youtube-feed_skill

Overview

This skill fetches recent uploads from a curated list of AI and tech YouTube channels and returns new podcast/video updates from the last N days. It lists each new item with title, publish date, brief description, and optional duration and view counts. It's designed for daily checks of followed AI/tech podcasts and creators.

How this skill works

The skill runs a script that polls the configured channel list and collects videos published within the requested window (default 2 days). For each video it gathers metadata: title, channel, publish date, description, and optionally duration and view count. It can output plain text or Markdown links and maps natural language requests like "recent podcast updates" to the appropriate time window. Selected items can be forwarded to a downstream podcast workflow for subtitle extraction and content digesting.

When to use it

• Daily quick check for new AI/tech podcast episodes
• Before meetings or content reviews to spot fresh interviews
• When you want a short list of recent uploads from followed creators
• To prepare selection for deeper processing (transcripts, summaries)
• When curating a newsletter or feed of recent AI discussions

Best practices

• Run with default --days 2 for daily monitoring; increase with --days 7 for weekly sweeps.
• Use --markdown when you need clickable links for sharing or publishing.
• Enable --views only when you need popularity signals, as it increases the request time.
• Keep the CHANNELS list updated to add or remove creators and use canonical channel IDs for reliability.
• After selecting videos, hand off to the podcast-workflow for transcript extraction and content-digest processing.

Example use cases

• Daily morning command: get the last 48 hours of AI podcast uploads and skim titles and summaries.
• Weekly roundup: run with --days 7 and generate Markdown links for a newsletter.
• Pre-interview prep: check if a guest has a new conversation published in the last few days.
• Curate candidates to send into the podcast-workflow for automated transcript and summary production.
• Spot emerging interviews from major research labs (DeepMind, Anthropic) to flag for team attention.