ai-gen
A Claude Code skill plugin for AI image and video generation prompt engineering.
What it does
Teaches Claude how to craft optimized generation prompts for Google (Imagen, Veo) and OpenAI (DALL-E, GPT-image) models. Claude automatically activates this skill when visual assets are needed — during web development, page design, content planning, or any context where images or videos would help.
This skill does not call any API. It outputs paste-ready prompts you take to your generation tool of choice.
Install
# From GitHub
/plugin install gh:anton-haskevych/ai-gen
# Or copy locally
cp -r . ~/.claude/skills/ai-gen
How it works
When Claude recognizes that visual assets are needed, it:
- Identifies what kind of asset fits the context (hero image, team portrait, background video, etc.)
- Builds an optimized prompt using an 8-layer architecture (Subject, Action, Setting, Lighting, Mood, Style, Technical, Quality)
- Adapts the prompt style for Google or OpenAI models
- Outputs the prompt in a standard format with aspect ratio and placement notes
Example output
When building a dance studio website, Claude might output:
[Hero Section]: Full-width hero image for the landing page
Prompt: A couple performing an elegant waltz in a grand studio with warm hardwood floors and soft golden light streaming through tall windows, romantic and inviting mood, editorial photography style, shot on 35mm with shallow depth of field, high resolution professional photography
Aspect ratio: 16:9
Notes: Leave negative space on the left for headline text overlay
License
Apache 2.0
anton-haskevych
anton-haskevych
