概要
The validation-task-artifacts skill acts as a quality assurance gate within the human-in-loop multi-agent framework. It analyzes task mappings and implementation files to identify gaps in planning, such as missing vertical slices, incorrect TDD ordering, or broken traceability between user stories and specific tasks. By classifying issues by severity and providing actionable feedback, it ensures that Claude Code has a solid, testable roadmap before execution begins, preventing architectural debt and missed requirements.