Super Extract — Structured Information Extraction with Source Grounding

Extract structured entities from any document (PDF, DOCX, or text) with character-level source grounding and interactive visualization. Inspired by google/langextract.

Sub-Skills

This is a workflow skill that composes:

• pdf skill — for reading and extracting text from PDF files
• docx skill — for reading and extracting text from Word documents

When to Use

• User wants to extract entities, facts, or structured data from documents (PDF, DOCX, text)
• User needs named entity recognition (NER), relation extraction, or information extraction
• User wants extracted data grounded to exact positions in the source text
• User has a document (file, URL, pasted text) and wants structured output

How It Works