Who it is forA creator who posts photo carousels on TikTok and wants them generated from a consistent set of persona photos.
The problemChoosing photos, writing short bold lines and adding text to each slide is repetitive when you post carousels often.
The outcomeA set of 1080x1920 slide images with text on each, plus a zip file, ready to post as a photo post.
What you need first
- Docker Desktop, or Docker Engine with Compose v2
- Your own vertical B-roll clips, placed under assets/ or uploaded in the UI (ClipFactory ships no footage)
- An OpenAI key and an ElevenLabs key for the content factory, or Offline dry run with no keys to test the pipeline
- For the AI Lab: an image provider key and a video provider key (Google Gemini API and/or fal.ai)
Steps
- Run
docker compose up -dand open http://localhost:3000. - Add your photos under B-roll, then Photos, for the active persona.
- On Generate, pick the Slideshow template and enter your topic.
- Open the project page, review the 5 to 10 slides, and approve the set.
- Download the zip and post the images as a photo post in the TikTok app, then pick a trending sound there.
Prompt for your agent
Use the repo feyzilim/clipfactory with the Slideshow template to create a photo carousel for the persona <persona name> on the topic <topic>. Expected output is a set of 1080x1920 slide images plus a zip file. Verify that the zip contains one image per slide and that each image has text on it.
Caveat. ClipFactory exports the slides but does not post them, and the trending sound has to be chosen inside TikTok itself.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for feyzilim/clipfactory and "Build a TikTok photo carousel from a persona photo library"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer