Agent skill

Immersive Short Video

by xrkseek in xrkseek/XRK-AGT

Produce immersive vertical short videos (口播/科普/产品讲解) without AI-slop aesthetics.

MITAuto-check passedMedia & Creative

Install Immersive Short Video

skills CLI
$ npx skills add xrkseek/XRK-AGT --skill immersive-short-video -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install xrkseek/XRK-AGT immersive-short-video --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/xrkseek/XRK-AGT.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.cursor/skills/immersive-short-video .claude/skills/immersive-short-video && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
immersive-short-video
GitHub stars
138
Token cost
~1k tokens
SKILL.md length
284 words
Files
7 (incl. scripts)
Skills in repo
12
Repo updated
First seen
Licence
MIT

At a glance

Produce immersive vertical short videos (口播/科普/产品讲解) without AI-slop aesthetics.

  • Works in 6 steps: Script & VO → Vertical storyboard → Generate images (magenta key) → …
  • The XRK video pipeline
  • SKILL.md covers Non-negotiables, End-to-end flow, Step 1 — Script & VO and Step 2 — Vertical storyboard, plus 7 more sections
  • Runs Python and JavaScript scripts from its folder; calls python, npx and node

What it does

Immersive Short Video is an agent skill from xrkseek/XRK-AGT. Produce immersive vertical short videos (口播/科普/产品讲解) without AI-slop aesthetics. Covers magenta-key assets, chroma-key, JS/Remotion/Vue seekable timelines, TTS, and the canonical Playwright frame-seek + h264nvenc assemble path. Use for 短视频, 口播, 科普视频, explainer, Remotion, 生图抠图, 动画帧, 录屏, NVENC, or the XRK video pipeline.

Its SKILL.md is about 1k tokens, which your agent loads only when the skill is triggered. The skill folder holds 7 other files, including scripts (for example `reference-capture.md`, `reference-prompts.md` and `reference-stacks.md`).

It sits in Media & Creative, covering Video production, Text to speech and voice and Browser testing. It works with Remotion, Vue.js and Playwright. The repository describes itself as: 这是基于多语言所搭建的基于web,工作流,新型算法组件的智能体. The licence is MIT.

When your agent uses it

  • The XRK video pipeline
  • Tasks that involve Video production
  • Tasks that involve Text to speech and voice

Example prompts

  • “/immersive-short-video”

Requirements

  • Python 3
  • Node.js

Workflow steps

6 steps, taken from the step headings in SKILL.md.

  1. Script & VO
  2. Vertical storyboard
  3. Generate images (magenta key)
  4. Chroma key
  5. Scene tech stack(选一)
  6. Assemble(权威录屏 · 必须遵守)

What it can do on your machine

Read from SKILL.md and the folder at commit 63f004e. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 2 files in scripts/ (Python and JavaScript), which the agent can run.

    Shell commands in SKILL.md call:

    • python
    • npx
    • node

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use npx, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Immersive Short Video loads about 1k tokens when it runs. Until then it costs about 86 tokens; SKILL.md has 284 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~86
When it runs · the whole SKILL.md, loaded when a task matches
~1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from xrkseek/XRK-AGT at commit 63f004e, republished under its MIT licence (© xrkseek). 284 words, ~1,033 tokens.

Download SKILL.mdSave it as .claude/skills/immersive-short-video/SKILL.md (or your agent's skills folder). This skill also uses 6 other files; get the full folder from GitHub.
name
immersive-short-video
description
Produce immersive vertical short videos (口播/科普/产品讲解) without AI-slop aesthetics. Covers magenta-key assets, chroma-key, JS/Remotion/Vue seekable timelines, TTS, and the canonical Playwright frame-seek + h264_nvenc assemble path. Use for 短视频, 口播, 科普视频, explainer, Remotion, 生图抠图, 动画帧, 录屏, NVENC, or the XRK video pipeline.

