SkillFoxx

srt-vox-director skill

Skill turns finished SRT subtitles into a pack of image and video prompts: 21 styles, timecode storyboard, no media generation

  • 125 ★ on the repository
  • Last commit
Skill

Minimal access

Why this level

  • The skill writes only text prompts and runs nothing.
All reasons and checks

geeklee/srt-vox-directorAuthor:geeklee

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add skills/srt-vox-director

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Minimal accessThe tool runs nothing and makes no network calls.

  • The skill writes only text prompts and runs nothing.

This is third-party code. Review the repository files before installing.

What it does

srt-vox-director starts from narration that already exists. It reads the SRT, picks a visual system, splits the text into shots by meaning and snaps them to 4, 6, 8 and 10 second steps, then outputs reference-image prompts and image-to-video prompts. All prompts are written in Chinese. There are five modes: pick a style from 21, split the storyboard, produce the full pack (the main mode), redo one shot and diagnose generation failures by symptom. The usual path runs through modes 1, 2 and 3 and stops for confirmation. Long videos write a `state.json` file and the command «continue» resumes work. The validators `check_storyboard.py`, `check_prompts.py` and `check_state.py` verify that shots cover the whole timeline with no gaps or overlaps.

Who it is for. Authors of explainer videos whose narration text and subtitles are ready while images and clips are generated on their own platforms.

Good fit when

  • Narration is recorded and an SRT needs cutting into shots with exact lengths.
  • You need to choose a style out of 21 based on the text topic and mood.
  • Frame generation failed and a minimal prompt fix by symptom is needed.
  • A long video of dozens of shots is produced in batches with resumption.

Not a fit when

  • You want the skill to draw images or build an MP4: it outputs text only.
  • There is no narration or subtitles yet, only a topic.
  • You need prompts in a language other than Chinese.

Example request

Turn story.srt into a prompt pack for an explainer video: suggest three styles with samples and cut it into shots.

Limitations

The skill creates no image or video: any request for prompts stays text and the rule is not relaxed in any mode. Prompts are written in Chinese, and styles and examples target it. After prompts are produced the path runs through third-party services: the recommended chain is ChatGPT Images, Google Flow and an editor, and substitutes are possible. The first frame of a generated clip often differs from the uploaded reference image, which is normal for such models. The repository does not state a license. The Vox style belongs to Vox Media.

How to disable. Delete the skill folder from your agent's skills directory. Prompt packs and the `state.json` file stay in the project.

FAQ

How is srt-vox-director different from video generators?

It makes no clip and writes no script. Narration already exists and the skill produces prompts from subtitles to shots.

How many styles are available?

Twenty-one. Each has a sample, a close-up of differences and a six-panel board to upload to a model.

How does the skill fit shots to generation?

Lengths round up to 4, 6, 8 or 10 seconds, and the difference from narration is calculated explicitly and built into the prompt.

What about a long video?

The skill writes `state.json` and resumes where it stopped on the command «continue».

Can the result be verified?

Yes. The scripts `check_storyboard.py` and `check_prompts.py` check timeline coverage and prompt format, and there are `unittest` tests.

Discussion

No questions or reviews yet.

Skills

Skill for Codex and Gemini: turns a phrase, text or SRT into a Vox-style storyboard with prompts and JSON for Google Flow

SkillMinimal access126

video-production-skill takes an agent from script to video: slides, speech-verified narration, subtitles and a cover.

SkillExtended access106

A skill that builds a narrated paper-collage explainer video from a topic, a talking-head clip or a single photo

SkillBroad access347

A skill that turns a Chinese voice-over script into animated HTML for screen recording, in manual mode or SRT autoplay mode

SkillMinimal access247

A skill that turns one topic into a finished Vox-style paper-collage video: script, keyframes, motion, voice-over, music and captions

SkillExtended access2.8K

The video-explainer skill turns a script into a vertical 9:16 explainer: narration, captions, Remotion and a pre-render review

SkillExtended access124

README in short

The repository is organized as a skill with a routing `SKILL.md`, reference files for styles and storyboards, sample images and checker scripts.

