executing-plans

To execute a plan with human-in-the-loop validation, load a written plan like docs/plan.md to assess dependencies and constraints. The system coordinates staged reviews and checkpoints, ensuring auditable task execution across multiple steps within a single session.

Managing task dependencies during complex workflow execution involves automatically identifying prerequisites and environmental assumptions from your loaded plan. Execution pauses to request human confirmation when dependencies are missing or constraints are unclear.

Handling blockers and ambiguities during plan execution requires pausing the workflow to request human confirmation. This prevents misexecution by enforcing structured reviews at staged checkpoints before proceeding with dependent tasks.

Yes, you can use a markdown plan file for automated task management by loading docs/plan.md as the entry point. The system reviews dependencies, breaks down tasks into executable steps, and tracks completion status with structured logging.

After completing multi-step plan execution, the system reports progress and can launch downstream skills like finishing-a-development-branch for branch merging and release preparation, ensuring a controlled transition from execution to delivery.