Parser for .NET Parser for Java Parser for Node.js Parser for Python
Parser for AI Agents (MCP)

Document data extraction for AI agents via MCP server

Ask Claude, Cursor or Copilot to pull text, tables, images and barcodes out of PDF, Word, Excel and email files. Runs in Docker on your machine, and nothing is uploaded.
Docker Image Start Free Trial
One-click install: VS CodeCursorAll clients
Version 26.9.0  released
See what’s new

Extract invoice tables with an AI agent

More examples >
{
  "mcpServers": {
    "groupdocs-parser": {
      "command": "docker",
      "args": [
        "run", "--rm", "-i",
        "-v", "/path/to/documents:/data",
        "ghcr.io/groupdocs-parser/parser-net-mcp"
      ]
    }
  }
}
Then ask: Extract the tables from invoiceA.pdf as JSON and total the amount column
docker pull ghcr.io/groupdocs-parser/parser-net-mcp

GroupDocs.Parser MCP server at a glance

An MCP server that gives AI agents the GroupDocs.Parser engine as tools they can call

Illustration parser

Structured data the agent can trust

GroupDocs.Parser reads the document’s own text layer, tables and barcodes, so the agent works with exact values instead of guessing from a screenshot. Tables come back as Markdown or JSON.

Local by design

Your AI client starts the server as a local process over stdio. It opens no ports, sends no telemetry, and reads and writes only the folder you configure, so contracts, records and internal documents never leave your machine.

Works with every MCP client

Use it from Claude Desktop, Claude Code, VS Code with GitHub Copilot, Visual Studio 2022, Cursor, Windsurf, Cline, JetBrains Rider and Codex CLI. One-click installs and ready-made configurations are available for each client.

Runs where your agent runs

GroupDocs.Parser MCP server runs in Docker on Windows, Linux and macOS, with every native dependency in the image

Docker
VS Code
Windows
Linux
macOS
Azure
Amazon

Docker image

  • Nothing to install besides Docker
  • linux/amd64 and linux/arm64, native dependencies bundled
Get the image

MCP Registry

  • Listed in the official MCP Registry
  • Discoverable by clients that browse the registry
View the listing

VS Code and GitHub Copilot

  • One-click install with a pre-filled configuration
  • Set your documents folder after install
Install in VS Code

Cursor

  • One-click install with a pre-filled configuration
  • Runs in evaluation mode until you add a license
Add to Cursor

Supported file formats

GroupDocs.Parser MCP server works with the following file formats.

Microsoft Office formats

  • Word: DOCX, DOC, DOCM, DOT, DOTX, DOTM, RTF
  • Excel: XLSX, XLS, XLSM, XLSB, XLTM, XLT, XLTM, XLTX, XLAM, SXC, SpreadsheetML
  • PowerPoint: PPT, PPTX, PPS, PPSX, PPSM, POT, POTM, POTX, PPTM

Images & Other Formats

  • Portable: PDF
  • Images: JPG, BMP, PNG, TIFF, GIF
  • Other office formats: ODT, OTT, OTS, ODS, ODP, OTP, ODG

Other formats

  • Web: HTML, MHTML
  • Archives: ZIP, TAR, 7Z
  • e-Books: CHM, EPUB, FB2, MOBI

What the MCP server can do

Seven tools that turn documents into text, tables and values your agent can work with

Feature icon

Extract text

The extract_text tool returns plain text for the whole document or page by page.

Feature icon

Extract tables

The extract_tables tool returns tables as Markdown or JSON, ready to total, filter or load.

Feature icon

Extract images

The extract_images tool saves embedded images as separate files.

Feature icon

Read barcodes and QR codes

The extract_barcodes tool decodes Code128, QR, PDF417, DataMatrix, EAN, UPC, Aztec and more.

Feature icon

Extract metadata

The extract_metadata tool returns author, title, dates, EXIF, XMP and IPTC as JSON.

Feature icon

Email and 50+ formats

Works with PDF, DOCX, XLSX, PPTX, HTML, EPUB, EML, MSG and 50+ more formats.

Feature icon

Check before extracting

The get_document_info tool reports type and page count, so the agent knows what to expect.

Feature icon

Whole folders

Extract the same fields from every document in a folder and get one consolidated table.

Feature icon

Flexible licensing

Start in evaluation mode, then use a license file or metered keys. The get_license_status tool shows which one is active.

What you can ask your agent

Typical GroupDocs.Parser MCP scenarios, each with the prompt that drives it

Extract tables into structured data

Invoices, statements and reports hold their data in tables. The agent extracts them as JSON and works with exact numbers, not with text it has to reparse. Step by step: guide and blog walkthrough.

Prompt

Extract the line items from invoice.pdf as JSON
and total the amount column

Read barcodes and QR codes from scans

Delivery notes and labels carry their key in a barcode. The agent decodes every code and matches it to the file and page it came from. Step by step: guide and blog walkthrough.

Prompt

Read the barcodes in these scanned delivery notes
and list the values with the file they came from

Extract data from a whole folder

Run the same extraction across a folder and get one consolidated answer, with a check first for files that have no text layer. Step by step: guide and blog walkthrough.

Prompt

For every PDF in my documents folder, extract the tables as JSON,
total the amount column, and give me one table of file, total and page count

Run on-premise with Docker

For regulated environments, run the server inside your perimeter. Mount the documents folder and the license, and pin the image version for change control. Architecture and security model: on-premise deployment.

docker-compose.yml

services:
  groupdocs-parser:
    image: ghcr.io/groupdocs-parser/parser-net-mcp:26.9.0
    volumes:
      - "/srv/documents:/data"
      - "/srv/licenses:/license:ro"
    environment:
      GROUPDOCS_MCP_STORAGE_PATH: /data
      GROUPDOCS_LICENSE_PATH: /license/GroupDocs.Total.lic
    stdin_open: true

Install in your AI client

Copy the configuration for your client, or use the one-click links at the top of the page

Register the server

Every client starts the same container. Mount the folder with your documents at /data, then restart the client. Add -e GROUPDOCS_LICENSE_PATH and a license mount to lift evaluation limits. More clients and options: installation guide.
Claude Desktop Claude Code VS Code Cursor Docker
{
  "mcpServers": {
    "groupdocs-parser": {
      "command": "docker",
      "args": ["run", "--rm", "-i", "-v", "/path/to/documents:/data", "ghcr.io/groupdocs-parser/parser-net-mcp:latest"]
    }
  }
}

claude mcp add groupdocs-parser -- \
  docker run --rm -i -v /path/to/documents:/data ghcr.io/groupdocs-parser/parser-net-mcp:latest

{
  "servers": {
    "groupdocs-parser": {
      "type": "stdio",
      "command": "docker",
      "args": ["run", "--rm", "-i", "-v", "/path/to/documents:/data", "ghcr.io/groupdocs-parser/parser-net-mcp:latest"]
    }
  }
}

{
  "mcpServers": {
    "groupdocs-parser": {
      "command": "docker",
      "args": ["run", "--rm", "-i", "-v", "/path/to/documents:/data", "ghcr.io/groupdocs-parser/parser-net-mcp:latest"]
    }
  }
}

docker run --rm -i \
  -v /path/to/documents:/data \
  ghcr.io/groupdocs-parser/parser-net-mcp:latest

Ready to get started?

Download GroupDocs.Parser for free or get a trial license for full access!

Useful resources

Explore documentation, code samples, and community support to enhance your experience.

Documentation
API reference
Code samples
Free support
Paid support

Temporary license tips

1
Sign up with your work email.
Free mail services are not allowed.
2
Use Get a temporary license button on the second step.
Go to license
✕