README badge

Original SKILL.md

---
name: srt-vox-director
description: 把已经写好的字幕（SRT / 配音稿 / 解说词 / 旁白稿）变成 Vox 风格解释视频的分镜与提示词包——参考图提示词、图生视频提示词、分镜表、关键词台账、视觉圣经、风格选择。只交付文本提示词，不生成任何图片或视频。适用于用户已有成片旁白、想做成分镜或配画面、需要切分镜头与处理时长差值、选定视觉风格，或出图/出片失败后诊断修复（字错了、画面没动、风格跑偏等）。
---

# SRT 解释视频导演

把一份已经写好的 SRT 字幕，变成可以直接复制去生图、生视频的提示词包。
**所有交付的提示词都用中文写**——风格块、参考图提示词、视频提示词、负面约束，一律中文。

「Vox 风格」在这里指：强解释性的信息编排、可触摸的实体信息物件、清晰的编辑层级、有物理意义的运动。
不得暗示与 Vox Media 有关联、不使用其标识、不逐镜复刻任何已发布视频。

## 路由：选最小够用的模式

1. **选风格** —— 从风格库里推荐三种并展示这三张样图，定下风格 DNA 与视觉圣经。
   **不要一次抛出全部**，用户想看更多时才按分组展开。
2. **切分镜** —— 读 SRT，按语义切分并就近吸附到 4/6/8/10 秒，输出分镜表。
3. **出完整包** —— 主模式。分批输出参考图提示词与视频提示词。
4. **局部补出** —— 只重出某一镜的提示词。
5. **诊断修复** —— 出图或出片失败时，诊断并做最小必要改写。

默认链路 `1 → 2 → 3`，每步停下来等用户确认。用户直接扔 SRT 且未说别的时，从模式 1 开始。

**模式 4 与模式 5 怎么分**——用户说「第 7 镜不对」时不要猜：

```
画面已经生成了、结果不对          → 模式 5，先定位症状再最小改写
画面还没生成 / 要换内容或换角度   → 模式 4，重出该镜提示词
分不清就问一句：你是已经生成了觉得不对，还是还没生成想先改？
```

模式 5 **需要用户描述症状**。只说「不对」时，先拿 troubleshooting 的症状索引表反问是哪一种，
不要凭空猜着改——猜着改会同时动几处，用户无法判断是哪一处生效。

「给我提示词」不等于允许生成媒体。本 skill 在任何模式下都不生成图片或视频。

## 按需加载参考

| 什么时候 | 读什么 |
|---|---|
| 任何完整交付 | [references/delivery-contract.md](references/delivery-contract.md) —— 输出段落结构、目录结构、state.json、用户自检表 |
| 模式 1 | [references/style-gallery.md](references/style-gallery.md) 样图索引 · [references/style-selector.md](references/style-selector.md) 七维打分 · [references/style-library.md](references/style-library.md) 风格块本体（只取选中那一条） |
| 模式 2，**必读** | [references/storyboard-algorithm.md](references/storyboard-algorithm.md) —— 不得凭经验切分镜 |
| 需要题材专属的切镜策略 | [references/storyboard-blueprints.md](references/storyboard-blueprints.md) |
| 写任何参考图或视频提示词之前，**必读** | [references/prompt-templates.md](references/prompt-templates.md) |
| 画面有任何上屏文字（中文、英文、数字皆算），**必读** | [references/prompt-keywords.md](references/prompt-keywords.md) |
| 涉及**参考图画法**、运动、组装方向、桥接帧、余量收束与补差 | [references/prompt-motion.md](references/prompt-motion.md) —— 第 6 节是参考图角色与组装方向的唯一正典 |
| 模式 5，出图或出片失败 | [references/troubleshooting.md](references/troubleshooting.md) —— 按症状索引定位与修法 |
| 用户是新手，或问「从哪开始」「参考图哪里来」「传到哪」，**必读** | [references/delivery-handoff.md](references/delivery-handoff.md) |

Report a problem