| name | ai-news |
| description | Fetch and summarize recent AI news from curated RSS feeds (Hugging Face, VentureBeat, The Verge, OpenAI, Anthropic, DeepMind, etc.) and YouTube channels (Yannic Kilcher, Two Minute Papers, AI Explained, etc.). Also fetches full transcripts for specific YouTube videos. Use when the user asks about recent AI news, what's happened in AI lately, summaries of AI research or product announcements, or wants a digest of what's going on in the AI space. |
AI News
Fetch recent AI news from RSS feeds and YouTube channel feeds, then synthesize into a digest. Optionally fetch full transcripts for YouTube videos.
Setup (one-time)
cd ~/.letta/skills/ai-news/scripts && npm install
Workflow
1. Fetch recent news
npx tsx scripts/fetch-news.ts [--days 7] [--limit 5] [--sources all|rss|youtube]
- Outputs a JSON object with
items[] sorted newest-first
- Each item has:
source, type (rss/youtube), title, url, publishedAt, summary, videoId
- Failed sources are listed in
errors[] — safe to ignore individual failures
2. Synthesize a digest
Read the JSON output and write a digest grouped by theme (e.g. model releases, research, tools, policy). Lead with the most significant items. For YouTube videos, note they have transcripts available.
3. (Optional) Fetch a video transcript
For any YouTube item worth deeper coverage:
npx tsx scripts/fetch-transcript.ts <video-url-or-id> [--summary] [--lang en]
--summary returns first ~4000 chars (good for quick context)- Without
--summary, returns full transcript + timestamped segments
- Works with any YouTube URL format or bare video ID
Customizing Sources
See references/sources.md for the full list of default RSS feeds and YouTube channels, how to find channel IDs, and additional feed recommendations.
To add/remove sources, edit the RSS_FEEDS and YOUTUBE_CHANNELS arrays at the top of scripts/fetch-news.ts.
Tips
- Use
--days 1 for daily briefings, --days 7 for weekly digests
- Use
--sources rss if YouTube fetches are slow or failing
- Some feeds (arXiv) can be very high volume — use a lower
--limit or filter by keyword after fetching
- Transcripts require captions to be enabled on the video; auto-generated captions work fine
