workflow fit Shipped
What this lens looks for
Check each designed command against the engine's workflow shapes: authoring synthesizes ONE document per invocation via a read-only agent + playbook; review (specialist-pipeline) judges a git changeset and emits findings. Flag: an authoring command whose outcome is really many documents per run (that is an orchestration/builtin shape, not v1 vocabulary); a review command over content that never lives in a repo; commands that duplicate the generator's standard report command or the shared top-level sessions inspection surface; interactive/dialogue outcomes forced into a batch workflow; and a lead mismatch (an authoring command without an author playbook, a review command without a report-writer lead).
What its verifier checks
Every command was classified against its declared outcome; each mismatch names the command, the declared kind, and the shape it actually needs; no flag is raised on a command whose kind and outcome already agree.