What problem does it solve? Generated and drafted tracks often sound flat or recognizably AI-made because nobody made structural decisions: energy rises monotonically, instrumentation never changes, every section is a multiple of 8 bars, nothing is ever subtracted, and endings are default fade-outs. This Skill provides the decision framework for how a track is shaped over time, grounded in measurements from 62 real multitrack arrangements. ## Core Features & Use Cases - Energy curve and dramatic arc: Maps Freytag's pyramid onto music, enforcing the measured rule that real arrangements always contain an energy descent, with explicit exception conditions for dub and minimal styles. - Subtractive arranging and layering as form: Provides the fill-then-carve workflow, instrument entry/exit event tables, and rules backed by corpus data (59/59 tracks have mid-song entries and early exits). - Diagnosis and ARR-SPEC field mapping: Supplies a symptom-to-cause diagnostic table, section proportion conventions, hook placement rules, three ending strategies, and direct mapping to ARR-SPEC fields like energy_curve, subtraction_events, and roster. - Use Case: A generated song sounds flat and every section feels identical. Use this Skill to diagnose the cause (unchanged instrumentation, no subtraction), then rewrite the arrangement with an entry/exit roster, a non-monotonic energy curve, and a deliberate ending. ## Quick Start Ask the agent to plan the arrangement structure of your song, including the energy curve, instrument entry and exit points, and at least one subtraction event, and fill those into the ARR-SPEC arrangement fields.
