nutrient-dws

Nutrient DWS MCP Server exposes a programmable interface to the Nutrient Document Web Service Processor API, enabling AI agents to create, edit, convert, redact, sign, OCR, and otherwise manipulate documents via natural language instructions. After configuring an MCP client to connect to the nutrient-dws server, you can send commands through the tools exposed by the server, such as document_processor for general processing tasks, document_signer for digital signatures, and sandbox_file_tree or directory_tree for navigating files. The server handles tasks like merging PDFs, converting between formats, applying OCR, redaction of sensitive data, and applying digital signatures in a standards-compliant way. You can drive complex workflows by combining multiple operations in sequence, enabling AI agents to produce finished documents with minimal human intervention.

To use it, set up your MCP client with the appropriate environment variables (API key and sandbox path), then issue natural-language style instructions that map to the available tools. For example, you might request merging several PDFs, redacting PII, and exporting to Markdown, or you might instruct the signer to apply a visible PAdES signature on a specific page. The server returns results and file paths for the produced documents, which you can continue to process or store as needed.

Prerequisites:

• Node.js installed on your machine (for npx usage)
• Access to a Nutrient API key (NUTRIENT_DWS_API_KEY)

Installation steps:

1. Install a compatible Node.js environment if you don’t already have one.
2. Obtain a Nutrient API key from the Nutrient dashboard and save it where your MCP client can read it (or set it in the environment).
3. Start the MCP server using npx with the documented package:

export NUTRIENT_DWS_API_KEY=your_api_key_here
export SANDBOX_PATH=/path/to/your/sandbox
npx -y @nutrient-sdk/dws-mcp-server

4. Point your MCP clients to the running server configuration (see mcp_config example).
5. Verify connectivity by sending a basic processing command and checking the output in the sandbox directory.

Notes and tips:

• Sandbox mode is enabled by setting SANDBOX_PATH; relative input/output paths will be resolved within that directory. If SANDBOX_PATH is not set, the server may operate with unrestricted paths.
• Ensure your API key remains secret and is not checked into version control. Use environment variables in your deployment environment.
• The MCP server exposes a set of tools; you can use document_processor for most document editing and conversion tasks, and document_signer for digital signatures (PAdES/CAdES). Use sandbox_file_tree or directory_tree to inspect sandbox contents when needed.
• If you encounter connectivity issues, verify that the MCP client configuration points to the correct command and environment, and that the API key has the necessary permissions for the requested operations.
• For Windows users, ensure path values in SANDBOX_PATH are properly escaped.
• Restart your AI client after updating the MCP configuration so changes take effect.