What problem does it solve?
It removes the risk of shallow or incorrect test generation by executing a previously approved QA plan with isolated subagents and mandatory two-stage review before any task is considered done.
Core Features & Use Cases
- Task-by-task subagent execution: Runs each plan task in a fresh delegated worker to prevent context pollution and ensure deterministic, scoped work.
- Two-stage review gate per task: Enforces a spec-correctness review (risk coverage and red-phase expectations) followed by a test-quality review (observability, determinism, and tautology avoidance).
- Continuous execution loop: Processes tasks continuously with parallel “waves” based on the plan’s parallel/sequential markers, avoiding mid-plan user check-ins.
- Plan integrity cross-task review: Performs a final reviewer step across all task outputs to confirm collective risk coverage and absence of coverage seams.
Quick Start
Use this skill when you already have a signed QA plan and want the system to dispatch and execute each task in isolation with spec-and-quality reviews, finishing by handing off to sumo-qa-finishing-qa-work.
