What problem does it solve?
This skill automates the safe execution of written implementation plans so agents can carry out development work in controlled, reviewable batches rather than making ad-hoc edits or skipping steps.
Core Features & Use Cases
- Task bootstrap & verification: Load an existing .tasks.json or parse a plan file to create and verify native tasks and dependency relationships before any work begins.
- Batched execution with checkpoints: Execute tasks in small batches (default first 3), mark progress, run verifications, and stop for human feedback between batches.
- Safe workflow integration: Enforces dependency relationships, updates task status, persists .tasks.json changes, and references required follow-up skills for branch finishing.
- Use Case: Implement a multi-step feature from a saved plan while preserving dependency order, running tests, and pausing for architect review after each batch.
Quick Start
Ask the agent to run the executing-plans skill on the saved plan at ./plans/feature-plan.md to execute the first batch and report results for review.