What problem does it solve?
Turning a feature plan into an executable, well-structured task list is error-prone: tasks may miss dependencies, lack test coverage, or fail to map back to the plan. This Skill automates the tasks phase of an iterative story pipeline by generating a full tasks.md from plan.md and research.md, then independently validating it for compliance.
Core Features & Use Cases
- Two-stage subagent execution: Renders two isolated subagent prompts—one calling speckit-tasks to generate tasks.md, another calling speckit-analyze to produce a compliance report (tasks-report.md).
- Content gating and verdict routing: Checks task format, user-story slicing, TDD coverage, dependency ordering, and plan completeness; routes pass / needs_fix / plan_insufficient / spec_insufficient verdicts to retry or rollback workflows.
- Reentrant retry protocol: Supports up to 3 rounds per attempt and rollback re-entry to plan or specify phases, with meta.yaml state tracking and append-only process.log.
- Use Case: During a TAPD story pipeline, after the plan phase completes, invoke this Skill to produce a validated tasks.md where every user story has independently testable, dependency-ordered tasks before implementation begins.
Quick Start
Run the tasks phase for the current story to generate tasks.md from plan.md and research.md, then validate it with the analyze report.
