One corpus.
Every source.
Uoink starts with video and podcasts.
Articles and threads join the same local library.
Same source card. Same citations.
YouTube
Capture: Button, right-click, playlist queue, and Shorts.
Corpus: Transcript, frames, comments, metadata, channel context.
Best for: Creator research, hook study, lecture notes, agent citations.
Open source page ->
shippedX video
Capture: Right-click video posts and capture the source with creator credit.
Corpus: Transcript, post URL, author handle, media metadata, local citation.
Best for: Credited threads, clip research, launch swipe files.
Open source page ->
shippedPodcasts
Capture: Add an RSS feed or episode URL and transcribe with local Whisper.
Corpus: Speaker labels, transcript, episode metadata, show context, entities.
Best for: Interview notes, claim trails, guest research, agent search.
Open source page ->Reddit threads
Capture: Save a thread and keep the comment tree as source material.
Corpus: Post body, nested comments, authors, score context, outbound links.
Best for: Audience language, objection mining, product research.
Tracked on roadmap ->Web Articles & Pages
Capture: Universal page extraction strips navigation, headers, ads, and trackers.
Corpus: Clean body markdown text, author details, metadata, outbound links.
Best for: Writing anchors, source-backed essays, style study.
Tracked on roadmap ->
The source changes. The library compounds.
A video, a podcast, and a newsletter issue can all mention the same person, tool, company, or idea. Uoink keeps those mentions on disk so your agent can search across the whole library later.