requirements-ingest

To extract and chunk requirements from a PDF, this skill parses the document and transforms scattered text into atomic, traceable chunks. It outputs a structured JSON schema containing unique IDs, location hints, and confidence scores for each requirement.

Atomic requirement chunking breaks down large specifications into standalone units, each with a unique ID and source location reference. This granularity ensures full traceability, allowing teams to audit and track individual requirements back to their original documents.

Yes, you can ingest requirements from Markdown and DOCX files, as well as PDFs and plain text. The skill processes these varied input formats to classify text and extract a structured JSON schema with project IDs and glossary suspects.

You can build a glossary from domain terms using the glossary extraction feature during requirement ingestion. The skill automatically identifies and isolates domain terminology, returning them as glossary_suspects within the final JSON schema for terminology management.

The best way to classify scattered plain text requirements is to ingest them through an automated chunking workflow that assigns per-chunk IDs and confidence scores. This process consolidates disparate texts into a traceable, structured JSON format for rigorous management.

Yes, the requirements extraction method supports confidence scoring for parsed chunks. The workflow evaluates each atomic chunk and includes a confidence score within the structured JSON output to help you assess extraction reliability.