Immersive Short Video Pipeline

后续短视频默认走本 skill。目标:竖屏 9:16、代入感强、少 AI 味(像真实 App/IM 界面,不像霓虹 HUD 演示片)。

Non-negotiables

  1. 画幅:默认 1080×1920。禁止左右双栏挤一屏;单列 + 安全区。
  2. 安全区:上 ≥100px · 下 ≥180px · 左右 ≥56px。一镜一个主视觉。
  3. 去 AI 味:禁止霓虹网格、粒子扫描线、赛博 HUD、紫渐变玻璃拟态堆叠、居中巨大描边字幕叠画面。
  4. 代入感:优先模拟真实产品——QQ/飞书式聊天、系统设置卡、CLI 终端、CI 面板;多用图标/头像 PNG,少纯 CSS 几何装饰。
  5. 字幕:默认不烧录;口播靠声音;画面内只保留短标题(一句)。
  6. 资源规范:需抠图的素材一律 纯品红底 #FF00FF 生成,再跑抠图脚本。

End-to-end flow

text
选题/口播稿 → TTS(mp3+时间轴)
  → 分镜表(竖屏·一镜一意)
  → 生图(品红底 icons/avatars/UI 切图)
  → chroma_key → 透明 PNG
  → 可 seek 场景页(JS 时间轴 / Remotion / Vue)
  → record: __XRK_SEEK__ 逐帧截屏 @60fps + h264_nvenc → 成片 mp4

工作目录建议:tmp/<slug>-video/(含 assets/ scenes/ voice.* record.* README.md)。

Step 1 — Script & VO

  • 中文口播约 180–220 字/分;抖音/视频号常见 40–60s。
  • 结构:钩子 3s → 问题 → 解释 → 正确做法 → 收束。
  • TTS:本机可用 edge-tts(如 zh-CN-YunyangNeural);写出 voice.mp3,可选 voice.srt 仅作时间轴参考(不烧进片)。

Step 2 — Vertical storyboard

每镜只答一个问题。表头:t_start t_end visual on_screen_line assets。

时段画面类型(优先)
钩子真实 IM 手机框 + 气泡入场
概念大图标 + 2–4 张竖向信息卡
风险警告图标 + 短列表
做法全宽竖卡堆叠(CLI / CI / Scope)
收束主图标 + 3 枚徽章

Step 3 — Generate images (magenta key)

用 GenerateImage(或等价生图)时 必须 在 prompt 末尾锁定:

text
Solid pure magenta #FF00FF fills the entire background canvas.
Subject centered. No other background. No text. No watermark.
Magenta everywhere outside the subject for chroma key.

命名:assets/<name>-magenta.png(如 icon-qq-magenta.png、avatar-bot-magenta.png)。

图标风格:圆角方标 App icon / 圆形头像;像商店上架图,不要插画风噪点堆满。

Step 4 — Chroma key

bash
python .cursor/skills/immersive-short-video/scripts/chroma_key.py \
  --input tmp/<slug>-video/assets \
  --glob "*-magenta.png"

产出同目录去后缀透明图:icon-qq.png。容差默认适配 AI 压缩品红边缘。

Step 5 — Scene tech stack(选一)

无论 A/B/C,动效必须可由绝对时间 t 复现(供 Step 6 seek 截屏)。禁止只靠实时 CSS @keyframes 播一遍再录。

分镜级运动(必做)
  • [data-fx] + data-d:JS render(t) 算入场(easeOutBack / easeOutQuart),像分镜不是 UI pop。
  • 切镜:交叉淡化 ~0.65s + 轻推拉/位移;全程 #cam 缓慢 drift/zoom。
  • 暴露 window.__XRK_DUR__ 与 window.__XRK_SEEK__(t);?capture=1 静帧,预览用 rAF。
  • 参考实现:tmp/xrk-agt-intro/scenes.html。
A. Remotion(React)
  • AbsoluteFill Sequence spring interpolate;竖屏 1080×1920。
  • 有 CLI 时 npx remotion render;否则导出可 seek 预览页走 Step 6。
B. Vue 3 + Motion / GSAP
  • data-a/data-b 对齐口播;组件化卡片;同样要可 seek 到任意 t。
C. 纯 HTML(快速成片可用)
  • 单列竖屏 + 上节时间轴;样例 tmp/xrk-agt-intro/。

Step 6 — Assemble(权威录屏 · 必须遵守)

完整说明:reference-capture.md · 模板:scripts/record-seek-nvenc.mjs

默认路径(有 NVIDIA):Playwright 按帧 __XRK_SEEK__ → JPEG@60fps → h264_nvenc + voice.mp3。

  1. 画面时长 = 口播时长(±0.3s);DUR 与 mp3 对齐。
  2. node record.mjs(从模板复制到 tmp/<slug>/,改输出文件名)。
  3. 禁止:Playwright screencast / page.video() 再 2× 或 fps 上采样到 60;禁止 animations: 'disabled'。
  4. 无独显:同 seek 流程,改 libx264 CRF 17,可降到 30fps。
  5. 成片自检:有声、1080×1920、约 60fps;勿交付 silent.mp4。

Anti-AI checklist(交付前勾选)

