Youtube seo
Generate YouTube titles, descriptions, timecodes and hashtags from video transcripts.From its SKILL.md
npx -y skills add AlekseiUL/sprut-agent-kit --skill youtube-seoAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- runs commandsInstructs the agent to run 2 commands, including `ffmpeg -i video.mp4 -vn -acodec libmp3lame audio.mp3` and 1 more.
SKILL.md
7.0 KB, ~1.8k tokens by cl100k_base, as published. Nobody here has run it
YouTube SEO Generator
Generate click-worthy, SEO-optimized titles, descriptions, timecodes and hashtags for YouTube videos.
✍️ Правила текста (стиль владельца)
Обязательно:
- Дефис (-) вместо длинного тире (—). ВСЕГДА
- Сильные глаголы, короткие предложения
- Личный опыт, метафоры из жизни
- Хук в заголовке, интрига в первых 2 строках описания
Запрещено:
- Длинное тире (—) - заменять на дефис (-)
- "Конечно", "Безусловно", "Стоит отметить", "Является"
- Канцелярит и пассивный залог
- Больше 1 эмодзи в описании
Полные правила:
skills/copywriter/SKILL.md
Workflow
Step 1: Get transcript
If user provides video/audio file:
# Extract audio if needed (video → audio)
ffmpeg -i video.mp4 -vn -acodec libmp3lame audio.mp3
# Transcribe via OpenAI Whisper API
{skillsDir}/openai-whisper-api/scripts/transcribe.sh audio.mp3 --language ru --json --out transcript.json
If user provides transcript text: Use directly.
Step 2: Analyze transcript
- Identify main topic and goal
- Extract key insights, numbers, facts
- Map narrative structure (for timecodes)
- Determine content type:
- Tutorial → focus on value and steps
- Case study / research → focus on results and insights
- Review / comparison → focus on comparison and choice
- Interview → focus on expert and their opinion
Step 3: Generate metadata
Style Guide
- First person (as channel author)
- Tone: sincere, professional, not pushy
- ONE emoji maximum in entire description
- Avoid clichés: "В этом видео я расскажу..."
- Natural keyword integration, no spam
Target Keywords (use contextually)
- Автоматизация, n8n, AI-агент
- Искусственный интеллект / AI
- Промпт / prompt engineering
- Additional keywords from transcript content
Avoid (YouTube demonetization triggers)
- Violence, weapons, drugs, gambling
- Forex/crypto (unless educational)
- Profanity, discrimination
- Misleading clickbait
- Pushy "подпишись/лайкни" — use neutral alternatives
Output Format (STRICT)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
📹 НАЗВАНИЕ И ОПИСАНИЕ ДЛЯ YOUTUBE-РОЛИКА
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🎯 ПЯТЬ ВАРИАНТОВ НАЗВАНИЯ:
Вариант 1: [Название]
Обоснование: [Текст]
Вариант 2: [Название]
Обоснование: [Текст]
[... до 5 вариантов]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
✅ РЕКОМЕНДУЕМОЕ НАЗВАНИЕ:
[Твой выбор]
Обоснование выбора: [Почему именно это название]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
📝 ОПИСАНИЕ РОЛИКА:
[Полный текст описания]
🔗 ОБЩАЕМСЯ ТУТ: https://t.me/YOUR_CHANNEL
⏱ Тайм-коды:
0:00 - Введение
[...]
#SPRUT #SPRUTAI [остальные хештеги]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🏷 ХЕШТЕГИ ДЛЯ ЗАГРУЗКИ РОЛИКА (через запятую, до 500 символов):
[список без решёток, через запятую]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Title Requirements
- 70-75 chars optimal, max 100
- First 70 chars = most engaging part
- Include 1-2 target keywords
- Use numbers when relevant ("7 принципов...")
- Intrigue + specificity + value
- Each variant needs rationale: SEO value, clickability, emotional trigger
Description Requirements
Structure (in order):
- Hook — powerful first sentence, NO "В этом видео..."
- What viewer learns — key insights, methods, specific value
- Topic blocks — structured content overview
- Soft CTA — non-pushy
- Community link:
🔗 ОБЩАЕМСЯ ТУТ: https://t.me/YOUR_CHANNEL - Timecodes
- Hashtags for description
Timecode Requirements
Format: ММ:СС - Название секции
- Based on actual transcript content and timestamps
- Clear, informative section names (Russian)
- 8-15 sections depending on length
- First:
0:00 - Введениеor similar
Hashtag Requirements
Основные хештеги (для описания)
- Format:
#хештег1 #хештег2 ... - Mandatory branded:
#SPRUT #SPRUTAI - Thematic based on content
- 15-20 hashtags total
- Mixed language (Russian + English for reach)
Example:
#SPRUT #SPRUTAI #искусственныйинтеллект #ai #автоматизация #n8n #prompting #GPT4 #нейросети #aiagent #promptengineering
Хештеги для загрузки (upload tags)
- Format: comma-separated, NO # symbol
- Max 500 characters
- Can overlap with description hashtags
- Add mid/low-frequency queries
- Mixed language
Example:
искусственный интеллект, автоматизация, n8n, AI, агент, промпт инженеринг, ChatGPT, GPT-4, нейросети, AI автоматизация, prompt engineering
Key Reminders
- Specificity and value over abstract promises
- First 70 chars of title = strongest part
- Description should be "tasty" but not pushy
- ONE emoji in entire description (max!)
- Timecodes must match actual video structure
- Avoid all YouTube stop-words and restricted topics
🔗 Связанные скиллы
| Скилл | Зачем |
|---|---|
analytics | Перед оптимизацией SEO - посмотри yt-deep-stats.py: какие видео в топе, откуда трафик (поиск vs рекомендации), ср. удержание. Оптимизируй под реальные данные |
copywriter | Описание и заголовок в стиле владельца |
creator-marketing | Стратегия продвижения видео |
What ships with it: 1 file
510 B alongside SKILL.md, 1 of them executable
scripts/
- extract-audio.shruns510 B
Gives 0 of the 12 instructions most marketing audience skills give in ~1.8k tokens
Counted across 690 of the 894 authors here whose files we hold, read 2026-08-07
- Apply Poppins font to headingsin 41 of 690, across 6 files
- Apply Lora font to body textin 41 of 690, across 6 files
- Use Arial fallback for headingsin 39 of 690, across 4 files
- Use Georgia fallback for body textin 39 of 690, across 4 files
- Maintain text hierarchy and formattingin 39 of 690, across 4 files
- Use accent colors for non-text shapesin 38 of 690, across 3 files
- Use RGB values for precise color matchingin 38 of 690, across 3 files
- Use brand colors for primary text and backgroundsin 36 of 690, across 1 file
- Read product marketing context file before asking questions, starting, or auditingin 35 of 690, across 23 files
- Use active voice instead of passive voicein 26 of 690, across 10 files
- Implement or generate appropriate JSON-LD structured datain 24 of 690, across 17 files
- Prioritize clarity over clevernessin 22 of 690, across 8 files
Said here and by no other author read
- use hyphen instead of em dash always
- write strong verbs in short sentences
- extract audio from video
- transcribe media using OpenAI Whisper
- write five title variants with rationale
- include branded hashtags in the description
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.