Social Post Carousel re-typesets a post as a set of 1080 by 1080 cards: black background, off-white text, gray attribution and one body size on every card. The agent plans the slides at paragraph boundaries, writes a small JSON spec and runs a Pillow script that writes the PNGs.
The SKILL.md asks for verbatim copy by default, a phone-scale contact sheet check and delivery of ordered PNGs. It does not post anywhere unless you separately authorize it.
What You Get
- Ordered slide-01.png, slide-02.png and so on at 1080 by 1080 pixels
- A slide plan that breaks the post at paragraph and idea boundaries
- A JSON spec of slides with optional name, handle and avatar
- A QA pass covering dimensions, order, clipping and blank cards
- A synthetic example spec to try the renderer
Common Use Cases
- Repurposing a long X or LinkedIn post as a swipeable carousel
- Producing cards from a screenshot by transcribing the visible text first
- Keeping a consistent look across many carousels without design work
- Checking a carousel at phone scale before sharing it
What Makes It Unique
The renderer is deterministic and offline. Type size stays fixed on every card, and a card that overflows raises an error so the copy is split, which keeps the set consistent. Cards carry no platform logo, verification badge or engagement counters, so they stay brand-neutral.
Under The Hood
The SKILL.md gives the agent a four-step method and one command. The agent writes a JSON spec file and runs a Python script from the skill folder, which needs Pillow and the paths of two fonts installed on the machine. The script reads the spec and an optional avatar image, and writes PNG files to a fresh output directory. It makes no network calls. In a full repository checkout the SKILL.md also allows a version check and consented usage logging through the repository helpers when network access is allowed.