markdown, and structured JSON. The JSON extraction includes the page title, meta description, canonical URL, language, main text, and links. Extraction is readability-first and does not use an LLM.

Each render response includes a temporary download URL and receipt metadata such as the final URL after redirects, HTTP status, capture time, duration, byte size, and SHA-256 hash. Download links remain valid for one hour. MCP responses can also include artifacts inline when they are 1 MB or smaller; screenshots are returned as images, while PDF, Markdown, and JSON are embedded resources.

How it works

The service is hosted at url-pdf.hammbox.com and renders pages on Cloudflare infrastructure. MCP-compatible clients connect to the Streamable HTTP endpoint at /mcp, while scripts and backend services can call individual /tools/* HTTP endpoints with POST requests. The same API key can be used for all four paid tools.

list_capabilities is free and reports available capabilities, tool prices, and where to obtain a key. The rendering and extraction tools require a key purchased through the service checkout. Authentication can be supplied with an x-api-key header or an Authorization: Bearer header.

Only public HTTP and HTTPS destinations are accepted. Internal and reserved network addresses are rejected at the edge. Generated outputs are automatically deleted after one hour.

Setup and configuration

Purchase an API key at https://url-pdf.hammbox.com/buy, then place it in the MCP client's HTTP server configuration. No local browser, headless Chrome installation, or local package installation is described as necessary. Claude Desktop, Cursor, and other clients that support Streamable HTTP can connect using the hosted MCP URL.

For direct automation, send JSON to the relevant HTTP endpoint and include the key in the request header. The HTTP API is suitable for scripts, backends, and low-code tools. A lost key can be recovered through the service's claim portal using the checkout email.

Tools and capabilities

• url_to_pdf supports A4, Letter, or Legal paper sizes, portrait or landscape orientation, print backgrounds, scaling, margins, and CSS page-size preferences.
• url_to_screenshot supports PNG or JPEG output, full-page capture, quality settings, element selectors, and clipping coordinates.
• url_to_markdown removes navigation-oriented content and can remove specified selectors or retain images.
• url_to_extract accepts text and link limits and returns structured page data.
• All render tools support viewport dimensions, device scale factor, dark mode, ad blocking, cookie-banner hiding, CSS-selector clicks, load conditions, selector waits, extra wait time, and navigation timeouts.

The service supports viewport widths from 320 to 3840 pixels, heights from 240 to 2160 pixels, and device scale factors from 0.5 to 3. It can wait for page states or specific elements before capturing dynamic content.

Limitations and notes

The paid tools require an API key and the documented subscription costs $5 per month, managed through Paddle. A short burst of heavy concurrent usage may receive HTTP 429 responses under the Cloudflare Browser Rendering free tier; retrying after a few seconds is recommended by the source material.

The service is intended for public pages, not private or internal URLs. Download links expire after one hour, so consumers that need longer retention must retrieve the artifact during that period. The README does not document local deployment, custom authentication providers, or support for authenticated target web pages.