  • 有真实图标/头像 PNG,不是全靠几何形
  • 背景是纸感/产品灰或真实 UI,不是赛博粒子场
  • 文字少、靠下或作短标题,无大字叠字
  • 动效像 App 过渡(slide/fade/spring),不像演示 HUD
  • 一镜一意,列表不超过 4 条可见项
  • 品红原图保留在 assets/*-magenta.png 便于重抠

Project conventions (XRK-AGT)

  • 权威录屏样例:tmp/xrk-agt-intro/(scenes.html 时间轴 + record.mjs seek/NVENC)。
  • 早期样例:tmp/strix-kepu-anim/。
  • 科普内容需与仓库文档一致,勿编造接口。
  • 不把临时成片提交进 core/ / src/。

Extra reference

© xrkseek, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 6 other files (scripts) in .cursor/skills/immersive-short-video of xrkseek/XRK-AGT.

  • SKILL.md
  • reference-capture.md
  • reference-prompts.md
  • reference-stacks.md
  • reference-vertical.md
  • scripts/chroma_key.py
  • scripts/record-seek-nvenc.mjs

Open the folder on GitHubat commit 63f004e

Compare with similar skills

Immersive Short Video next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Immersive Short Video compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Immersive Short Video this skillxrkseek/XRK-AGT138—~1kAutomated safety check: PassMIT
Whiteboard Videotrustfuture/simon-skills382—~1.8kAutomated safety check: NotesMIT
Playwright Browser Demo Recordingdigitalsamba/claude-code-video-toolkit2.2k1 repos~3.2kAutomated safety check: PassMIT
Motion Adfabricioctelles/skills106—~4.1kAutomated safety check: PassApache-2.0
Eva Feature Screenshotvvedantb/eva101—~3.2kAutomated safety check: NotesMIT
Cs Web Promo FilmChenShuo2004/cs-skills197—~1.1kAutomated safety check: PassMIT

Similar skills

  • Whiteboard Video

    trustfuture/simon-skills

    手绘白板风"边画边讲"讲解视频出片 skill(Excalidraw 风格逐笔动画 + 火山引擎配音 + 烧录字幕 + 品牌水印与片尾卡 + 横竖两张封面 + 各平台发布文案)。当用户说"做一期白板视频 / 边画边讲 / 手绘讲解视频 / 用 excalidraw 做视频 / whiteboard video",或要在本仓库里新建一期、改场景、换贴纸或真实…

    382 GitHub stars~1.8k tokensUpdated 17 days ago
    Media & CreativeAuto-check: notes
  • Playwright Browser Demo Recording

    digitalsamba/claude-code-video-toolkit

    Records browser interactions as video with Playwright, covering viewport sizing, cursor highlighting, and converting output for Remotion.

    2.2k GitHub starsUsed in 1 repo~3.2k tokens
    Media & CreativeAuto-check passed
  • Motion Ad

    fabricioctelles/skills

    Produce a short motion-graphics video ad — a 15s Facebook/Instagram/TikTok spot — as a rendered MP4.

    106 GitHub stars~4.1k tokensUpdated today
    Media & CreativeAuto-check passed
  • Grab a single clean HD screenshot of a new eva feature from the real running app (Playwright at deviceScaleFactor 2, 1280 layout captured crisp at 2560×1440, dev overlays hidden) and write a tweet…

    101 GitHub stars~3.2k tokensUpdated today
    Media & CreativeAuto-check: notes
  • Cs Web Promo Film

    ChenShuo2004/cs-skills

    把真实网页做成 30–60 秒产品演示宣传片:Playwright 采集页面长截图,Remotion 用虚拟浏览器窗口做推进、平移、滚动和点击,渲染可交付 mp4,并可附口播稿。触发词包括 $cs-web-promo-film、网页宣传片、产品演示片、把这个网页做成视频、产品功能演示。不用于网页幻灯片、静态海报、真人出镜、电商对标复刻(改用 $cs-auto-videl)、ChatCut…

    197 GitHub stars~1.1k tokensUpdated 3 days ago
    Media & CreativeAuto-check passed
  • Render Mosaic Grid Reveal

    gooseworks-ai/goose-skills

    Render a 'mosaic-grid-reveal' video from a config — a real-DOM FULL-BLEED N×N mosaic of real product tiles that pops in one tile at a time (scatter order, ease-out-back overshoot), the grid clears…

    1.2k GitHub stars~1.1k tokensUpdated 2 days ago
    Media & CreativeAuto-check passed

