executing-plans

Batch execution breaks your implementation plan into controlled groups (typically 3 tasks at a time), running each batch with verification gates—tests, typecheck, lint—followed by automated progress reports and explicit review prompts before proceeding to the next batch.

Use controlled batch execution with mandatory review checkpoints after each group of tasks. This approach prevents costly deviations by automatically reporting verification results and prompting for human feedback before commits, ensuring continuous alignment throughout implementation.

Yes. The skill tracks per-task state (in_progress, completed) and applies verification gates including tests, typecheck, and lint after each batch, reporting results before requesting human review and approval for the next batch.

Regular review checkpoints after each batch catch misalignment early. Automated verification ensures code quality gates pass before commits, and explicit human review prompts between batches create controlled decision points that reduce costly rework downstream.

Use batching for complex feature plans where continuous alignment matters, verification gates must pass between steps, and human review is necessary before proceeding. Batching suits development, integration, and deployment contexts where staged progress prevents cascading failures.