understanding-specs

Load hierarchical context by reading all specifications from product level down to the target story. This process ensures you have complete, aligned context by reading required capability, feature, and story spec files before implementation begins.

When required specification documents are missing during context validation, the process aborts immediately. This prevents any work from beginning without full understanding, ensuring complete, aligned context by loading all relevant specifications from product level down to the target work item.

PRD and TRD files are optional when loading work item context. The process reads required capability, feature, and story spec files, plus optional PRD/TRD and ADR/PDR documents only when present, generating a structured context summary with document counts and decisions.

ADRs and PDRs factor into loading constraints by providing optional architectural and product decision records. When present, they are loaded alongside required capability, feature, and story specs to validate product-wide constraints from specs/CLAUDE.md before starting work items.

The best way to verify complete context alignment is to read all specifications from product level down to the target work item. This applies to capability, feature, and story levels, loading required spec files and optional documents, then outputting a structured context summary with document counts and decisions.

You should not use hierarchical spec ingestion when required specification documents are unavailable, as the process will abort on missing files. It is designed to prevent work from beginning without full understanding by enforcing complete, aligned context from product level down to the target story.