Tubask/Docsv0.5.0

summarize_video

One-call understanding of a single video: summary paragraph, key points, chapters, quotable lines with timestamps, transcript excerpt, and view stats.

When to use

User wants to understand one specific video
Long talks, tutorials, podcasts, conference sessions
Need chapters + takeaways + a few citeable quotes in one response
Follow-up after youtube_query returned a video ID

When NOT to use

Multi-video creator analysis → use youtube_query(goal="advice"). Exact keyword hunt across full captions → use get_transcript.

Examples

Tool call

summarize_video(video_ref="https://www.youtube.com/watch?v=VIDEO_ID", depth="detailed")

Chat prompt

Break down this talk — main ideas, chapter outline, and 3 quotable lines with timestamps:
https://youtube.com/watch?v=...

Brief mode

summarize_video(video_ref="dQw4w9WgXcQ", depth="brief", include_timestamps=true)

Parameters

Parameter	Type	Default	Description
video_ref*	string	—	YouTube URL, youtu.be link, shorts URL, or 11-character video ID.
depth	brief \| detailed	brief	brief = shorter summary; detailed = richer key_points and more quotes.
include_timestamps	boolean	true	Include timestamps on chapters and quotes.
languages	string[] \| null	null	Preferred caption languages, e.g. ["en", "en-US"]. Auto-detect if null.

Response fields

Typical JSON/text structure in the result:

summary — prose overview of the video
key_points[] — bullet takeaways
chapters[] — topic segments with start times
quotes[] — notable lines with timestamps for citation
transcript_excerpt — representative snippet (not full captions)
metadata — title, channel, duration, views
next_steps[] — e.g. get_transcript for a specific keyword

Videos without captions

Summaries work best when captions exist (auto-generated or manual). If a video has no captions, you still get metadata and description-based summary, but chapters and quotes may be sparse. Check next_steps for guidance.

Quota note

Metadata uses YouTube Data API units. Transcript portion of the summary does not count against Data API quota. See Quota & billing.