技能 - v0.54.0 - 多参考合成、全新广告管线，以及 9:16 动态视频

Expand your agent's creative toolset with new composition capabilities, automated ad pipelines, and critical timing improvements for video generation.

Features & Improvements

• Multi-reference composition: Combine several reference images into cohesive scenes with the new scenario-multi-reference skill.
• Social and kinetic video: Generate mobile-ready content via the scenario-kinetic-music-video skill, featuring native 9:16 portrait rendering and scoped soundtrack editing.
• Expanded workflows: Automate repetitive tasks with new pipeline skills dedicated to rendering video thumbnails, creating ad variants, and generating interactive pet avatars.

Fixes

• Transparent cutouts in the scenario-gpt-image skill now reliably route to the 2.5 model family for better edge accuracy.
• Plate crops and chant timings properly align when rendering kinetic music videos.
• The scenario-video-assembly skill no longer disrupts pipelines by attempting to send subtitle (SRT) files to the asset upload endpoint.

Get the update

Run this command where your skills were initially installed to pull the latest instructions:

npx skills update

Not installed yet? Grab the entire collection to access all the new ad, thumbnail, and composition pipelines in one go:

npx skills add scenario-labs/skills --skill "*"