nutrient-document-processing_skill

Overview

This skill processes and automates document workflows using the Nutrient DWS Processor API. It handles conversions, OCR, extraction, redaction, watermarking, digital signing, and PDF form filling across PDFs, Office files, HTML, and common image formats. The skill is built for integration into agent pipelines or developer tools to speed up repetitive document tasks.

How this skill works

You send multipart POST requests to the Nutrient API endpoint with an instructions JSON payload describing input parts and actions. The API performs conversions, OCR, table/text extraction, redactions, watermarks, signatures, and form fills and returns the requested output format. Optionally, an MCP server can run locally to expose the same functionality for native tool integrations.

When to use it

• Convert documents between PDF, DOCX, XLSX, PPTX, HTML, and images
• Extract plain text, tables, or structured data from PDFs and images
• Perform OCR on scanned pages or non-text images
• Redact PII or patterns before sharing documents
• Add watermarks or apply batch digital signatures
• Fill PDF forms programmatically or prepopulate templates

Best practices

• Provide clear instructions JSON that lists parts and the exact output type to avoid format ambiguity
• Use preset redaction strategies for common PII (SSN, email, credit card) when possible
• Enable OCR language explicitly for best results on non-English or mixed-language documents
• Process large batches incrementally to monitor failures and reduce retry cost
• Keep API keys in environment variables or a secure secrets manager; do not hard-code keys

Example use cases

• Batch convert legacy DOCX and PPTX files to searchable PDFs for archiving
• Extract all tables from financial reports into XLSX for downstream analysis
• OCR scanned contracts in multiple languages and then redact client identifiers
• Add CONFIDENTIAL watermarks to draft reports and apply a digital signature for approval workflows
• Auto-fill employment onboarding PDFs with HR data and return flattened, signed documents