Capture the source.
Transcript with timestamps, screenshots, description, title, thumbnail, channel context, and comments in one markdown file.
See the workflow ->
uoink.app / v3.3
Uoink that shit.
Local corpus for creators and AI developers. One click saves the videos, podcasts, and articles you study to your own disk, then hands them to your AI as a cited corpus you write from in your voice.
Open source / MIT / zero telemetry / 64 MCP tools / Windows now / Mac build is queued after Windows stabilizes
karpathy-intro-to-llms.mdlocal corpus
# Intro to Large Language Models
source: YouTube / channel: Andrej Karpathy
saved: Desktop/Uoink/AI-and-ML/karpathy-intro-to-llms/
## Metadata
duration, views, upload date, channel context, source URL
## Transcript
[00:47] hook classified as curiosity_gap + stakes
## Screenshots
12 frames on disk / 4 in clipboard budget
## Comments
top 50 comments with themes and disagreements
one uoink / readable by Claude, ChatGPT, and agents
uoink dashboardlocal corpus
what you get per uoink
Why paste a transcript when you can feed the entire corpus?
Transcript-only tools flatten video into text. Uoink keeps the parts that make video useful: the words, the frames, the audience reaction, the channel context, and the source metadata.
Keep the asset.
Every capture writes to disk and into a local SQLite index. Search it later, cite it later, move it into your own vault.
Browse features ->Hand it to AI.
Paste into Claude or ChatGPT, send directly, or let an MCP agent call Uoink without touching the clipboard.
Open developer docs ->
local-first, by design
Read privacy ->
No account. No Uoink cloud. No telemetry.
The helper runs on localhost. Your corpus lands on your disk. Optional Hook Type, Comment Intelligence, and Entity Extraction calls use your own Anthropic key, not a Uoink proxy.
three doors, one corpus
Creators and developers use the same source card.
The card stays the shared object. The next move changes by job.
01
Creators.
Skip the rewatch. Audit hooks, pacing, screenshots, and comments. Write threads and posts from your sources in Writing Studio with creator credit baked in.
See creator workflow -> 02Developers.
Treat your distribution loop like a local dev tool. Get an offline SQLite database, an MCP server, and copyable Claude/Cursor configs to query your library.
Open developer docs -> 03Sources.
Uoink captures video, audio, and text. Index YouTube, podcasts, newsletters, and social thread discussions directly on your workstation.
Browse supported sources ->
from source to draft
Pick a source, then generate from it.
Choose something you saved, aim it at a tweet, thread, or script, and Writing Studio drafts from the corpus with the creator credit attached.
generate flowsilent loop
model agnostic
Uoink this video and compare the hook against my last ten saved competitor videos.
Calling uoink_video, then classify_hook, then search_uoinks. Your model reads the corpus. Uoink stays the capture layer.
mcp tool trace
uoink_video(url)
get_job_status(job_id)
get_uoink_corpus(corpus_id)
classify_hook(corpus_id)
find_mentions("Karpathy")
install in a minute
One helper. One extension. Then the U button.
Download the helper, install the extension, open a video, and click Uoink. The helper bundles Python, yt-dlp, and ffmpeg so you never install them yourself.
Windowslive path
Download the helper.
Uoink installer / Windows 10 and 11
Runs in your tray, writes to your local library, and exposes the local MCP server.
Extensionpending review
Install the browser button.
Chrome / Edge / Brave / Vivaldi / Arc / Opera GX
Chrome Web Store approval is pending. Sideload the extension by loading the unpacked folder located at %LOCALAPPDATA%\Uoink\extension.
Macqueued
Mac build queued.
DMG, Keychain, LaunchAgent
Mac build is queued after Windows stabilizes. The macOS path will use the same corpus format and MCP surface.
OINK
Take the video. Make it usable.