cyberneticlibrary
← library

Parse documents into structured elements

unstructured-clientmcp_serversetup L2★0
Unstructured-IO/unstructured-js-client ↗
What it does

Parse and partition documents into structured elements

Best for

When you need reliable document partitioning with a managed API and want MCP integration for AI workflows.

Inputs
  • · PDF, DOCX, images, HTML web content
  • · File uploads via SDK or API
  • · Strategy parameter (Auto, Fast, Ocr, etc.)
Outputs
  • · Structured element arrays with bounding boxes
  • · Table extraction with cell contents
  • · Metadata (author, title, created date, etc.)
Requires
  • · Unstructured Platform API (unstructured.io)
  • · API key required (1000 free pages/day for 14 days)
Preconditions
  • · npm or yarn
  • · Node.js 20+
  • · API key from unstructured.io
  • · Optional: self-hosted instance (enterprise)
Failure modes
  • · Daily quota exhaustion after 1000 pages
  • · Large file timeout (>50 MB)
  • · Unsupported file formats (proprietary or encrypted PDFs)
  • · Strategy mismatch (Auto may be slower than Fast)
Trust signals
  • · Official Unstructured Platform backing
  • · TypeScript/JavaScript SDK published
  • · Workflow endpoint in development
  • · Speakeasy-generated client library