# superPDF AI

> superPDF AI is an open-source MCP server for files and documents. Local PDF tools for agents: read, convert, merge and split files without uploading them. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, macOS, Windows, Linux, self-hosted
- **Website:** https://superpdf.ai/docs/mcp-api/mcp-tool-access
- **Repository:** https://github.com/arjunprabhulal/superpdf
- **MCP Registry name:** io.github.arjunprabhulal/superpdf
- **Categories:** files and documents
- **Listing updated:** 2026-11-15

## About

Setup starts with an npm install of the @superpdf/cli package, which needs Node.js 22 or later, then pointing your agent at `superpdf mcp --allow-dir` with a folder you create. The agent can only touch files inside that folder, and the server refuses to start if the folder does not exist.

Processing happens on your machine and nothing is uploaded. The README explains that the reader works from the text, fonts and drawings inside a PDF rather than a language model, rebuilding headings, tables and charts as Markdown, HTML or JSON, and sends scans through Tesseract OCR. One caveat the docs give: when the agent reads document text, that text goes to the agent's own model.

The same engine also ships as a command line tool, a Python and JavaScript SDK, a local REST API, and loaders for LangChain and LlamaIndex. The author reports a score of 65.0 on the ParseBench suite from its own run and notes it is not on the official leaderboard.

## Features

- **Folder-restricted access:** Agents can only open PDFs inside the directory you pass with --allow-dir.
- **Structure extraction:** Output as Markdown, HTML or JSON with headings, paragraphs and a section outline.
- **Merge and split:** Combine PDFs into one file or write one file per page or range.
- **Office to PDF:** Word, Excel, PowerPoint and OpenDocument files convert to PDF.
- **OCR for scans:** Scanned pages are read with Tesseract OCR.
- **Many client configs:** Documented setup for Claude Code, Claude Desktop, Codex and Cursor.

## Good for

- Turn a report with tables and charts into Markdown for an agent
- Split a long PDF into single pages and number them
- Merge several PDFs into one deliverable
- Fill or compare PDF forms from an assistant
- Convert Office documents to PDF without uploading them

## Install

### Claude Code (Terminal)

```sh
claude mcp add superpdf -- npx -y @superpdf/cli
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "superpdf": {
      "command": "npx",
      "args": [
        "-y",
        "@superpdf/cli"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "superpdf": {
      "command": "npx",
      "args": [
        "-y",
        "@superpdf/cli"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "superpdf": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@superpdf/cli"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "superpdf": {
      "command": "npx",
      "args": [
        "-y",
        "@superpdf/cli"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "superpdf": {
      "command": "npx",
      "args": [
        "-y",
        "@superpdf/cli"
      ]
    }
  }
}
```

## Tools

- `pdf_list_files`: List PDFs in an allowed folder
- `pdf_info`: Page count, sizes, form fields and properties
- `pdf_read_text`: Embedded text by page, with a character limit
- `pdf_extract_structure`: Markdown, HTML or JSON with headings, paragraphs and a section outline
- `pdf_merge`: Combine two or more PDFs into a new file
- `pdf_split`: One new file per range or per page
- `pdf_from_office`: Convert Word, Excel, PowerPoint or OpenDocument files to PDF

## FAQ

### Are my files uploaded?

No. superPDF runs on your computer and uploads nothing, though text your agent reads is passed to its model.

### What do I need to install it?

Node.js 22 or later and the npm package @superpdf/cli.

### Does it use a language model to read PDFs?

The README says the reader works by rules, not a language model, and uses Tesseract for scans.

### How do I limit what the agent can see?

Start the server with --allow-dir pointing at a folder you chose.

### Is superPDF AI free?

Yes. superPDF AI is open source, so you can use and self-host it for free.

### How do I add superPDF AI to Claude Code?

Run this command in your terminal: claude mcp add superpdf -- npx -y @superpdf/cli

### What are the best alternatives to superPDF AI?

The closest alternatives to superPDF AI are ASTra, CARTESIAN ECU, Invoices, SigPen and ConvertAudioToText.

## Alternatives

- [ASTra](https://ai-powered.com/mcp/astra): ASTra is a freemium MCP server for files and documents.
- [CARTESIAN ECU](https://ai-powered.com/mcp/cartesian-ecu): CARTESIAN ECU is a paid MCP server for files and documents.
- [Invoices](https://ai-powered.com/mcp/invoices): Invoices is a paid MCP server for files and documents.
- [SigPen](https://ai-powered.com/mcp/sigpen): SigPen is a freemium MCP server for files and documents.
- [ConvertAudioToText](https://ai-powered.com/mcp/convertaudiototext): ConvertAudioToText is a freemium MCP server for design and media.
- [WhoScribe](https://ai-powered.com/mcp/whoscribe): WhoScribe is a freemium MCP server for design and media.
- [filex](https://ai-powered.com/mcp/filex): filex is an open-source MCP server for files and documents.
- [DocJacket](https://ai-powered.com/mcp/docjacket): DocJacket is a freemium MCP server for productivity.

---
Source: https://ai-powered.com/mcp/superpdf-ai
