Skip to content
kura

Quick start

From an empty terminal to a self-contained offline YouTube archive you can click through and play.

This walks the core loop: capture a channel, look at what landed on disk, serve it back, then deepen the capture to a playable vault and keep it up to date.

1. Capture a channel

kura archive @mkbhd

kura resolves the channel, streams its uploads through the free InnerTube surface, writes each video as JSON, downloads the thumbnails and transcripts beside it, and renders the HTML and Markdown views. At the default meta depth it captures no stream bytes, so this is a fast, small catalog. The summary tells you where the archive landed:

@mkbhd
  repo:        /home/you/data/kura/youtube/@mkbhd
  depth:       meta
  videos:      1480 total (+1480 new)
  transcripts: 1322
  range:       2009-03-21T... … 2026-06-17T...
  media:       1480 thumbs

2. Look at what landed

ls $HOME/data/kura/youtube/@mkbhd

videos/        # videos/<id>.json, the source of truth, plus sidecars
html/          # per-video inert watch pages
md/            # per-video Markdown with the inline transcript
media/         # localised thumbnails, avatars, banners (and streams at media depth)
_assets/       # kura.css
index.html     # the browsable archive home
README.md      # the Markdown index
channel.json   # the captured channel
manifest.json  # counts, depth, range, capture history, gaps

Open index.html directly in a browser and it renders offline, with no network.

3. Serve it back

kura serve runs a local static server so links, media, and the <video> range requests resolve exactly as they would on a real host:

kura serve $HOME/data/kura/youtube/@mkbhd
# open http://127.0.0.1:8080

4. Go deeper, then keep it fresh

A meta catalog reads and greps but does not play. Upgrade it to a vault: kura add --depth media fetches only the streams for the videos already on disk, merging separate video and audio with ffmpeg when it is present:

kura add @mkbhd --depth media

Now the watch pages carry a passive HTML5 <video> element over the local file and play offline.

Later, re-run with kura add to fetch only the uploads that are new since the last capture and re-render just the affected pages:

kura add @mkbhd

Where to go next

  • The guides cover archiving a channel, capturing videos and playlists, the depth and stream model, the transcript corpus, incremental re-capture, and media and views in depth.
  • The CLI reference lists every command and flag.
2 min read