Papyrio PDF tools for AI assistants

Package

papyrio-mcp

Tools

12 operations

Transport

stdio, runs locally

Install

You need Node.js 18 or later. npx fetches the package the first time the client starts it, so there is nothing to install by hand. Pick your client:

{  "mcpServers": {    "papyrio": {      "command": "npx",      "args": [        "-y",        "papyrio-mcp"      ]    }  }}

claude mcp add papyrio -- npx -y papyrio-mcp

{  "mcpServers": {    "papyrio": {      "command": "npx",      "args": [        "-y",        "papyrio-mcp"      ]    }  }}

{  "servers": {    "papyrio": {      "type": "stdio",      "command": "npx",      "args": [        "-y",        "papyrio-mcp"      ]    }  }}

Claude Desktop

~/Library/Application Support/Claude/claude_desktop_config.json on macOS, %APPDATA%\Claude\claude_desktop_config.json on Windows. Restart the app afterwards.

Claude Code

Run the command in a terminal. No file to edit.

Cursor

~/.cursor/mcp.json for every project, or .cursor/mcp.json in one.

VS Code

.vscode/mcp.json in the workspace.

Any other client that speaks MCP over stdio works the same way: the command is npx and the arguments are -y papyrio-mcp.

API key and limits

The server works without a key. It is then metered as ordinary free web traffic: 3 operations a day per IP address. That is enough to see whether it is useful, and not enough for real work.

For more, subscribe on RapidAPI, copy your key from Apps → Authorization, and pass it as PAPYRIO_API_KEY. The server then routes every call through the gateway and your plan’s quota applies. The Basic plan is free and needs no card.

{  "mcpServers": {    "papyrio": {      "command": "npx",      "args": [        "-y",        "papyrio-mcp"      ],      "env": {        "PAPYRIO_API_KEY": "your-rapidapi-key"      }    }  }}

claude mcp add papyrio -e PAPYRIO_API_KEY=your-rapidapi-key -- npx -y papyrio-mcp

{  "mcpServers": {    "papyrio": {      "command": "npx",      "args": [        "-y",        "papyrio-mcp"      ],      "env": {        "PAPYRIO_API_KEY": "your-rapidapi-key"      }    }  }}

{  "servers": {    "papyrio": {      "type": "stdio",      "command": "npx",      "args": [        "-y",        "papyrio-mcp"      ],      "env": {        "PAPYRIO_API_KEY": "your-rapidapi-key"      }    }  }}

Each tool call is one request. A job like “OCR four scans and merge them” costs five. Files stop at 50 MB.

Tools

Every tool takes absolute paths. output_path is optional and defaults to the input name with a -papyrio suffix, except on webpage_to_pdf, which has no input file to name the result after and so requires it.

Example prompts

You do not name tools. Describe the job and the assistant picks and chains them.

• OCR every scan in ~/invoices, merge them into one PDF, and compress it under 5 MB.
• Turn ~/Desktop/report.docx into a PDF and protect it with the password I give you.
• Save https://example.com/pricing as a PDF in ~/Downloads.
• Pull the text out of contract.pdf and list every date in it.

Privacy

The server runs on your machine and only touches the paths the assistant passes it. Each file goes to the Papyrio API, which holds it in memory, processes it, and drops it when the response goes out. We keep no copy. The result is written back to your disk.