What problem does it solve? Turning an approved specification and implementation plan into an executable, dependency-aware task breakdown is error-prone: tasks get grouped together, parallel work conflicts over shared files, and failure scenarios are forgotten. This Skill produces a validated tasks.md containing a Directed Acyclic Graph (DAG) of atomic tasks with roles, dependencies, parallelism flags, and file ownership. ## Core Features & Use Cases - DAG task generation: Derives atomic tasks (Tnnn) from spec.md, plan.md, and discovery reports, each with Papel (researcher/analyst/executor), Dependências, Paralelizável, Arquivos, and Wiki-Keywords fields. - Approval gate and validation: Aborts unless spec.md has Aprovado: true, then validates the DAG for cycles, broken dependencies, orphan tasks, and file conflicts between parallel tasks. - Phase-by-phase interaction: Proposes tasks one phase at a time with user approval before continuing, and supports LATTE coordination graphs (graph-operators, heartbeats, agent assignments) plus optional --with-memory experiential hints. - Use Case: After approving a feature spec and plan, ask the assistant to generate the execution matrix; it produces a tasks.md where Gherkin success and failure scenarios each map to separate executor tasks with disjoint file sets. ## Quick Start Generate the tasks.md execution matrix for the feature in specs/features/004-chat using the approved spec and plan.
