What problem does it solve? Multi-file code changes often fail due to uncoordinated edits, skipped verification, and untested commits. This Skill enforces a structured workflow that plans, parallelizes, verifies, and commits multi-file implementations without dropping planned work. ## Core Features & Use Cases - Tiered Model Assignment: Routes orchestration to Opus, implementation to Sonnet, and mechanical edits to Haiku to reduce cost while preserving judgment quality. - Wave-Based Parallel Execution: Groups independent tasks into waves dispatched in a single message, with mandatory verification gates between waves. - Mandatory Verification Gates: Enforces architecture analysis (Phase 0), scope validation for epics (Phase 0½), wave gates, and a test-then-stage-then-commit sequence (Phase 3). - Use Case: When implementing a feature spanning ten files across services and tests, the Skill dispatches a code-architect agent first, groups tasks into dependency-ordered waves, runs each task's verify command at wave gates, and blocks commits until typecheck, lint, and tests pass. ## Quick Start Ask the AI to plan and implement a multi-file change using the parallel-implementation workflow with wave gates and tiered model assignment.
