From podcast
Generates Korean podcast episodes from URLs, tweets, articles, PDFs: analyzes sources, writes script, OpenAI TTS audio, MP4 conversion, YouTube auto-upload. Partial execution supported.
npx claudepluginhub team-attention/plugins-for-claude-natives --plugin podcastThis skill uses the workspace's default tool permissions.
Analyze sources, generate a Korean podcast script, produce audio via OpenAI TTS, and auto-upload to YouTube.
Guides Next.js Cache Components and Partial Prerendering (PPR) with cacheComponents enabled. Implements 'use cache', cacheLife(), cacheTag(), revalidateTag(), static/dynamic optimization, and cache debugging.
Guides building MCP servers enabling LLMs to interact with external services via tools. Covers best practices, TypeScript/Node (MCP SDK), Python (FastMCP).
Generates original PNG/PDF visual art via design philosophy manifestos for posters, graphics, and static designs on user request.
Share bugs, ideas, or general feedback.
Analyze sources, generate a Korean podcast script, produce audio via OpenAI TTS, and auto-upload to YouTube.
[Source Collection] → [Analysis/Fusion] → [Script Writing] → [TTS Generation] → [MP4 Conversion] → [YouTube Upload]
Collect and analyze user-provided sources. Processing by type:
api.fxtwitter.com (replace domain in X/Twitter URL)When 2+ sources are provided, always spawn parallel subagents for each.
# [Episode Title]
> [Duration] podcast script | [Date]
> Sources: [source list]
---
## Opening (1 min)
- Hook: one sentence on why this topic matters
- Introduce sources
- Lead with conclusion (state core message upfront)
## Body Part 1 (3 min)
- Deep analysis of first source/perspective
## Body Part 2 (3 min)
- Deep analysis of second source/perspective
## Fusion/Intersection (3 min)
- Emergent insights from combining sources
- Patterns, commonalities, contrasts
- Generalizable implications
## Closing (30 sec)
- One-sentence summary of core message
- Sign-off
<output-dir>/
├── script.md ← Script
├── episode.mp3 ← Audio
├── episode.mp4 ← Video (for YouTube)
└── metadata.json ← Title, description, tags, YouTube URL
The output directory can be any user-specified path. A sensible default is podcast/YYYY-MM-DD-[slug]/ relative to the current working directory.
Convert script to audio using scripts/generate_tts.py:
python3 <plugin-path>/skills/podcast/scripts/generate_tts.py \
--input <script.md path> \
--output <episode.mp3 path> \
--api-key <OpenAI API key>
Replace <plugin-path> with the actual path where this plugin is installed (use ${CLAUDE_PLUGIN_ROOT} if available, or the resolved plugin installation path).
Check OPENAI_API_KEY environment variable first. If not set, ask the user.
| Setting | Value | Note |
|---|---|---|
| Model | gpt-4o-mini-tts | Latest model with instructions support |
| Voice | marin | Best for Korean. cedar as alternative |
| Chunk size | 1500 chars | 2000 token limit, Korean ~1.5 char/token |
| Instructions | Auto-generated per script | See default below |
Default TTS instructions:
"따뜻하고 친근한 한국어 팟캐스트 호스트. 명확한 발음으로 또박또박 읽되, 자연스러운 억양과 적절한 감정을 담아서. 중요한 포인트에서는 약간 힘을 주고, 인용구에서는 톤을 살짝 바꿔서 구분감을 준다. 전체적으로 지적이면서도 편안한 분위기."
If the user specifies a tone, customize via --instructions.
Convert MP3 to MP4 with a static title card:
python3 <plugin-path>/skills/podcast/scripts/convert_mp4.py \
--input <episode.mp3 path> \
--output <episode.mp4 path> \
--title "Episode Title" \
--subtitle "Subtitle"
Generates a 1920x1080 video with dark background (#1a1a2e) and Korean title/subtitle overlay.
python3 <plugin-path>/skills/podcast/scripts/upload_youtube.py \
--video <episode.mp4 path> \
--title "Episode Title" \
--description "Description" \
--privacy unlisted
~/Downloads/client_secret_*.json or ~/.config/google/client_secret_*.json--token-path)unlisted (unless user specifies otherwise)After upload, report to user:
Done!
- Script: <path>/script.md
- Audio: <path>/episode.mp3
- Video: <path>/episode.mp4
- YouTube: https://youtu.be/VIDEO_ID (unlisted)
Play episode.mp3 with afplay so the user can listen immediately.
Users may request only part of the pipeline:
homebrew-ffmpeg/ffmpeg tap may be needed for full codec supportOPENAI_API_KEY env var or provided by user)/System/Library/Fonts/AppleSDGothicNeo.ttc for Korean text overlay. On other platforms, adjust FONT_PATH in convert_mp4.py