What problem does it solve?
This Skill prevents reviewers from accepting incomplete or watered-down pull requests by requiring the PR to match the plan exactly and treating any in-scope undone work as a hard blocker.
Core Features & Use Cases
- Plan-to-PR execution grading: Buckets findings into in-scope/done, in-scope/undone (blocking), and cosmetic/no-value (silently ignored).
- Strict blocker discipline: Marks every in-scope + undone item as ALWAYS BLOCKING with a concrete “what to fix” payload sent back to the worker.
- Evidence-anchored review: Requires quoting or referencing the subtask/annotation and checking PR diffs or PR URL evidence rather than judging code quality or intent.
Use case: After a worker generates a PR from a detailed multi-repo task plan, use this Skill to verify that every explicit ask in the plan was actually delivered, and to immediately send any gaps back to the worker as blockers so pipeline iterations don’t leak scope.
Quick Start
Run the review against the task’s plan by calling ttal go for the planning hex and then sending NEED_WORK blockers back to the worker with ttal send based on the Category 2 items.