Guides
Task-oriented walkthroughs for the things people actually do with kura: archiving a channel, capturing videos and playlists, choosing capture depth, building a transcript corpus, keeping an archive current, handling media and views, and setting your defaults once.
Each guide is built around a job rather than a flag: capturing a whole channel, pulling individual videos, playlists, and searches, choosing how deep to go, making a channel's spoken content searchable, re-running a capture without re-fetching, choosing what media and views to keep, and setting the defaults you reach for on every run. They assume you have worked through the quick start.
Archiving a channel
Capture a whole channel's uploads as a catalog or a playable vault, widen it with shorts, live streams, playlists, and community posts, and bound it by time and count.
Capturing videos and playlists
Archive a single video, a playlist in order, a search query, or a music album, each into its own self-contained repository.
Depth and streams
The meta, media, and audio depth model, selecting a stream format with yt-dlp grammar, the optional ffmpeg merge, and how a missing stream degrades a video without aborting the run.
Transcripts and search
Build a greppable spoken-word corpus from a channel's transcripts, capture transcripts only for a light archive, pick languages, and grep across an entire channel.
Incremental and resumable captures
Re-run a capture to fetch only what is new, resume an interrupted run or a half-downloaded video, upgrade a catalog to a vault, force a clean recapture, and pin the stamp for reproducible output.
Media and views
How thumbnails and streams are localised and deduped, choosing which views to render, the inert HTML watch page with its local video player, and re-rendering from stored JSON with no network.
Configuration
Set kura's defaults once with a config file and KURA_* environment variables, understand the defaults to config to env to flag precedence, tune the shared response cache, and point kura at an external yt-dlp or ffmpeg.