reversa-to-do

To convert a roadmap into executable actions, load the roadmap.md file and decompose each item into single-turn atomic actions. The process generates sequential IDs, dependency mappings, and per-action file tracking, outputting a complete actions.md file for agent execution.

Task decomposition with a dependency graph breaks feature work into atomic actions linked by explicit dependencies. This mechanism ensures agent-driven implementation follows clear sequencing across phases while marking independent tasks touching different files for parallel execution.

Yes, generating an actions.md task plan requires reading .reversa/state.json for output paths and validating .reversa/active-requirements.json. The process aborts with clear guidance if these prerequisite inputs are missing, ensuring the roadmap has proper context before decomposition begins.

Parallel execution hints are determined by checking if tasks touch different files and have no explicit dependencies on each other. Tasks meeting both criteria get marked for parallel execution, enabling efficient agent-driven implementation without file conflicts or sequencing blockages.

When active requirements are missing, the task plan generation aborts with clear guidance. The validation checks .reversa/active-requirements.json before processing the roadmap, ensuring all prerequisite inputs exist and preventing incomplete action generation.

Yes, custom hooks can be applied before and after generating the actions.md plan. The Skill applies before-to-do and after-to-do hooks during the roadmap decomposition process, enabling customization of the task generation workflow and output handling.