Agents
Most AI video problems are not badly written prompts. They are shots that were never thought through: why does it move, who is the move for, what should the audience feel when it lands. Until those three are settled, more parameters only make the decoration heavier. This skill settles them first, then translates them into numbers.
一个会挑毛病的导演。你给它一场戏，它先判断这场戏要观众做什么，再决定镜头怎么动 —— 包括告诉你「不该」怎么动。 大多数 AI 视频的问题不在提示词写得不够漂亮，而在镜头本身没想清楚：为什么要动、动给谁看、动完观众该有什么变化。这三个问题不清楚，参数写得再全也是装饰。这个技能先把这三件事问清楚，再落到数值上。
把超过 15 秒的镜头想法，变成一段一段能跑出来的分镜方案。 MiniMax H3 一次只能生成 15 秒。想做一个 40 秒的短片，你面对的不是"写一条更好的提示词"，而是拆段、定模式、锁一致性、处理接缝这一整套工程问题。这个技能就是把这件事做完，直接给你可提交的东西。
Plan MiniMax H3 videos longer than 15 seconds. Splits a concept into chained segments, picks the right generation mode for each (T2VA / I2VA / FL2VA / L2VA / Ref2VA), locks global consistency anchors, and returns one ready-to-submit prompt per segment — plus audio-seam and continuity QC.