What problem does it solve? Turning a written implementation plan into working code requires coordinating multiple tasks, enforcing prerequisites, and tracking evidence across sessions. This Skill executes the task checklist in plan.md through specialist subagent archetypes (system-architect, core-implementer, quality-guardian) with deterministic readiness gates, write-set isolation, and resumable state tracking, so multi-task implementation runs stay consistent and auditable. ## Core Features & Use Cases - Subagent-driven task execution: Dispatches plan tasks to named specialist archetypes with computed parallel groups and write-set collision detection via the DoFlow runtime. - Readiness and approval gates: Validates task contracts (READY, NEEDS_EVIDENCE, NEEDS_USER_DECISION, BLOCKED) and resolves workflow approval gates before any source mutation. - Evidence and state tracking: Batches per-phase evidence and claims, checkpoints progress in state.md, and records stage handoffs for cross-session resumability. - Scaffold generation: With --scaffold, emits a reviewable signature-only code scaffold under the feature directory without touching the source tree. - Use Case: After running /do-brainstorm, /do-design, and /do-plan, invoke /do-execute-plan --scope phase:2 to build phase 2's tasks through subagents while the runtime enforces the feature readiness contract. ## Quick Start Ask the AI to execute the plan for the current feature, for example by saying "run /do-execute-plan --scope next to start building the next pending task from plan.md".
