AlphaMoatによると、2026-09時点でTube Cogのダウンロード数は3,309、スター数は11で、AIスキル全体5,078件中3,421位、コンテンツ制作では362件中232位です。
| プラットフォーム | インストール数 | ダウンロード数 | スター数 | コメント |
|---|---|---|---|---|
| SkillHub ↗ | 235 | 3309 | 11 | — |
プラットフォームごとに集計方法と集計時点が異なります。
このスキルは CellCog サービスを通じて YouTube コンテンツを作成します。ショート動画、長編チュートリアル、サムネイル、スクリプト/アウトラインをカバーします。create_chat の呼び出し方(OpenClaw は fire-and-forget、他エージェントはブロッキング)、コンテンツタイプ別のプロンプト例、YouTube 仕様表(1080p 動画 / 9:16 ショート / 1280x720 サムネイル)、スタイルと chat mode の推奨(通常は agent、ドキュメンタリー系は agent team)、そして AI 動画生成は結果が予測できず大量クレジットを消費しても満足な結果にならない可能性があるという明示的な注意書きを含みます。
資料公開日 2026-10-27
スキルのドキュメントをもとに生成した例であり、実際の実行結果ではありません。
2026-09 から集計方法を変更し、同名スキルを作者ごとに分けて集計し、全スキルを対象にしました。破線は変更前の月で、変更後のデータとは直接比較せず、変更をまたぐ増減も計算していません。
| 月 | ダウンロード数 | 前月比 | スター数 |
|---|---|---|---|
| 2026-09 | 3309 | — | 11 |
| 2026-08 | 2990 | +3.0% | 11 |
| 2026-07 | 2904 | +8.5% | 11 |
| 2026-06 | 2676 | +4.7% | 11 |
| 2026-05 | 2557 | — | 11 |
| 2026-03 | 2129 | — | 11 |
作者が公開した元のファイルをテキストのまま表示しています。AlphaMoat による編集や審査はしていません。リンクはクリックできません。コマンドは内容を確認してから実行してください。取得日 2026-10-23。
---
name: tube-cog
description: "AI YouTube content creation powered by CellCog. YouTube videos, Shorts, thumbnails, video scripts, tutorials, vlogs, educational videos, product reviews, video essays. From script to finished video with voiceover and music."
metadata:
openclaw:
emoji: "📺"
os: [darwin, linux, windows]
requires:
bins: [python3]
env: [CELLCOG_API_KEY]author: CellCog
homepage: https://cellcog.ai
dependencies: [cellcog]
---
# Tube Cog - YouTube Content Powered by CellCog
Create YouTube videos that get views - from Shorts to long-form tutorials to eye-catching thumbnails.
## How to Use
For your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.
**OpenClaw (fire-and-forget):**
```python
result = client.create_chat(
prompt="[your task prompt]",
notify_session_key="agent:main:main",
task_label="my-task",
chat_mode="agent",
)
```
**All agents except OpenClaw (blocks until done):**
```python
from cellcog import CellCogClient
client = CellCogClient(agent_provider="openclaw|cursor|claude-code|codex|...")
result = client.create_chat(
prompt="[your task prompt]",
task_label="my-task",
chat_mode="agent",
)
print(result["message"])
```
---
## What Content You Can Create
### YouTube Shorts
Vertical, snackable content for discovery:
- **Quick Tips**: "Create a 45-second Short explaining one JavaScript trick"
- **Reactions/Commentary**: "Make a Short reacting to a viral tech moment"
- **Teasers**: "Create a Short teasing my upcoming full video on AI"
- **Trending Formats**: "Make a 'Things that just make sense' Short for programmers"
- **Behind-the-Scenes**: "Create a BTS Short of my content creation process"
**Example prompt:**
> "Create a 50-second YouTube Short: 'The VS Code shortcut that changed my life'
>
> Hook: 'Stop coding like it's 2010'
> Demo: Show the multi-cursor feature
> Reaction: Mind-blown moment
> CTA: 'Full tutorial on my channel'
>
> Fast-paced, screen recording style with facecam energy."
### Long-Form Videos
In-depth content that builds audience:
- **Tutorials**: "Create a 4-minute tutorial on building a REST API with Python"
- **Explainers**: "Make a 3-minute video explaining how blockchain works"
- **Reviews**: "Create a comprehensive review of the new MacBook Pro"
- **Documentaries**: "Make a mini-documentary about the history of video games"
- **Essays**: "Create a video essay on why minimalism is becoming popular"
**Example prompt:**
> "Create a 4-minute YouTube video: 'Build Your First AI App in 2026'
>
> Structure:
> - Hook (15 sec): Show the finished app
> - Intro (30 sec): What we're building and why
> - Setup (45 sec): Environment and dependencies
> - Core tutorial (2 min): Step-by-step coding
> - Testing (1 min): Run and demo
> - Outro (30 sec): Next steps and subscribe CTA
>
> Style: Educational, friendly, clear explanations. Code on screen with voiceover."
