What problem does it solve? HyperFrames video projects need background music, sound effects, images, icons, brand logos, voiceovers, and color grades, but sourcing each asset manually scatters files across folders with no metadata, no reuse, and no provenance. This Skill turns any media need into one frozen local file plus a ledger record, keeping search noise on disk and out of the agent's context. ## Core Features & Use Cases - One-verb resolution: resolve --type <bgm|sfx|image|icon|logo|voice|grade|lut> --intent "..." returns a single frozen local path, with a deterministic cache floor that auto-reuses exact matches across projects via a content-addressed global cache at ~/.media/. - Generation and operations: TTS voiceover, local FLUX image generation, transcription (Parakeet/whisper), transcript-driven cutting, GSAP audio ducking keyframes, and ffmpeg-based trim/reframe/loudness recipes, with outputs registered back into the ledger via --from. - Color grading: resolves paste-ready data-color-grading blocks and validated .cube LUTs, including measured grade --for analysis of footage via ffmpeg signalstats. - Use Case: While building a product-launch video, run one media opportunity pass, resolve a BGM track, transition whooshes, real icons, and a warm cinematic grade; each asset lands in .media/ with an index.md inventory the whole team can read. ## Quick Start Ask the agent to resolve a background music track for your HyperFrames project with an intent like "upbeat tech launch" and use the returned local path in your composition.
