Creator templates

Six proven formats — TikTok highlight, podcast clip, UGC ad, product demo, cinematic story, music cutdown — applied in one prompt.

A creator template is a proven format the agent can execute in one shot: canvas direction, visual style, hook, and caption look — plus a playbook the agent follows to cut and finish the edit for that format. Instead of assembling a TikTok look piece by piece, you name the outcome and review the result.

Prompts to paste

"Turn this into a TikTok highlight — 30 seconds, best moments, captions."

"Make a 60-second podcast clip from the strongest insight in this interview."

"Turn this product footage into a UGC ad. Hook with the main benefit, CTA at the end."

"Cut a clean product demo from this screen recording — keep all the UI readable."

"Give this a cinematic story treatment, around a minute."

"Beat-led music cutdown, 20 seconds, high energy."

The six templates

Template	Canvas	Default length	Style pack	Caption skin
tiktok-highlight	portrait 9:16	30s	viral-cut	tiktok-bold
podcast-clip	portrait 9:16	60s	warm-founder	boxed-contrast
ugc-ad	portrait 9:16	35s	ugc-ad	boxed-contrast
product-demo	landscape 16:9	60s	clean-educator	minimal-clean
cinematic-story	keeps your current canvas	75s	cinematic-doc	documentary-lower
music-cutdown	portrait 9:16	20s	neon-gaming	neon-pop

ugc-ad and product-demo also include a CTA beat (default text "Try it today" / "Follow the steps") — override it or skip it in your prompt.

What the agent actually does

Applying a template itself makes four changes: sets the canvas direction, applies the template's style pack, inserts the hook overlay (you can replace the text or skip it), and applies the caption skin — but only if captions already exist. Templates never transcribe or generate captions on their own.

The rest comes from the template's playbook. Each template tells the agent which steps fit the format — for a TikTok highlight that's reviewing the footage, extracting highlights, face-focused reframing, styled captions, zoom punches, and sound design, then verifying the result. So an outcome-level prompt ("turn this into a TikTok") gets the full treatment, while "apply the TikTok template" on its own just gets canvas + style + hook.

Everything lands on the real timeline and is undoable.

Refining after the template

Templates are starting points — iterate conversationally:

"Swap the style to clean-educator but keep everything else."

"Different hook — quote the line about pricing instead."

"Tighter: get it under 25 seconds."

When to do it manually

You can build any of these looks yourself: set the canvas in project settings, add effects and text overlays, and style captions in the Properties panel. The template is the same treatment without the assembly.

Limits

Captions require audio with speech; the caption skin only applies once captions exist.
Hook text defaults are generic — the agent replaces them with something grounded in your transcript, but check the hook is specific to your content before publishing.
cinematic-story keeps your current canvas (it doesn't force 16:9 or widescreen).
A template applied twice overwrites the first application's style choices.