create-epics

To create development epics from approved GDDs and architecture modules, this skill translates architectural ownership boundaries into one scoped epic per module. It processes GDD content and maps each epic to its system, layer, and module responsibilities.

Requirement tracing and ADR coverage validation identify governing ADRs, compute engine risk from ADRs, and flag untraced requirements needing ADRs before execution. This ensures architectural decisions map directly to generated epic boundaries.

The correct layer sequence for planning epics runs Foundation, then Core, then Feature, and finally Presentation. This dependency-safe ordering provides guidance on when to hold Feature layer work until Core layer epics stabilize.

You cannot generate user stories alongside epics in this step. The process explicitly writes EPIC.md files and an index without generating stories, requiring a separate downstream step to break approved epics into actionable stories.

Scoping and generating epics requires reading GDD summaries for scoping, loading in-scope GDDs, and gathering relevant architecture, ADR, control-manifest, tr-registry, and engine version inputs to ensure complete requirement coverage.

Presenting per-epic approvals before writing files ensures teams review and validate each module's scoped boundaries. The skill writes production/epics/[epic-slug]/EPIC.md and index.md only after you approve the generated epic definitions.