runlisted

Take the approved plan all the way: $ARGUMENTS (plan file path) Stop only at three places: an escalation question, a fix loop exceeding 3 rounds, and merge approval (when the policy resolves to local/manual). Otherwise proceed without asking the CEO. 1. Run `build` via the Skill tool (argument: $ARGUMENTS). Do not continue until it finishes. 2. Run `review` via the Skill tool (argument: base branch, normally main). 3. Run `ship` via the Skill tool (argument: plan slug). If a skill's stop condition triggers, stop there and report what is blocked. When all three finish, report once in the final report format (what / how verified / remaining risks / decisions needed).