More from xrkseek/XRK-AGT

All 12 skills in this repo
  • Xrk Crawl

    xrkseek/XRK-AGT

    当你需要开发/排查 HTTP 抓取、SSRF、Playwright 受控浏览器、本地字体增强截图,或判断 webfetch 与 browser 工作流如何选型时使用。

    138 GitHub stars~1.3k tokensUpdated 10 days ago
    Auto-check passed
  • Xrk HTTP API

    xrkseek/XRK-AGT

    当你需要开发或排查 HTTP API(core//http/.js)、理解 HttpApi 基类、HttpApiLoader、业务层约定时使用。

    138 GitHub stars~569 tokensUpdated 10 days ago
    Auto-check passed
  • Xrk Node Runtime

    xrkseek/XRK-AGT

    编写或审查 core/src 代码时,确保使用 Node 26 稳定 API,禁止旧写法(fetch/exec/错误/二进制)。AI 改 Core 前必读。

    138 GitHub stars~907 tokensUpdated 10 days ago
    Auto-check passed
  • Xrk GitHub Research

    xrkseek/XRK-AGT

    在 GitHub 搜索成熟开源实现、对比方案、读 issue/PR 时使用。架构选型、unfamiliar 领域、找业界最佳实践时主动调用。

    138 GitHub stars~202 tokensUpdated 10 days ago
    Auto-check passed
  • Xrk LLM

    xrkseek/XRK-AGT

    当你需要配置/新增/排查 LLM 提供商(OpenAI/Azure/Gemini/Anthropic/Ollama/各类兼容网关)时使用;确保 YAML/Schema/代码一致。

    138 GitHub stars~304 tokensUpdated 10 days ago
    Auto-check passed
  • Xrk Www Compat

    xrkseek/XRK-AGT

    编写或审查 core//www 静态页、校园 WebView 兼容、HttpResponse 前端解包时使用。Core www 走浏览器兼容层(xrk-www-compat)。含零配置静态与有 sign.json(纯静态/产物/反代,sign↔server 合并)挂载。

    138 GitHub stars~544 tokensUpdated 10 days ago
    Auto-check passed

Questions about Immersive Short Video

What does Immersive Short Video do?

Produce immersive vertical short videos (口播/科普/产品讲解) without AI-slop aesthetics. Immersive Short Video is an agent skill from xrkseek/XRK-AGT. Produce immersive vertical short videos (口播/科普/产品讲解) without AI-slop aesthetics.

When should I use Immersive Short Video?

Immersive Short Video fits situations like: the XRK video pipeline; tasks that involve Video production; tasks that involve Text to speech and voice.

How do I install Immersive Short Video in Claude Code?

Run `npx skills add xrkseek/XRK-AGT --skill immersive-short-video -a claude-code`. Or copy the skill folder (.cursor/skills/immersive-short-video in xrkseek/XRK-AGT) into .claude/skills/immersive-short-video in your project. Claude Code loads it when a task matches its description.

How do I install Immersive Short Video in Codex?

Run `npx skills add xrkseek/XRK-AGT --skill immersive-short-video -a codex`. Or copy the skill folder (.cursor/skills/immersive-short-video in xrkseek/XRK-AGT) into .agents/skills/immersive-short-video in your project. Codex loads it when a task matches its description.

Can I use Immersive Short Video in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add xrkseek/XRK-AGT --skill immersive-short-video -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/immersive-short-video, .gemini/skills/immersive-short-video, .github/skills/immersive-short-video and .opencode/skills/immersive-short-video in your project.

What does Immersive Short Video need to run?

Going by SKILL.md and its folder, Immersive Short Video needs Python and JavaScript for the scripts in its folder and the command-line tools its instructions call (python, npx and node). Our summary lists: Python 3; Node.js.

Does Immersive Short Video access the network?

SKILL.md contains no URLs. Its commands use npx, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Immersive Short Video safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Immersive Short Video use?

Immersive Short Video is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Immersive Short Video use?

About 1k tokens (SKILL.md is roughly 4.1k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Immersive Short Video?

Skills that share tags, products or a category with Immersive Short Video: Whiteboard Video (trustfuture/simon-skills, 382 stars), Playwright Browser Demo Recording (digitalsamba/claude-code-video-toolkit, 2.2k stars), Motion Ad (fabricioctelles/skills, 106 stars) and Eva Feature Screenshot (vvedantb/eva, 101 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Immersive Short Video?

xrkseek (a GitHub user) maintains it in xrkseek/XRK-AGT, which has 138 GitHub stars. The repository holds 12 skills in this directory. The repository was last updated on October 1, 2026.

Source: xrkseek/XRK-AGT on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.