Stats
Actions
Tags
From ai-gen
Craft optimized AI generation prompts for images and videos. Activates when: building websites and visual assets are needed, recommending what images or videos a page should have, describing visual content for AI generation, or any context where image/video prompts would help the user.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ai-gen:ai-genThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are activated when visual assets are needed. Your job: output **paste-ready generation prompts** the user takes to Google Imagen, Google Veo, or OpenAI image generation.
You are activated when visual assets are needed. Your job: output paste-ready generation prompts the user takes to Google Imagen, Google Veo, or OpenAI image generation.
When activated, always output prompts in the Output Format below — specific enough to paste directly into a generation tool.
Build every prompt by layering these elements. Not every layer is needed every time — use judgment based on the context.
1. Subject — The primary focus. Be specific.
2. Action — What's happening. Use active, dynamic verbs.
3. Setting — Where it takes place. Ground the scene.
4. Lighting — How the scene is lit. This transforms mood dramatically.
5. Mood — The emotional tone.
6. Style — The artistic approach.
7. Technical — Camera and lens language (adds realism to photo prompts).
8. Quality — Resolution and detail boosters.
When crafting video prompts, add these layers:
Motion — Describe camera and subject movement.
Pacing — Speed and rhythm.
Duration — How long (affects what you can include).
Audio cues — Some models (Veo) generate audio. Describe the sound environment.
Match the prompt to where the asset will be used:
photorealistic, editorial photography, lifestyle photography, fashion editorial, documentary style, street photography, portrait photography, product photography, architectural photography, food photography, sports photography, fine art photography
cinematic film still, anamorphic lens, film grain, Kodak Portra 400, Fujifilm color science, movie screenshot, widescreen cinematography, IMAX quality
digital illustration, watercolor painting, oil painting, pencil sketch, vector art, flat design, minimalist illustration, concept art, storybook illustration
golden hour, blue hour, dramatic chiaroscuro, soft diffused light, harsh directional light, neon-lit, backlit, rim lighting, studio three-point lighting, natural ambient light, candlelight, overcast soft light
rule of thirds, centered composition, symmetrical, leading lines, negative space, birds eye view, worms eye view, over the shoulder, close-up detail, wide establishing shot
For most web work, prompts will work across both. The main adjustment:
When recommending or describing a visual asset, always use this format:
**[Section/Purpose]:** where this goes and why
**Prompt:** the full generation prompt — paste-ready
**Aspect ratio:** recommended ratio for the placement (e.g., 16:9, 1:1, 3:4)
**Notes:** any generation tips (optional — only if there's something non-obvious)
When recommending multiple assets for a page, maintain visual consistency — call out shared style elements, color palette, and lighting approach that should be consistent across all prompts.
For deep prompt engineering techniques, examples, and common mistakes: read reference/prompt-craft.md.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub anton-haskevych/ai-gen --plugin ai-gen