/speckit.clarify — “Clarify underspecified areas.” Through iterative dialogue the AI asks clarifying questions while research agents investigate technical context, maturing the vague specification: edge cases and boundary conditions, acceptance-criteria precision, removal of the [NEEDS CLARIFICATION] markers left by speckit-specify, and identification of organizational constraints (database standards, auth requirements, deployment policies). It creates no distinct artifact — spec.md evolves in place.

The ordering inversion

clarify is Spec Kit’s node in the wiki’s grilling cluster (pattern-grilling) — interrogate the human until decisions resolve — alongside:

Note the ordering difference: those frameworks grill before producing the spec, whereas Spec Kit drafts the spec first (speckit-specify) with ambiguities flagged, then grills to resolve them. Same generic activity (stage-align), spec-first sequencing.

See Also