| name | youtube-digest |
| description | Fetch latest videos from YouTube channels, extract transcripts, and generate a digest with key insights. Use when asked for YouTube digest, channel updates, or video summaries. |
| homepage | https://github.com/hesamsheikh/awesome-openclaw-usecases |
| metadata | {"openclaw":{"emoji":"📺","requires":{"bins":["uv"]},"primaryEnv":"YOUTUBE_API_KEY","install":[{"id":"uv-brew","kind":"brew","formula":"uv","bins":["uv"],"label":"Install uv (brew)"}]}} |
YouTube Digest
Fetch latest videos from YouTube channels, extract transcripts, and produce a digest.
Generate a channel digest
uv run {baseDir}/scripts/digest.py --channels "@Fireship,@lexfridman,@ThePrimeTimeagen" --hours 48
With transcripts and summaries:
uv run {baseDir}/scripts/digest.py --channels "@TED,@Fireship" --hours 48 --transcript --max-videos 10
Keyword search mode:
uv run {baseDir}/scripts/digest.py --search "OpenClaw,AI agents" --hours 72 --transcript
JSON output:
uv run {baseDir}/scripts/digest.py --channels "@Fireship" --output json
Features
- Resolves
@handle to channel ID via YouTube Data API v3
- Fetches recent uploads via channel uploads playlist
- Extracts transcripts via
youtube-transcript-api (free, no API key, no binary)
- Tracks seen videos in
--seen-file to avoid reprocessing
- Outputs markdown with title, channel, link, date, and transcript bullets
Seen-video tracking
uv run {baseDir}/scripts/digest.py --channels "@Fireship" --seen-file seen-videos.txt
Only new (unseen) videos are processed. After processing, video IDs are appended to the file.
Cron
Every morning at 8am, fetch latest videos from @TED, @Fireship, @lexfridman and give me a digest with key insights.
Save my channel list to memory so I can add/remove channels later.
API keys
Notes
- Transcript extraction is free (no API key, no credits) via
youtube-transcript-api.
channel/latest listing requires YOUTUBE_API_KEY.- Use timestamps in filenames:
yyyy-mm-dd-yt-digest.md.
