Get the latest updates from followed YouTube creators/podcasts.
List of followed channels
Data source:Zara's AI Learning Library
AI Education & Technical Deep Dives
| Podcast name | Channel |
|---|---|
| Andrej Karpathy | @AndrejKarpathy |
| Anthropic | @anthropic-ai |
| Lex Fridman | @lexfridman |
AI Products & Startups
| Podcast name | Channel |
|---|---|
| Lenny's Podcast | @LennysPodcast |
| Peter Yang | @PeterYangYT |
| The MAD Podcast (Matt Turck) | @DataDrivenNYC |
| Every | @EveryInc |
VC & Investors
| Podcast name | Channel |
|---|---|
| Y Combinator | @ycombinator |
| Latent Space | @LatentSpacePod |
| South Park Commons | @southparkcommons |
| No Priors | @NoPriorsPodcast |
| a16z | @a16z |
Big Tech & Research
| Podcast name | Channel |
|---|---|
| Google DeepMind | @googledeepmind |
| Google for Developers | @GoogleDevelopers |
| Stanford GSB | @stanfordgsb |
Vibe Coding & Tools
| Podcast name | Channel |
|---|---|
| Mckay Wrigley | @realmckaywrigley |
| Tiago Forte | @TiagoForte |
| The Pragmatic Engineer | @ThePragmaticEngineer |
AI News & Trends
| Podcast name | Channel |
|---|---|
| The AI Daily Brief | @TheAIDailyBrief |
| TBPN | @TBPNLive |
| Brett Malinowski | @TheBrettWay |
Quick Start
# 默认获取最近 2 天更新
python3 scripts/get_updates.py
# 获取最近 7 天(一周)的更新
python3 scripts/get_updates.py --days 7
# Markdown 格式输出(带可点击链接)
python3 scripts/get_updates.py --markdown
# 获取播放量(会增加请求时间)
python3 scripts/get_updates.py --views
# 组合使用
python3 scripts/get_updates.py --days 7 --markdown --views
Natural Language Mapping
When the user says:
- "get podcast updates" →
--days 2 - "get podcast updates from the past week" / "what new podcasts are there this week" →
--days 7 - "get podcast updates from the past three days" →
--days 3 - "get podcast updates for this month" →
--days 30
Output Content
Each podcast update includes:
- Title: clickable YouTube link
- Channel: source channel name
- Date: publish date
- Duration: video duration (use the --views parameter)
- View count: number of video views (use the --views parameter)
- Summary: a content summary of about 300-400 characters (automatically extracted from the video description)
Chinese Output Requirements
When presenting podcast updates to the user, the Agent needs to:
- Title translation: translate English titles into Chinese, preserving key terms
- Guest introduction: introduce the guest's background in detail (100-150 characters), including:
- Guest name and current position
- Their company/institution and its prominence
- The guest's representative works or achievements
- Why this person is worth following
- Content summary: describe in Chinese the main topics discussed in the podcast (150-200 characters)
- Recommendation reason: explain why this episode of the podcast is worth listening to
Output template
### {序号}. [{中文标题}]({链接})
**{频道}** | {日期} | ⏱ {时长} | 👁 {播放量}
**嘉宾介绍**:{嘉宾姓名} 是 {公司} 的 {职位}。{公司介绍}。{嘉宾成就}。{为什么值得关注}。
**内容摘要**:本期播客讨论了...(150-200字中文描述)
Workflow
When the user says "get podcast updates":
Step 1: Get the update list
Run the script to get updates from the past 2 days:
python3 /Users/ugreen/.claude/skills/youtube-feed/scripts/get_updates.py --days 2
Step 2: Display the update list
Show the user a formatted update list:
📺 最近 2 天共有 N 个新播客更新:
1. 【Lenny's Podcast】
📌 视频标题
🕐 2026-01-30 15:00
🔗 https://www.youtube.com/watch?v=xxx
📝 简要描述...
2. 【No Priors】
...
Step 3: User selection
Ask the user:
请选择要处理的播客(输入序号,如 1 或 1,3):
Step 4: Connect to podcast-workflow
After the user selects, call the podcast-workflow Skill for full processing:
- Extract subtitles
- Content-digest processing
- Save to Feishu
Add a new channel
Edit the scripts/get_updates.py in CHANNELS list:
CHANNELS = [
("频道名称", "channel_id", "@handle"),
# ...
]
How to get channel_id:
- Open the channel page
- View the page source and search for "channelId"
- Or use an online tool such as https://commentpicker.com/youtube-channel-id.php
Work with other Skills
This Skill is the first step in the podcast processing workflow:
youtube-feed → 用户选择 → podcast-workflow → 飞书
↓ ↓ ↓
获取更新 挑选播客 字幕+处理+发布
Complete workflow example
User: Get podcast updates
Agent:
- Run get_updates.py
- Display the update list
- Ask the user which one to process
User: Process the 2nd one
Agent:
- Call podcast-workflow
- Extract subtitles
- Ask whether to content-digest
- Process and save to Feishu