PrompTom
All skills

Carousels for Reels and TikTok

carousel-conveyorPROcontentsocial

Builds carousels around one recurring character: the face is generated once, captions are composited in code for free.

What it does

  • Triggers when you ask to build carousels for a character, stand up a new one, or schedule posts.
  • Sets the character up first: one anchor selfie and about six scenes made from it. That is the only place money is spent — roughly $0.60 per character.
  • Every carousel then reuses those photos, with the text composited on top in code. Rewriting a caption, changing the @handle or redoing the layout costs nothing.
  • Each scene is edited from the original selfie, never from another scene: editing an edit drifts the face, and after a few hops the character stops being recognisable.
  • Before anything is published it shows the finished slides and waits for approval — the check is that the face is the same on every slide. Every post is flagged as AI-generated.

Why you'd want it

In this kind of pipeline only face generation costs real money, and it is tempting to pay for it again every time. The skill does the opposite: the face is bought once per character, while captions, layout and scheduling are done in code for free. The other reason it exists is the ban on publishing unreviewed work: a face that drifted between slides is visible only to the eye, so the slides go to you first.

Where to put it

  1. 1Create the folder .claude/skills/carousel-conveyor in your project
  2. 2Put a SKILL.md file in it with the text below
  3. 3That's it. Claude Code loads the skill itself when a task matches the description

To make the skill available in every project rather than one, put it in ~/.claude/skills instead of the project folder.

SKILL.md file

# Carousel Conveyor

A factory for turning hook ideas into finished, scheduled photo carousels for a
recurring AI persona. It is the production half of the content system: the
**trendwatch** skill produces the hooks/formats; this skill renders them as
7-slide carousels in a persona's voice and face, then schedules them to TikTok.

> The script names below (`gen_persona_ref.py`, `cover_bank.py`, builders,
> `schedule_batch.py`) refer to a reference implementation — a few hundred lines
> of Python around the Gemini image API, PIL, and the Postiz API. The scripts are
> trivial to rebuild; the **operating principles and the prompt blocks are the
> transferable asset**.

The whole design is built around one economic insight (the "AI UGC army"
playbook popularized by Adrià Martinez): the only expensive step is generating

This skill's file is part of the PRO collection

Unlock access
Carousels for Reels and TikTok — PrompTom