brand-promo-video-generator — Claude Code skill: brand promo videos

What brand-promo-video-generator does

The skill runs a project from brief to finished cut. It starts with intake: you upload or link logos, fonts, brand colors, product images, UI screenshots and official pages, and confirm duration, aspect ratio, audience and copy language. From there it builds a brand truth sheet, picks a narrative direction, plans beats and shots, generates the imagery, video, voiceover or music it needs, and assembles the final edit with a pre-delivery review.

The work is done through hub_* tools: image search, media analysis, image, video, audio and music generation, speech synthesis, video editing and canvas grouping. It needs the MiniMax Hub agent with a canvas workspace, so it does not carry over to a generic agent harness. When a runtime is missing one of the generation or editing tools, the skill downgrades the deliverable to a pre-production package rather than pretending a final video exists.

A provenance manifest is part of the flow. Every identity-bearing asset gets an ID, a source URL or note, a source type, a verification target and a rights note. The manifest itself grants no publication rights. If a logo, UI or font cannot be authenticated, the skill stops and asks for an authorized original instead of generating a plausible stand-in.