What problem does it solve? Complex coding and review tasks often ship with undetected defects because a single agent both implements and verifies its own work. This Skill enforces a rigorous multi-agent workflow that separates candidate generation from validation, so findings and code changes are independently reviewed and challenged before completion is declared. ## Core Features & Use Cases - Parallel Exploration and Mapping: Fan out explorer subagents across disjoint scopes for repository mapping, security review, dependency review, and test analysis. - Independent Reviewer and Critic Gates: Validate candidate findings with skeptical reviewer subagents, then challenge confirmed findings with a separate critic before synthesis. - Mandatory Implementation Closeout Gate: Any task editing code, tests, docs, or package metadata must pass objective validation, an independent implementation reviewer, and a final critic on the latest diff before completion. - Use Case: When refactoring an authentication module, the agent parallelizes subsystem investigation, has a reviewer verify each candidate issue as CONFIRMED or DISPROVED, and blocks completion until a critic approves the final diff and evidence. ## Quick Start Enable swarm mode for this session and use parallel explorer, reviewer, and critic subagents for my next complex coding task.
