What problem does it solve? Skill collections grow messy over time: overlapping skills, dead references, unverified behavior claims, and new skills added without checking existing ones. This Skill governs the full lifecycle of creating, consolidating, registering, and evaluating discipline skills so the registry stays accurate and evidence-based. ## Core Features & Use Cases - Consolidation-first boundary decisions: Checks whether an existing skill mode or reference extension solves the need before creating a new skill, preserving unique inputs, outputs, and approval boundaries. - Behavior evaluation coverage: Requires every active skill to have primary and non-selection behavior coverage, with results tied to exact commit, registry SHA-256, and an independent reviewer context. - Evidence classification: Distinguishes EXECUTABLE_EVIDENCE, CONTRACT_EVIDENCE, and MISSING_EVIDENCE so file existence is never promoted to actual model behavior proof. - Use Case: When a team wants to add a new workflow skill, use this Skill to inventory existing skills, decide whether to extend a mode or create a new unit, register it with triggers and aliases, and run behavior evaluations before claiming it works. ## Quick Start Use the evolving-project-discipline-skills skill to review the current skill registry and decide whether my proposed new skill should be created or merged into an existing one.
