Notion Infographic

Rylai Notes Infographic

Create a coherent infographic series from source material. This is an original Rylai workflow for Codex and Hermes. The visual direction is a general notebook-style aesthetic and is not an official Notion product or brand asset.

Workflow

1. Read the complete source before proposing images.
2. Extract the central promise and the smallest set of independent supporting ideas.
3. Honor a user-specified image count. Otherwise create three to eight images; use at most twelve.
4. Write a short series plan before generation:

[
  {
    "file": "infographic-01.png",
    "role": "cover",
    "headline": "Short visible headline",
    "message": "One idea the image must communicate",
    "scene": "Concrete visual metaphor or diagram"
  }
]

5. Generate images one at a time so the next prompt can correct consistency problems.
6. Inspect every result. Regenerate images with unreadable text, accidental logos, inconsistent characters, clipped content, or a different palette.
7. Deliver the ordered files and a one-line description of the series.

## Prompt Construction

Build each prompt from a fixed style block plus one variable scene block.

Fixed style block:

```tex
Rylai notebook infographic, editorial hand-drawn linework on warm white paper,
confident imperfect marker strokes, simple flat shapes, generous whitespace,
black ink with teal, coral, and golden-yellow accents, clear visual hierarchy,
friendly human proportions, minimal decoration, no gradients, no 3D rendering,
no photorealism, no logos, no watermark.

Variable scene block:

- State the one idea for this image.
- Describe a concrete subject, diagram, sequence, or comparison.
- Specify only the essential visible labels.
- Repeat any recurring character, clothing, props, and palette exactly.
- Add the requested aspect ratio and language.

Keep on-image text short. When exact typography matters, generate the artwork without text and add text later with a local deterministic editor.

## Series Structure

- Cover: topic and clear promise.
- Body: one independent idea per image.
- Summary: synthesis, checklist, or call to action when useful.

Do not add a cover or summary merely to increase the count. Merge overlapping ideas and split overloaded images.

## Runtime Mapping

- Codex: use `image_gen`.
- Hermes: use `image_generate` or the configured image backend.
- If a backend cannot preserve a consistent character, use the best approved image as a visual reference for later generations.
- Save outputs in the user's workspace with zero-padded names such as `infographic-01.png`.

## Quality Gate

- The image communicates one idea within three seconds.
- Text is readable at the target publishing size.
- Every claim comes from the supplied source or is clearly labeled as interpretation.
- Palette, line quality, character design, margins, and aspect ratio match across the series.
- No external logo, trademarked UI, remote asset, or creator signature appears unless the user supplied and authorized it.

Authored by Rylai for Codex and Hermes.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [Rylaispirit](https://github.com/Rylaispirit)
- **Source:** [Rylaispirit/rylai-codex-hermes-skills](https://github.com/Rylaispirit/rylai-codex-hermes-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.