What problem does it solve? AI music generators fall back to enumerable defaults when a spec does not constrain them, producing tracks that sound synthetic even when the spec was followed faithfully. This Skill provides the final acceptance gate: a fourteen-item checklist of machine fingerprints, a calibrated compliance-rate threshold, and a diagnostic path from symptom to the owning skill. ## Core Features & Use Cases - Fourteen-Item AI-Tell Enumeration: Check each generated track against defaults like 8-bar-multiple sections, monotonically rising energy, grid-locked rhythm, identical chorus repeats, and perfect pitch, with machine-measurable items separated from ear-checked ones. - Two Separate Metrics: Keep compliance rate (did the backend follow the spec) and AI-tell flag count (does the result sound machine-made) as distinct numbers, with zero-point calibration setting the pass threshold at 82 rather than 60. - Remediation Routing: Map each symptom (flat, muddy, mechanical, forgettable) to the specific composition skill responsible for fixing it. - Use Case: A Suno generation comes back with a 95% compliance rate but still sounds like AI. Run the fourteen-item audit, find the energy curve never decreases and the choruses are identical, then route to the arrangement and vocal-direction skills for remediation. ## Quick Start Audit the generated track I just received against the fourteen AI-tell items and tell me whether to accept or reject it.
