# H3 Prompt Writing

> Write MiniMax H3 video generation prompts for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. Use when rewriting multimodal requests into H3 prompt structures, composing integrated_multimodal_description, overall_soundscape, and non_diegetic_music, aligning keyframes, or defining reference labels for images, videos, and audio.

- Skill: `zmgid/h3-prompt-writing` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add zmgid/h3-prompt-writing`
- Raw SKILL.md: https://api.skillmd.com/api/skills/zmgid/h3-prompt-writing/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: zmgid (https://skillmd.com/u/zmgid)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/zmgid/h3-prompt-writing

---


# H3 Prompt Writing

## Workflow

1. If the input is only a rough idea, story, or image without an executable shot plan, read the sibling `video-director` skill first. Use its selected directing plan as the source of truth; do not invent a second concept during prompt conversion.
2. Identify the input mode: T2VA, I2VA, FL2VA, L2VA, or full-reference Ref2VA.
3. For base text/keyframe modes, read `references/base-en.txt` and follow its final prompt structure.
4. For full-reference mode, read `references/ref-en.txt` and follow its six-section rewrite format.
5. Preserve the exact field names, section order, labels, timing, actions, transitions, and ending from the selected guide and directing plan.

## Base Modes

- T2VA: build the full audiovisual timeline from text.
- I2VA: start from the first frame and develop forward from it.
- FL2VA: describe the continuous path between the first and last frames.
- L2VA: infer a plausible opening and converge to the supplied last frame.

Use `integrated_multimodal_description`, `overall_soundscape`, and `non_diegetic_music` in the order shown in `references/base-en.txt`.

## Full-Reference Mode

Ref2VA rewrites use `subject_definitions`, `summary`, `retention_analysis`, `detailed_description`, `overall_soundscape`, and `non_diegetic_music` in that order. Reference labels stay consistent across all sections.

Read `references/ref-en.txt` for label rules, retention analysis, and complete examples.

## Output Rules

- Write rewrite sections in English; preserve dialogue, lyrics, and visible scene text in their original language.
- Describe each shot by composition, subjects, environment, actions, camera, sound, and the exact point where referenced content appears.
- Avoid plot summaries, unresolved reference labels, and timing that does not match the requested duration.

## Dsivio 内置版

对话使用本 Skill 的原始脚本和 MCP。查找、保存或修改模板时读 [共享模板说明](../../TEMPLATES.md)，直接读写共享文件夹；不创建页面任务或同步草稿。运行环境由应用提供，内置副本随应用更新。

## 视频配置在哪

视频页面和本插件共用一份 `providers.json`，页面保存后下次脚本调用直接生效，不需要重复配置。路径优先取 `DSVIDEO_CONFIG_PATH`；否则 Windows 为 `%APPDATA%/dsvideo/providers.json`，macOS/Linux 为 `${XDG_CONFIG_HOME:-~/.config}/dsvideo/providers.json`。
用户问模型或配置时，运行 `python <插件根目录>/scripts/dsvideo_config.py show`（密钥脱敏），查看 `providers.grok` / `providers.minimax` / `providers.comfy` 的 `model` 和 `base_url`。MiniMax-H3 路线模型固定为 MiniMax-H3；ComfyUI 模型由工作流决定。旧环境变量显式覆盖时说明来源。生成仍使用插件自己的脚本和 MCP。