Task quality review before development

★ 6.9 · vibe-coding

breakdown-review-tasks is a Claude Code skill that validates generated task files for quality and completeness before a development layer is marked as done. Invoked automatically by the /breakdown skill after task generation, it parses XML task files against seven critical criteria: presence of all required sections, inline self-containment (no references to external docs), complete interface contracts with type annotations and imports, specific requirements with exact file paths and class names, concrete test values and assertions, runnable verification commands, and a maximum of three files per task scope. A single failure on any critical criterion blocks the entire layer from proceeding. Results are written to a `{layer}_review.json` file containing per-task pass/fail details, issue locations, and warnings, along with a final PASSED or FAILED verdict that gates pipeline continuation.