### Thumbnails
Click-worthy images that drive views:
- **Reaction Thumbnails**: "Create a thumbnail with shocked face and bold text"
- **Tutorial Thumbnails**: "Make a clean thumbnail for a coding tutorial"
- **Versus Thumbnails**: "Create an X vs Y comparison thumbnail"
- **Listicle Thumbnails**: "Make a '10 Things' style thumbnail"
- **Story Thumbnails**: "Create an intriguing thumbnail that hints at drama"
**Example prompt:**
> "Create a YouTube thumbnail for: 'I Built an App in 24 Hours'
>
> Elements:
> - Person looking stressed/determined
> - Timer showing 24:00:00
> - Code on screen in background
> - Bold text: '24 HOURS'
>
> Colors: Dark background, yellow/orange accents
> Style: High contrast, readable at small size"
### Scripts & Outlines
Content planning before production:
- **Full Scripts**: "Write a complete script for a 4-minute video on productivity"
- **Video Outlines**: "Create a detailed outline for a product review"
- **Hook Variations**: "Give me 5 different hooks for my video about investing"
- **CTA Scripts**: "Write compelling end-screen call-to-action scripts"
---
## YouTube Specs
| Format | Dimensions | Duration |
|--------|------------|----------|
| Standard Video | 1920×1080 (16:9) | Up to 4 minutes |
| Shorts | 1080×1920 (9:16) | Up to 60 seconds |
| Thumbnail | 1280×720 | - |
---
## Video Styles
| Style | Best For | Characteristics |
|-------|----------|-----------------|
| **Educational** | Tutorials, explainers | Clear, structured, helpful |
| **Entertainment** | Vlogs, reactions | Personality-driven, dynamic |
| **Professional** | B2B, corporate | Polished, trustworthy |
| **Documentary** | Essays, deep dives | Cinematic, researched |
| **Fast-Paced** | Shorts, compilations | Quick cuts, high energy |
---
## Chat Mode for YouTube Content
| Scenario | Recommended Mode |
|----------|------------------|
| Shorts, thumbnails, scripts, standard tutorials | `"agent"` |
| Documentary-style content, complex video essays, multi-part series planning | `"agent team"` |
**Use `"agent"` for most YouTube content.** Shorts, thumbnails, and standard videos execute well in agent mode.
**Use `"agent team"` for complex narratives** - documentaries, video essays, or content requiring deep research and storytelling craft.
---
## Example Prompts
**Educational tutorial:**
> "Create a YouTube tutorial: 'Docker for Beginners - Full Course'
>
> Length: 20 minutes
> Audience: Developers who've never used Docker
>
> Chapters:
> 1. What is Docker and why use it? (3 min)
> 2. Installing Docker (2 min)
> 3. Your first container (4 min)
> 4. Dockerfile basics (4 min)
> 5. Docker Compose (4 min)
> 6. Real-world example (3 min)
>
> Style: Screen recording with clear voiceover. Beginner-friendly, no jargon."
**Engaging Short:**
> "Create a YouTube Short: 'AI wrote my entire codebase'
>
> Hook: 'I let AI write 100% of my code for a week'
> Content: Quick montage of AI coding, my reactions, the chaos
> Twist: 'And it actually... worked?'
> CTA: 'Full video on my channel'
>
> Fast cuts, meme energy, relatable programmer humor."
**Click-worthy thumbnail:**
> "Create a thumbnail for: 'Why I Quit My $300K Job'
>
> Show: Professional person walking away from office building
> Expression: Confident, peaceful
> Text: '$300K → QUIT' in bold
> Style: Cinematic, slightly desaturated, text pops
>
> Must be readable at small sizes."
---
## ⚠️ Important — Video Generation Expectations
AI video production for YouTube is still at the frontier of what's possible. While some users generate high-quality videos that are ready to publish, others may spend significant credits and still not achieve a usable result. Even spending thousands of credits does not guarantee a satisfactory outcome — this is the nature of where AI video technology stands today.
There is a real learning curve to generating videos with CellCog. It takes time, money, and patience. Your prompting skill, the complexity of what you're trying to create, and how well the foundation models perform on your specific request all play a role. Results improve as you develop intuition for what works, but we want to be upfront: video generation is inherently unpredictable, and there is always a risk that the output may not meet your expectations.
---
## Tips for Better YouTube Content
1. **Hook in 5 seconds**: YouTube viewers decide instantly. Front-load the value or intrigue.
2. **Thumbnails matter more than titles**: A great video with bad thumbnail won't get clicked. Invest in thumbnail quality.
3. **Retention > Length**: A tight 8-minute video beats a padded 20-minute video. Respect viewer time.
4. **Chapters improve watch time**: Structure long videos with clear chapters. Helps SEO too.
5. **End screens work**: Ask for subscribes when viewers are most engaged (after delivering value).
6. **Shorts feed long-form**: Use Shorts to drive traffic to your main channel content.
---
## If CellCog is not installed
Run `/cellcog-setup` (or `/cellcog:cellcog-setup` depending on your tool) to install and authenticate.
**OpenClaw users:** Run `clawhub install cellcog` instead.
**Manual setup:** `pip install -U cellcog` and set `CELLCOG_API_KEY`. See the **cellcog** skill for SDK reference.
文章に残る AI らしさを見つけて取り除き、人が書いたように読める文章にします。
AI が書いた文章の典型的な痕跡を見つけ、意味を保ったまま自然な文章に書き直す文章編集スキルです。
AI が生成した文章の痕跡を見つけて取り除く中国語の編集スキルです。
wechat-search は WeChat の人気記事を検索し、関連度、人気、鮮度で結果を整理します。
公開された短編動画の共有リンクからタイトル、概要、口播テキストを抽出し、改写版も作ります。
humanizer は中国語と英語の文章を自然な表現に整えるための編集ガイドです。
Web 小説の各章をファイルに書き、会話では要約と位置だけを返します。
複数のマーケティング手法をひとつのスキルにまとめ、依頼内容に合うモジュールへ振り分けるスキルです。
データ月:2026-09 · ダウンロード数、スター数、インストール数は公開スキルレジストリ(ClawHub、SkillHub)から月単位で集計しています。詳しくはデータの方法論をご覧ください。