# PSPDFKit/nutrient-dws-mcp-server

**Category:** 🛠️ Other Tools and Integrations  
**Repository:** https://github.com/PSPDFKit/nutrient-dws-mcp-server  
**GitHub Stars:** 71  
**npm Downloads (last month):** 396  
**Views:** 5  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/pspdfkit-nutrient-dws-mcp-server

## Description
MCP server for the Nutrient DWS Processor API. Convert, merge, redact, sign, OCR, watermark, and extract data from PDFs and Office documents via natural language. Works with Claude Desktop, LangGraph, and OpenAI Agents.

## Tools
Capabilities this server exposes over MCP:

- **document_processor** — Document processing for conversions, OCR, watermarking, rotation, annotation flattening, and redaction workflows
- **parse_document** — Structured data extraction (DWS Data Extraction API): typed JSON elements with bounding boxes and confidence, or whole-document Markdown
- **extract_fields** — Schema-guided field extraction (DWS Data Extraction API): pulls specific named fields into a JSON shape you define, with per-field citations
- **document_signer** — PDF signing with CMS / PKCS#7 and CAdES signatures plus visible or invisible appearance options
- **ai_redactor** — AI redaction for detecting and permanently removing sensitive content such as names, addresses, SSNs, emails, and custom criteria
- **check_credits** — Read-only account lookup for current DWS credits and usage. No document content is uploaded
- **sandbox_file_tree** — Read-only view of files inside the configured sandbox directory
- **directory_tree** — Read-only view of local files when sandbox mode is disabled. Sandbox mode is strongly recommended

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "nutrient-dws-mcp-server": {
    "command": "npx",
    "args": ["-y","@nutrient-sdk/dws-mcp-server"],
    "env": {
      "SANDBOX_PATH": "",
      "NUTRIENT_DWS_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `SANDBOX_PATH`, `NUTRIENT_DWS_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What PSPDFKit/nutrient-dws-mcp-server MCP server does

PSPDFKit/nutrient-dws-mcp-server MCP server exposes Nutrient DWS document-processing capabilities to MCP-compatible AI clients. It connects to the Nutrient Document Web Service Processor API for document transformations and the Data Extraction API for parsing and field extraction. The server is intended for workflows where an agent needs to work with PDFs or Office documents using natural-language requests.

Supported workflows include conversion, OCR, watermarking, rotation, annotation flattening, redaction, document merging, and PDF signing. Extraction can produce whole-document Markdown, typed JSON with bounding boxes and confidence values, or a JSON structure containing named fields and per-field citations.

## How it works

The package runs as a local stdio MCP server. An AI client invokes one of its tools, and the server reads input files from the configured local directory before calling the relevant Nutrient DWS API. Results can be written to explicit output paths or returned as extracted content, depending on the operation.

The first request that uses the Nutrient API can start a browser-based OAuth sign-in. API-key authentication is also available for CI, headless environments, and scripted use. Data extraction offers four parse modes: `text`, `structure`, `understand`, and `agentic`; the README identifies `understand` as the default and lists different credit costs for each mode.

## Setup and configuration

Install the package through an MCP client configuration with `npx -y @nutrient-sdk/dws-mcp-server`, or install it from Claude Desktop Extensions. Set `SANDBOX_PATH` to the directory containing documents the server may access. Sandbox mode is recommended because it limits file handling to the configured directory. When sandbox mode is disabled, the read-only `directory_tree` tool can inspect local files visible to the user account.

Create a Nutrient account before using the DWS APIs. For interactive desktop use, the browser sign-in flow can provide authentication without an API key. For non-interactive operation, set `NUTRIENT_DWS_API_KEY` with a key created in the Nutrient dashboard. Restart the MCP client after changing its configuration, then use explicit file names or paths in requests.

The README provides configurations for Claude Desktop, Cursor, Windsurf, and VS Code, and states that other MCP-compatible clients can use stdio transport.

## Tools and capabilities

PSPDFKit/nutrient-dws-mcp-server MCP server provides these tools:

- `document_processor` handles conversions, OCR, watermarks, rotation, annotation flattening, and redaction workflows.
- `parse_document` returns Markdown or structured extraction data with spatial coordinates and confidence values.
- `extract_fields` extracts named fields into a developer-defined JSON shape and can include citations for each field.
- `document_signer` applies CMS, PKCS#7, or CAdES PDF signatures with visible or invisible appearances.
- `ai_redactor` detects and permanently removes sensitive content, including common personal data and custom criteria.
- `check_credits` reads DWS credit and usage information without uploading document content.
- `sandbox_file_tree` lists files in the configured sandbox directory.
- `directory_tree` lists local files when sandbox restrictions are disabled.

The package also documents prompts for chained tasks such as signing after watermarking, extracting fields, redacting personally identifiable information, parsing documents for retrieval, and converting Office files to PDF/A.

## Limitations and notes

Document-processing and extraction requests depend on Nutrient DWS access and account credits. The `check_credits` tool is read-only and does not upload document content, but other tools may send documents to the corresponding DWS APIs. API-key authentication is optional for interactive OAuth use but needed for the documented CI and headless fallback. Sandbox mode is strongly recommended, and explicit paths are safer than vague file-browsing instructions.

_Full upstream README: https://allmcps.com/mcp/pspdfkit-nutrient-dws-mcp-server/readme_