task-decomposition

To decompose a feature, break the high-level objective into atomic, verifiable tasks with explicit preconditions and a dependency graph. This enforces task atomicity and splits large changes to keep scope manageable for Python and React development.

Task atomicity means breaking complex objectives into small, independently testable units. Each atomic task lists explicit preconditions and a concrete 'Done when' verification, ensuring deterministic validation and manageable scope during feature implementation.

You can track planning state by generating persistent task files, specifically task_plan.md and progress.md. These files record task state and execution progress, allowing you to maintain context and resume planning across different sessions.

Yes, generating a dependency graph is a core function of this task decomposition approach. It maps task preconditions and outputs a text-based dependency graph that guides the deterministic execution order for cross-team planning.

The best way to define verification criteria is to specify a concrete 'Done when' condition for every atomic task during decomposition. This ensures deterministic validation and confirms that each small, testable task is fully complete before moving forward.

You split large changes by enforcing size-aware planning that limits task scope during decomposition. This approach automatically breaks oversized objectives into smaller, atomic tasks to ensure they remain testable and verifiable throughout implementation.