Navi+ Longform Images
Use this skill whenever the task involves article images for longform.naviplus.io, especially generating or replacing files under images/{article-slug}/*.webp.
Goal
Create one complete 1280x720 editorial storytelling illustration per article figure. Avoid old placeholder-looking images: sparse generic diagrams, low-detail screenshots, tiny labels, flat stock-like icons, or files that compress to very small WebP sizes.
Visual System
Use this consistent visual family:
- clean modern flat/vector editorial SaaS and ecommerce illustration;
- warm off-white background;
- large editorial headline on the left;
- deep navy typography;
- one coral-red problem keyword when useful;
- bright blue UI accents;
- teal/green solved path, success state, or recommendation path;
- pale blue-gray UI cards;
- rounded browser, phone, and product frames;
- simple ecommerce product cards;
- soft shadows and subtle gradients;
- dotted paths, arrows, callouts, and attention rings;
- simple vector shopper/store-owner/UX character only when it clarifies the story;
- tiny bottom-right watermark in a translucent rounded white label:
@Navi+ AI Menu Builder | naviplus.io.
Prompt Template
Use a self-contained prompt like this:
Create one standalone 1280x720 horizontal blog illustration for longform.naviplus.io.
Style inspiration only: https://cdn.naviplus.app/AI/templates/chatGPT/ChatGPT-generate-images.webp
Article: "{article title}"
Section: "{section heading}"
Topic: {specific concept}
Headline text, large on the left: "{short editorial headline}"
Make the word "{problem_or_focus_word}" coral red; keep the rest deep navy.
Subtitle below headline: "{one-sentence explanation}"
Three chips below: "{chip 1}", "{chip 2}", "{chip 3}".
Main visual on center/right: {one coherent scene that explains this figure's section. Describe the UI, path, warnings, and solved state.}
Visual system: clean modern flat vector editorial SaaS/ecommerce illustration, warm off-white background, large editorial headline on the left, deep navy typography, coral red for the problem, bright blue UI accents, teal/green solved path, pale blue-gray cards, rounded device frames, soft shadows, subtle gradients, dotted paths and clear arrows.
Watermark: tiny faint bottom-right watermark inside a small translucent white rounded rectangle. Exact text: "@Navi+ AI Menu Builder | naviplus.io".
Negative constraints: no photorealism, no 3D render, no dark theme, no stock photo style, no dense dashboard, no collage, no random logos, no real brand names unless the article explicitly requires them, no tiny unreadable text, no non-English text, no footer bar.
Workflow
- Read the article section around each
<figure>. - Reuse the existing target filename unless the article markup needs correction.
- Generate one image per figure with the image generation tool.
- If the image contains random logos, brand names, unreadable dense text, wrong language, or a stock/3D look, regenerate before saving.
- Save the latest generated PNG to WebP:
python3 scripts/save_latest_generated_image.py images/{article-slug}/{filename}.webp
- Inspect dimensions and size:
identify images/{article-slug}/{filename}.webp
- Update article
alt,title, or caption only when the new image meaning differs from the old metadata. - Update
IMAGE_GENERATION_PROGRESS.mdwith replaced files, dimensions, sizes, and concepts. - Run:
bundle exec jekyll build
Validation
Use this scan to catch likely old placeholder assets:
find images -type f -name '*.webp' -size -30000c -exec ls -l {} +
After a complete replacement pass, this command should return no article illustration placeholders unless a small file is intentionally accepted and documented.