execute-build

name	execute-build
description	Use after the plan is signed off, to drive the autonomous, plan-gated build to completion. Orchestrates per-task implementation, review, and verification in customer-experience-first order, stopping only on genuine blockers.

Execute Build

Announce at start: "I'm using claude-engineer:execute-build to drive the autonomous build."

Iron Law

Only run after the user signed off on the plan (claude-engineer:plan-project). If there is no signed-off plan in docs/plan/, stop and go back.

The drive loop

1. Read state first (.claude-engineer/memory.md, progress.md, current-phase.txt). Trust them over recollection.
2. Isolate the workspace via superpowers:using-git-worktrees before writing code.
3. Work phases in CX-first order (from the plan): Phase 0 scaffold → 1 design system → 2 frontend → 3 backend → 4 integration → 5 quality/CI/finish. Invoke the matching claude-engineer domain skill for each phase:
   • project-conventions, monorepo-and-tooling, docker-dev-environment (scaffold)
   • design-system → frontend-development → backend-development
   • testing-and-quality, ci-cd-pipeline
4. Execute tasks with superpowers:subagent-driven-development - a fresh implementer subagent per task; each task uses superpowers:test-driven-development. Use superpowers:dispatching-parallel-agents for independent workstreams. Fall back to superpowers:executing-plans only where subagents are unavailable.
5. After each task: update .claude-engineer/progress.md; record decisions in memory.md.
6. When something breaks: superpowers:systematic-debugging (root cause before fixes).
7. Review gate per phase / major feature: superpowers:requesting-code-review + superpowers:receiving-code-review.
8. Before claiming any phase "done": superpowers:verification-before-completion AND run the four gates via claude-engineer:testing-and-quality.
9. Finish: claude-engineer:ci-cd-pipeline then superpowers:finishing-a-development-branch.

Autonomy & stopping

Plan-gated then autonomous: run to completion without check-ins. Stop only on genuine blockers - ambiguous requirements not covered by the brief, a destructive/irreversible action, missing credentials, or a decision that changes the agreed plan. Surface the blocker crisply with options.

Red flags - STOP

• Building before reading the ledger → re-read it.
• Re-doing a task already marked done in progress.md.
• Marking a phase done without running the four gates.
• Changing the plan unilaterally → that's a blocker; ask.