What problem does it solve?
Mandatory execution discipline protocol for AI agent workflows. Use this skill whenever an agent needs to plan work, track progress, verify completion, or handle execution drift. Provides the standard operating procedures for: (1) Plan Mode Default — creating explicit plans/checklists before non-trivial tasks (2+ steps or architectural decisions), (2) Replan on Drift — stopping and reassessing when execution diverges from plan, (3) Verification Before Done — never marking work complete without evidence (tests, logs, diffs, rubric checks), (4) Track Progress — marking checklist items complete as work progresses, (5) Demand Elegance — preferring the simplest robust design and avoiding over-engineering. Keywords: plan mode, replan, drift, verification, evidence, checklist, track progress, elegance, non-trivial task, planning discipline.
Core Features & Use Cases
- Explicit planning and checklists before complex work to ensure clarity and accountability.
- Drift detection and re-planning mechanisms to recover from deviations.
- Verification-before-done with evidence requirements (tests, logs, diffs, rubrics) to prevent premature completion.
- Progress tracking with structured checkpoints and decision logs.
- Encourages elegant, minimal-impact designs that balance rigor with practicality.
Quick Start
Always create an explicit plan or checklist before starting any non-trivial task.
