What problem does it solve? Agent-generated plans often fail at two altitudes: they are either too vague ("add caching", "handle auth") or lost in low-level detail with no product vision. This Skill subjects a plan to an adversarial senior-engineer review grounded in the actual codebase and current best practices, then rewrites it into a scoped, evidence-backed version. ## Core Features & Use Cases - Evidence-based adversarial review: Every finding cites a file and line from the repository or a dated external source, with blocker/major/minor severity levels. - Altitude diagnosis: Classifies plans as fog (too vague), tunnel (too granular), or mixed using structured checklists and a vague-word blacklist. - Plan promotion: Produces a rewritten plan with goals, decisions with rationale, sequencing, rollback strategy, and open questions for a human. - Use Case: After an AI assistant drafts a migration plan, invoke this Skill to verify the plan against pinned dependency versions, check for deprecated APIs via web research, and receive a hardened v2 plan with a delta summary. ## Quick Start Review this plan like a staff engineer and promote it to a senior-level version.
