Nutrient Document Engine

The Nutrient Document Engine MCP Server connects your AI agents, like Claude or Cursor, directly to your documents. It lets you perform complex document processing tasks using natural language.

Instead of writing scripts for things like data extraction or redaction, you just tell your AI what to do. You can use it to pull text from a PDF, split a large contract into sections, or remove sensitive information from tax forms, all through a simple conversation.

Features List

• Content Access: Extract text, search through documents, render pages as images, and pull out form data or tables.
• Document Editing: Split documents into smaller files, add watermarks, make copies, and fill out PDF forms automatically.
• Security: Find and redact sensitive data like social security numbers or credit card information using pre-built patterns or your own custom rules.
• Annotations: Add, read, or remove document annotations such as highlights, sticky notes, and stamps.
• System Management: Discover what documents are available, look up file metadata, and check the server’s operational status.

Use Cases

• Automated Invoice Processing: An AI agent can scan a folder of invoices, extract details like names, invoice numbers, and amounts, and then organize that data.
• Secure Document Sharing: A legal aide can ask an AI to redact all personal information from a batch of contracts before sharing them. The server handles the redaction and prepares clean versions of the files.
• Interactive Report Analysis: You can upload a lengthy report and ask your AI to summarize a specific section or extract a particular data table. The server retrieves just the information you need, right when you need it.
• Batch PDF Form Filling: A developer can set up a workflow where an AI agent takes customer data from a database and uses the server to populate hundreds of PDF registration forms automatically.

How to Use It

1. Clone the Nutrient Document Engine MCP server from GitHub and start the engine:

git clone https://github.com/PSPDFKit/nutrient-document-engine-mcp.git
cd nutrient-document-engine-mcp
docker-compose up -d

2. Tell Claude Desktop where to find the MCP server. Go to Settings → Developer → Edit Config in the Claude app and add the following JSON configuration:

{
  "mcpServers": {
    "nutrient-document-engine": {
      "command": "npx",
      "args": ["-y", "@nutrient-sdk/document-engine-mcp-server"],
      "env": {
        "DASHBOARD_USERNAME": "admin",
        "DASHBOARD_PASSWORD": "password",
        "DOCUMENT_ENGINE_BASE_URL": "http://localhost:5000",
        "DOCUMENT_ENGINE_API_AUTH_TOKEN": "secret"
      }
    }
  }
}

3. Close and reopen Claude Desktop for the new configuration to take effect.

4. Upload and Process Documents

1. Open a web browser and navigate to http://localhost:5100/dashboard.
2. Drag and drop the documents you want to work with into the dashboard.
3. Go back to Claude Desktop and start asking it to perform tasks, like “List my current documents and extract the text from the contract file.”

FAQs

Q: Do I need a paid license to try this?
A: No, the server runs in an evaluation mode by default. This version has some limitations and will place watermarks on documents. To remove these, you need to purchase a license from Nutrient.

Q: Where are my documents stored and processed?
A: Your documents are handled locally. The Nutrient Document Engine runs in a Docker container on your own machine, and the dashboard for uploading files is also hosted locally.

Q: What kind of information can it redact automatically?
A: It comes with preset patterns for common sensitive data like SSNs and credit card numbers. You can also create your own custom redaction rules using regular expressions (regex) for more specific needs.