What problem does it solve? It turns an approved implementation plan into disciplined execution by loading task beads, working them in small batches, and pausing for human review between batches instead of running an entire plan unchecked. ## Core Features & Use Cases - Batch execution with checkpoints: Executes the first 3 tasks by default, then reports results and waits for feedback before continuing. - Beads-driven task tracking: Claims, closes, and blocks task beads via beads_update, beads_close, and beads_mol_ready so dependencies unblock automatically. - Critical plan review and stop conditions: Reviews the plan before starting, verifies the plan-approved gate, and stops to ask when blocked or when the plan is wrong. - Use Case: After a planning session produces an approved multi-task implementation plan, use this Skill in a fresh session to claim each task bead, implement with TDD, run verifications, and hand off to the finishing-a-development-branch workflow. ## Quick Start Ask the AI to execute the approved implementation plan for the current branch using the executing-plans skill, working tasks in batches and reporting after each batch.
