Prompt Details
Model
Chat - (gpt-6-luna)
Token size
1,227
Example input
[PASTE TRANSCRIPT HERE]: 00:13:12 Host: Chapters are the second job and they serve a different person than clips do. 00:13:23 Guest: Clips recruit strangers, chapters serve the people already watching. 00:13:34 Guest: A chapter list is a table of contents, and people scan it before they commit. 00:13:45 Host: If the list looks useful they stay, if it looks vague they leave. 00:13:56 Host: So chapter titles should name the question, not the segment number. 00:14:07 Guest: A title like part three tells the viewer nothing and costs you the watch. 00:14:18 Guest: A title like why the second week is harder tells them whether they need it. 00:14:29 Host: I start every chapter list at zero seconds, without exception.
Example output
CHAPTERS
00:13:12 | Chapters serve existing viewers
CLIP CANDIDATES
00:13:12 -> 00:13:34 | 9 | Clearly distinguishes the purpose of chapters from clips and closes by explaining that a chapter list works like a table of contents.
00:13:34 -> 00:13:56 | 8 | Explains that viewers scan chapter lists before committing and closes with the practical rule that titles should name the question.
00:13:56 -> 00:14:18 | 9 | Contrasts an unhelpful numbered chapter title with a specific title that tells viewers whether the section is relevant.
CLIP EVIDENCE
CLIP 1
QUOTE: "Clips recruit strangers, chapters serve the people already watching."
WHY IT STANDS ALONE: The quote directly explains the different roles of clips and chapters without needing earlier context.
CONTEXT NEEDED? no.
CLIP 2
QUOTE: "A chapter list is a table of contents, and people scan it before they commit."
WHY IT STANDS ALONE: The quote explains how viewers use chapter lists, and the range closes with a clear rule for naming chapters.
CONTEXT NEEDED? no.
CLIP 3
QUOTE: "A title like part three tells the viewer nothing and costs you the watch."
WHY IT STANDS ALONE: The range gives both the bad example and the contrasting specific-title example, making the point understandable by itself.
CONTEXT NEEDED? no.
COVERAGE CHECK
CHAPTERS: 1
TOTAL DURATION: 00:13:12 to 00:14:29
TIMESTAMPS VERIFIED AGAINST INPUT: yes
UNVERIFIED: first cue is not 00:00:00
Get this prompt + 9 more with SelectFrom $14/mo · Cancel anytimeFrom $14/mo · Cancel anytime
View SelectBy purchasing this prompt, you agree to our terms of service
GPT-6-LUNA
Reads a timestamped transcript of a long video and returns chapters that begin at 00:00:00, three to eight clip candidates of 20 to 60 seconds scored on whether they stand alone, and a written reason for each clip that quotes the transcript word for word.
Every timestamp is copied from the transcript and never calculated, so the output can be checked mechanically. Includes a timestamp format guide and a clip candidate table format.
...more
Added 1 week ago
