images_skill
- TypeScript
1
GitHub Stars
9
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill cdeistopened/skill-stack --skill images- gen_4s_concepts.py3.9 KB
- gen_4s_styles.py7.3 KB
- gen_4s.py2.4 KB
- gen_batch_thumbnails.py6.2 KB
- gen_prism_styles.py5.1 KB
- gen_prism_v2.py5.3 KB
- generate_image.py5.8 KB
- skill-stack-style.md3.8 KB
- SKILL.md5.4 KB
Overview
This skill generates on-brand blog thumbnails for Skill Stack using the Gemini image API and a strict editorial aesthetic. It follows an iterative workflow: brainstorm 4–5 concepts, get approval, expand a prompt, generate the image, then review and iterate. The thumbnails are minimal, paper-inspired, and use a single terracotta accent.
How this skill works
I read the post title and content to find a concise visual metaphor that fits the paper/document aesthetic. I produce 4–5 one-sentence thumbnail concepts for approval, then expand the chosen concept into a full Gemini prompt with composition, color palette, texture, and avoid rules. After you approve the prompt I generate the image via the Gemini API and save it to the thumbnails folder for review and iteration.
When to use it
- When you need a consistent, editorial thumbnail for Skill Stack posts
- When a minimal, paper-inspired visual will help explain a post’s core idea
- When you want a fast concept-and-approve workflow before image generation
- For posts where brand consistency and single-color accents matter
- When thumbnails must be ready in 16:9 landscape for blog headers
Best practices
- Always read the post to extract the single core concept before brainstorming
- Produce 4–5 concrete, non-generic visual concepts and wait for approval
- Use the provided prompt template: composition, palette, texture, avoid list, and 16:9 format
- Limit the terracotta accent to exactly one element per image
- Request specific, narrow edits (composition, accent placement, element count) rather than vague feedback
Example use cases
- Framework post: stacked paper cards representing steps, top card terracotta
- Philosophy piece: crumpled-to-crisp paper before/after metaphor with terracotta highlight
- Tool tutorial: simple paper interacting with a geometric tool icon in neutral grays
- How-to guide: a single focused action (folding/stacking) with generous negative space
- Newsletter hero: editorial paper stack on warm cream background with one terracotta focal piece
FAQ
Yes — I always present 4–5 one-sentence concepts and wait for your approval before creating a prompt or generating images.
How strictly should I follow the color rules?
Strictly: background warm cream, layered paper grays, and terracotta on one element only. This ensures brand consistency across the site.
What if the first generation isn’t right?
Provide targeted feedback (e.g., move focal element to right third, reduce elements to one, remove texture). I’ll revise the prompt and regenerate until it matches the intended vibe.
