create-epics

You create bounded epics by parsing GDD summaries and loading in-scope ADRs, architecture metadata, and TR registries. The process maps each architectural module to an epic with ADR coverage validation, writing EPIC.md files without creating stories.

ADR traceability cross-references ADR domains with the TR registry and engine metadata to determine governing ADRs for each epic. It detects untraced requirements lacking Accepted ADR coverage, blocking or warning before writing EPIC.md files.

Yes, you can filter epic creation by specific architectural layer or system name. The workflow supports in-scope filtering across foundation, core, feature, and presentation layers, generating one bounded epic per matching architectural module.

When generating epics, untraced technical requirements are identified by validating TR-IDs against Accepted ADR coverage. The workflow blocks or warns prior to writing, guiding you to run architecture-decision or proceed using placeholders.

The best way to structure epic definitions for production planning is generating one bounded epic per architectural module across foundation, core, feature, and presentation layers. This aligns engineering scope before story decomposition begins.

Epic generation stops before creating user stories to enforce separation between scope definition and story decomposition. It writes EPIC.md and index.md files only, ensuring architectural module boundaries and ADR governance are validated before story breakdown.