Verified against Claude · 2026-07-28
Pull the actual Shorts-worthy moments out of a long-form transcript
Identifies specific timestamp ranges in a long-form transcript that work as standalone Shorts, and rejects moments that only make sense with the surrounding context the clip would cut away — instead of mechanically clipping the video's "highlights."
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are identifying which specific moments in a long-form video's transcript are genuinely strong candidates for standalone YouTube Shorts, and rejecting ones that only seem interesting because of surrounding context a Short would cut away. Not every quotable or funny moment in a long video survives being extracted — the test is whether it works with zero prior context, not whether it was a highlight of the original. FULL TRANSCRIPT WITH TIMESTAMPS [2:14] ...and that's when I realized the whole recipe was wrong from the start. [2:31] So here's what actually works: you brown the butter FIRST, before anything else touches the pan. [2:58] Watch what happens when I do it the other way... ORIGINAL VIDEO TOPIC A 22-minute video troubleshooting why a classic cookie recipe kept coming out flat, testing five variables one at a time NUMBER OF CLIPS NEEDED 5 CHANNEL SHORTS STRATEGY Shorts are meant to drive discovery toward the long-form channel — prefer clips that clearly reference "the full breakdown" without giving away the full answer SELECTION RULES For each candidate moment, apply a strict test before including it: read only the 15-45 seconds around the clip's proposed boundaries, with zero surrounding context, and confirm it makes complete sense and lands its point on its own — if understanding it depends on something explained ten minutes earlier in the original video, reject it as a Shorts candidate even if it was a great moment in the full video. Prefer moments that already have a self-contained arc — a claim, then a demonstration or payoff, within the same short window — over moments that are just a strong isolated line with no setup or resolution nearby, since a strong line alone often needs the surrounding video to land and a self-contained arc doesn't. Note the exact in and out timestamps for each clip, and check the few seconds immediately before your proposed start point — if the actual sentence or thought genuinely begins slightly earlier than the "best" line, adjust the start point backward so the clip doesn't open mid-thought, which is a common editing mistake when clipping purely by "where the good part starts." Flag where a clip's context needs a single added on-screen caption to work without the surrounding video — for instance, naming what's being reacted to — rather than either forcing the clip to stand fully alone with no assist or silently including confusing unstated context. Rank the selected clips by how well each satisfies the zero-context test, not by how funny or dramatic the moment seemed in the original, since virality potential and originality of the point matter, but nothing outperforms actually being comprehensible in isolation. OUTPUT FORMAT For each of the requested number of clips: exact in/out timestamps, a transcript excerpt of just that range, a one-line note on why it passes the zero-context test, and any single caption needed to make it stand alone. Rank the full list from strongest to weakest zero-context candidate.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
The zero-context test is the actual mechanism this whole prompt is built around, because the single most common mistake in mechanical highlight-clipping is mistaking "this was a great moment in the full video" for "this works as a standalone piece of content" — a line that lands because of ninety seconds of buildup earlier in the video will confuse or simply fail to land for a Shorts viewer who has none of that setup and won't go find it, no matter how good the line sounded in context, which is why the rule is explicit that great-in-context and works-alone are different properties and only the second one matters for this format. Preferring a self-contained claim-then-payoff arc over an isolated strong line targets a subtler version of the same problem: an isolated quotable line often relies on the audience already caring about the stakes established earlier, while a moment with its own mini setup and resolution inside a 30-45 second window carries its own stakes and doesn't need to borrow interest from content the clip doesn't include. Checking the few seconds before the proposed start point for a mid-thought opening addresses a specific, common technical clipping error: identifying the "best line" first and cutting right before it frequently starts the clip mid-sentence or mid-thought, because the actual idea often begins a beat earlier than the specific words that made the moment memorable, and a viewer who opens on an already-in-progress thought loses the exact clarity the zero-context test is trying to protect. Flagging where a single caption could rescue an otherwise-borderline clip, rather than only accepting fully self-sufficient moments, matters because it distinguishes two genuinely different fixes: some content problems need a caption supplying one piece of missing context — who's being replied to, what's being reacted to — which is a legitimate and common Shorts editing technique, while other content simply cannot be made to work standalone at any caption budget and should be rejected outright rather than patched with an increasingly elaborate caption trying to compensate for content that fundamentally depends on its original context.
Verified against
Claude Sonnet 4.6 · 2026-07-28
ChatGPT GPT-5.1 · 2026-08-06
Changelog
- 2026-07-28 — Initial publish, verified against Claude (Sonnet 4.6) and ChatGPT (GPT-5.1).
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
