Problem
The `software-kb` skill guides agents through the kanban workflow (claim, work, submit, review). Now that `sw_claim` and `sw_transition` return gate results (DoR/DoD criteria), the skill needs to teach agents how to interpret and act on them.
Currently agents won't know to:
Check gate results after claiming an item
Fill DoR gaps (add acceptance criteria, link to ADR, estimate effort) before starting work
Verify DoD items (tests passing, KB docs updated) before transitioning to doneScope
Update the `software-kb` skill prompt to explain gate results in `sw_claim` / `sw_transition` responses
Add guidance for handling `judgment` criteria (self-evaluate) vs `checker` failures (fix before proceeding if enforce policy)
Add guidance for the `warn` vs `enforce` policy distinction
Include example gate response in the skill so agents know the shape of the data
Document the DoR self-check loop: claim → read gate → fill gaps → proceedAcceptance Criteria
Agent following the skill correctly interprets gate results after `sw_claim`
Agent addresses judgment items proactively (e.g., adds ## Acceptance Criteria section)
Agent does not get stuck on warn-policy failures