← Docs

Bloodweb tools for AI assistants

Claude, Cursor, VS Code and other assistants that speak the Model Context Protocol (MCP) can use all 20 Bloodweb services as tools: make PDFs, convert and resize images, read PDFs, make QR codes and barcodes, and convert units, colours and currencies. Each tool call is paid from your Bloodweb account like an API call, and a failed call costs nothing.

Set it up

Create a free account and make an API key. Accounts come with free tokens every month.
Add the server to your assistant, below. It is a remote server (streamable HTTP) at this address, with your key as a header:
```
https://bloodweb.net/mcp
Authorization: Bearer YOUR_KEY
```
Ask for something. The assistant picks the tool and shows you the result.

Claude Code

One command in a terminal:

claude mcp add --transport http bloodweb https://bloodweb.net/mcp --header "Authorization: Bearer YOUR_KEY"

Then /mcp inside Claude Code lists it with its tools.

Cursor

In ~/.cursor/mcp.json (every project) or .cursor/mcp.json (one project):

{
  "mcpServers": {
    "bloodweb": {
      "url": "https://bloodweb.net/mcp",
      "headers": { "Authorization": "Bearer YOUR_KEY" }
    }
  }
}

VS Code

In .vscode/mcp.json. VS Code asks for the key the first time and keeps it out of the file:

{
  "inputs": [
    { "type": "promptString", "id": "bloodweb-key", "description": "Bloodweb API key", "password": true }
  ],
  "servers": {
    "bloodweb": {
      "type": "http",
      "url": "https://bloodweb.net/mcp",
      "headers": { "Authorization": "Bearer ${input:bloodweb-key}" }
    }
  }
}

The tools then show in Copilot Chat's agent mode.

Claude Desktop

Claude Desktop's settings file starts local servers, so a small bridge, mcp-remote, connects it to ours (it needs Node.js). In Settings > Developer > Edit Config:

{
  "mcpServers": {
    "bloodweb": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://bloodweb.net/mcp", "--header", "Authorization:${BLOODWEB_AUTH}"],
      "env": { "BLOODWEB_AUTH": "Bearer YOUR_KEY" }
    }
  }
}

Restart Claude Desktop after saving.

Other clients

Any client that can add a remote MCP server over streamable HTTP with a custom header works the same way: the address above, and Authorization: Bearer YOUR_KEY. Clients that only accept sign-in by OAuth (ChatGPT's connectors, for example) cannot use it yet.

Things to ask for

"Turn this HTML into an A4 PDF and give me the link."
"Make a QR code for https://example.com, 600 pixels, as a PNG."
"Convert these iPhone photos from HEIC to JPG."
"Resize this image to 1200 by 630 for a social card, cropping from the centre."
"What does page 2 of this PDF say?"
"Is 'Gifts' a good enough customs description for a parcel to Germany?"

Files in and out

The assistant sends a file to a tool as base64 text, up to about 15MB per call.
A file result comes back as a private download link that works for 10 minutes; the file is then deleted. Images also come back inline, so the assistant can see them.
Nothing else is kept: we keep a record of each call, never the file. The terms have the details.

The tools

Tool	What it does
`favicon-generate`	Favicon generator: Turn one image into a full favicon set: favicon.ico, PNG sizes, the Apple touch icon, Android icons, site.webmanifest and the HTML tags.
`image-compress`	Image compressor: Shrink an image for the web: lower quality, strip metadata, cap the width, or switch to WebP or AVIF.
`image-convert`	Image converter: Convert an image to JPG, PNG, WebP, AVIF, GIF, TIFF or BMP. Reads HEIC (iPhone photos) and ICO too.
`image-metadata`	Image metadata reader: Read an image's format, dimensions, frames and colour details, plus EXIF camera data, date taken and GPS location.
`image-resize`	Image resizer: Resize an image to fit inside a box, fill it and crop, fit and pad, or stretch to exact dimensions.
`html-to-pdf`	HTML to PDF: Turn HTML and CSS into a PDF: invoices, receipts, reports, tickets. Rendered by Chromium, so modern CSS (flexbox, grid, web fonts) works.
`images-to-pdf`	Images to PDF: Combine photos, scans or receipts into one PDF, one image per page, in the order sent.
`pdf-compress`	PDF compressor: Make a PDF smaller by downsampling its images. Choose how far: screen (smallest), ebook, printer or prepress.
`pdf-merge`	PDF merger: Combine up to 20 PDFs into one, in the order you send them. Pages are copied as they are, with no loss of quality.
`pdf-split`	PDF splitter: Extract pages into a new PDF, remove pages, or split a PDF into one file per page.
`pdf-to-image`	PDF to image: Render PDF pages as PNG, JPG or WebP images at the resolution you choose. Good for thumbnails and previews.
`pdf-to-text`	PDF to text: Extract the text from a PDF, keeping the layout or as plain reading order, for all pages or a range.
`barcode-generate`	Barcode generator: Make product and shipping barcodes: EAN-13, EAN-8, UPC-A, UPC-E, Code 128, Code 39, Code 93, ITF-14 and Codabar, as PNG or SVG.
`qr-generate`	QR code generator: Make a QR code from a URL or text, as PNG or SVG, in your colours, with your logo in the centre if you like.
`csv-json`	CSV and JSON converter: Convert CSV to JSON or JSON to CSV. Handles quoted fields, line breaks inside cells, semicolon and tab files, BOMs and nested JSON.
`color-convert`	Colour converter: Convert a colour between hex, RGB, HSL, HSV, CMYK and CIE Lab, find its nearest CSS name, and check WCAG contrast against another colour.
`currency-convert`	Currency converter: Convert between about 30 currencies using the European Central Bank's daily reference rates.
`unit-convert`	Unit converter: Convert between 104 units of length, mass, volume, area, temperature, speed, time, data, pressure, energy, power and angle.
`audio-convert`	Audio converter: Convert audio between MP3, OGG, Opus, M4A, WAV and FLAC. Trim it, make it mono, or change the sample rate.
`customs-description-check`	Customs description checker: Check customs goods descriptions before you ship. Flags the vague wording EU customs (ICS2) rejects, category-only descriptions, product codes and missing materials.

Each call costs the same as the same API call, from the same balance. Prefer code? Every service has tested examples in curl, Node.js, Python and PHP, and the whole API is in the docs.