What problem does it solve?
Unvetted $deliver execution plans often contain hidden blockers, scope gaps, dependency order errors, or missing acceptance criteria that only surface during implementation, leading to wasted engineering effort and rework.
Core Features & Use Cases
- First-Principles Adversarial Council Review: Runs a structured internal adversarial review across multiple analytical lanes to pressure-test plans against first-principles standards.
- Safe Auto-Fix Mode: Automatically applies safe, scope-preserving fixes to the execution plan for confirmed blocker and material issues by default.
- Approval Gate Mode: Supports a read-only flag that writes proposed fixes to a review log for user approval before any plan edits are made.
- Use Case: Before implementing a new feature planned via $deliver, run $review-plan to catch missing validation steps, unclear checkbox items, or scope shrinkage before any implementation code is written.
Quick Start
Use the review-plan skill to run an adversarial first-principles review of your active $deliver execution plan and apply safe plan-only fixes before starting implementation.