YouTube video → context for an LLM
A raw transcript dilutes the model's attention. We return a structured summary — so the agent's answer quality goes up.
On every video, without leaving YouTube → Chrome extension
Use skribr on every video
Free Chrome extension — on every YouTube video.
FAQ
Why not just paste the full transcript?
20-30k tokens of raw text dilute attention (Lost-in-the-Middle) and lower answer quality; a ~2k distilled summary keeps the model sharp.
Which models does it suit?
Any — ChatGPT, Claude, Gemini or your IDE agent. The output is plain context you paste in.