What problem does it solve? Turning a high-level feature specification into an actionable implementation plan is error-prone: tasks end up too large, acceptance criteria stay vague, dependencies get missed, and later review phases lack a clear definition of done. This Skill structures the planning phase so every epic becomes a set of small, verifiable, traceable tasks. ## Core Features & Use Cases - Task Decomposition: Breaks an epic (read via bd show) into 3-7 small, individually verifiable tasks with mapped dependencies and no circular references. - Acceptance Criteria Generation: Extracts testable criteria from EARS requirements and appends an Acceptance Criteria table to the epic description, with each requirement mapped to at least one criterion. - Verification Contract: Derives a per-epic definition of done (profile, surfaces, risks, required evidence such as test, lint, build) so review phases know exactly what evidence is required. - Research Integration: Searches memory (ca search, ca knowledge) and spawns analyst subagents to ground the plan in past decisions, ADRs, and codebase patterns. - Use Case: Given an epic describing a new CLI feature, produce beads tasks with dependencies, an AC table tracing each EARS requirement, a verification contract for the cli profile, and blocking review/compound gate tasks. ## Quick Start Ask the agent to plan the implementation of an epic by decomposing it into testable tasks with acceptance criteria and a verification contract.
