Quickstart: what changes for an operator (#3469)

After this mission lands, the WP approval workflow stops forcing false records and accepts natural flag names.

Issue-matrix: context-only citations no longer force a false verdict

Before: a spec.md citing a parent epic, sibling missions, or a PR #NNNN scaffolded a gating row for each, and you had to pick a false verdict (fixed / in-mission / …) to pass approval.

After:

PR #NNNN references are classified non-gating — they do not block the approved transition.

  • Context-only citations (lines marked parent, epic, see #, Follow-up:, baseline-red) and
  • Any residual or operator-recorded row can carry the new truthful verdict:

``bash spec-kitty agent issue-verdict --mission <handle> --issue 1234 \ --verdict not-applicable --actor claude \ --evidence-ref "Cited as context in spec.md; mission owes no work." ``

explicit signal, so real implementation targets are never silently skipped.

  • not-applicable is terminal — it also passes the merge/done gate; you never re-verdict it.
  • Fail-safe preserved: an unmarked bare #1234 in prose still gates — the gate only relaxes on an

move-task: natural flag names now work

# Both forms are now equivalent:
spec-kitty agent tasks move-task WP01 --to doing --actor claude --reason "starting WP01"
spec-kitty agent tasks move-task WP01 --to doing --agent  claude --note   "starting WP01"

Subtask completion: use mark-status, not checkboxes

Ticking - [x] in tasks.md does not move the for_review gate (subtask completion is event-sourced by design). Complete subtasks with:

spec-kitty agent tasks mark-status T0NN --status done --mission <handle>

The move-task guidance and issue-verdict --help now state these rules up front, and the specify/plan/tasks/analyze prompts warn early that approvals gate on issue-matrix verdicts.