OAuth
(no NUTRIENTDWSAPIKEY set): the same browser-flow token used by every other tool also covers Data Extraction. No extra configuration is needed.
How to use it
OAuth is exposed by the Nutrient Dws MCP Server MCP server. Add the server to your MCP client (Claude Desktop, Cursor, Windsurf and others), and the OAuth tool becomes available to the model automatically. See the full listing for setup details and every tool this server provides.
Install Nutrient Dws MCP Server
npx @nutrient-sdk/dws-mcp-serverOther tools in Nutrient Dws MCP Server (16)
Spatial or Markdown
AI redaction for detecting and permanently removing sensitive content such as names, addresses, SSNs, emails, and custom criteria
Read-only account lookup for current DWS credits and usage. No document content is uploaded
Read-only view of local files when sandbox mode is disabled. Sandbox mode is strongly recommended
Document processing for conversions, OCR, watermarking, rotation, annotation flattening, and redaction workflows
PDF signing with CMS / PKCS#7 and CAdES signatures plus visible or invisible appearance options
Schema-guided field extraction (DWS Data Extraction API): pulls specific named fields into a JSON shape you define, with per-field citations
(server default true), strict (default false), and multimodal (default false, increases cost/latency) are only sent when you set them explicitly.
(free text, up to 10000 characters) adds guidance for ambiguous fields. language/maxLanguages/maxScripts tune OCR the same way as parsedocument.
OCR language(s) for structure/understand/agentic modes; left unset, the API auto-detects. maxLanguages / maxScripts cap how many languages/scripts auto-detection considers, and only apply when language is left unset.
runs the parse feeding the extraction: structure (1.5 credits/page), understand (default, 9 credits/page), or agentic (18 credits/page) — no text mode, since schema-guided extraction needs the structural parse text mode skips. Total cost per page is that parse component plus a fixed extract componen
is always returned inline, pretty-printed — it is the answer, bounded by your own schema. Alongside it, a citation match summary reports how each field was grounded (idmatch, idmatchmultiblock, idmatchpartial, fuzzymatch, notfound) and lists which field paths came back notfound (capped at 10, then "
Structured data extraction (DWS Data Extraction API): typed JSON elements with bounding boxes and confidence, or whole-document Markdown
Read-only view of files inside the configured sandbox directory
Spatial or Markdown
Markdown only