Claude Code plugin for Gemini-powered image generation, video generation, editing, and visual analysis. Powered by Gemini 3.1 Pro (gemini-3.1-pro-preview), Nano Banana Pro (gemini-3-pro-image), Veo 3.1 (veo-3.1-generate-preview), and Gemini 3 Flash. 169 art styles, text-to-video, image-to-video, pixel avatars, presentation decks, and more.
Every image and video on this page was generated using gemskills.
/plugin marketplace add b-open-io/claude-plugins
/plugin install gemskills@b-open-ioIndividual skills (for other agentic frameworks)
bunx skills add b-open-io/gemskills --skill generate-image
bunx skills add b-open-io/gemskills --skill generate-video
bunx skills add b-open-io/gemskills --skill browsing-styles
bunx skills add b-open-io/gemskills --skill avatar-portrait
bunx skills add b-open-io/gemskills --skill team-group-photo
bunx skills add b-open-io/gemskills --skill generate-icon
bunx skills add b-open-io/gemskills --skill edit-image
bunx skills add b-open-io/gemskills --skill upscale-image
bunx skills add b-open-io/gemskills --skill segment-image
bunx skills add b-open-io/gemskills --skill optimize-images
bunx skills add b-open-io/gemskills --skill generate-svg
bunx skills add b-open-io/gemskills --skill section-dividers
bunx skills add b-open-io/gemskills --skill deck-creator
bunx skills add b-open-io/gemskills --skill ask-gemini
bunx skills add b-open-io/gemskills --skill setupRequirements: GEMINI_API_KEY (get one) is the baseline and powers every skill.
Optional keys unlock additional providers for image/video/edit:
| Key | Unlocks | Get one |
|---|---|---|
GEMINI_API_KEY |
Gemini Nano Banana Pro images, Veo 3.1 video, all 169 styles (default) | aistudio.google.com |
OPENAI_API_KEY |
OpenAI gpt-image-2 image generation + masked editing |
platform.openai.com |
XAI_API_KEY |
xAI Grok Imagine image + grok-imagine-video-1.5 video |
console.x.ai |
REPLICATE_API_TOKEN |
Icon background removal; Veo reference-image / last-frame video | replicate.com |
generate-image, generate-video, and edit-image accept --provider gemini|openai|xai.
Omit it and gemskills auto-picks the best provider whose key is present and that
supports the request — e.g. plain image → gpt-image-2, video → grok-imagine-video-1.5,
but anything needing style tiles, reference images, transparency, or negative
prompts routes to Gemini (the only provider that supports them).
Set your own defaults interactively with /gemskills:setup (or the setup
skill), or pin them via env (GEMSKILLS_IMAGE_PROVIDER, GEMSKILLS_VIDEO_PROVIDER,
GEMSKILLS_EDIT_PROVIDER) or a .gemskills.json (project) / ~/.config/gemskills/config.json
(global) file. Per-provider prompt templates live in providers/prompts/ and are
tuned independently. Keys are read from one canonical env var each — gemskills
never falls back to alternate names; a missing key fails loudly.
Generate videos from text or animate existing images with Veo 3.1. Native audio, 720p/1080p/4K, 4-8 second clips, and all 169 art styles.
| Style | → | Generated Image | → | Generated Video |
 --style kusm |
+prompt |  Nano Banana Pro |
+prompt |  Veo 3.1 · 8s |
 --style impr |
+prompt |  auto-generated |
+prompt |  Veo 3.1 · 8s |
The two-step pipeline gives full creative control: generate a styled image first, then animate it. Or use --auto-image to do both in one command.
Prompt alone → complete scene with audio. No input image needed.
"Ocean waves crashing on dark volcanic rocks at golden hour..." · 8s · ~67s gen
Text-to-video demo | Image-to-video demo | Auto-image pipeline demo
Every image and video generation supports --style to apply any of 169 curated art styles. Each style includes an AI-generated tile reference image sent to Gemini alongside the prompt for dramatically better style adherence.
Kusama kusm |
 H.R. Giger gigr |
 Spider-Verse spdr |
 Ukiyo-e ukiy |
 Murakami mrkm |
 Stained Glass stgl |
 Wes Anderson wesa |
 Studio Ghibli ghbl |
Impressionism impr |
 Cyberpunk cybr |
 Art Deco deco |
 Vaporwave vapr |
 Made of Sand sand |
 Claymation clay |
 Film Noir fnoi |
 Steampunk stpk |
 Pixel Art pixl |
 Solarpunk solr |
Transform headshots into styled portraits, then composite into group scenes.
| Maya Creative Director |
Kai Lead Engineer |
Yuki Art Director |
Carlos Audio Director |
 |
 |
 |
 |
 |
 |
 |
 |
Generate OG images cropped to platform specs, or production-ready icon sets for iOS, Android, web, and desktop from a single prompt.
| Input | OG 1200x630 | Master Icon | No BG | App Store |
 |
 |
 |
 |
 |
Generate complete pitch decks with consistent visual style, then restyle with any of the 169 art styles.
| 01 Title | 02 Problem | 05 Portfolio | 09 Traction |
 |
 |
 |
 |
Same deck, restyled with --style pixl |
|||
 |
 |
 |
 |
Use ask-gemini for prompt refinement, design critique, and full page redesigns from HTML + inspiration screenshot.
Rough concept → Gemini refinement → Generated image (the hero image above was made this way)
HTML + inspiration → Redesigned code (input → output)
| Skill | Description |
|---|---|
| generate-image | AI image generation with 169 art styles, multi-reference (up to 14 images), img2img |
| generate-video | Text-to-video, image-to-video with Veo 3.1, native audio, auto-image pipeline |
| browsing-styles | Browse, search, and preview all 169 art styles |
| avatar-portrait | Likeness-preserving avatar portraits in any requested style |
| pixel-avatar | Compatibility alias for pixel-style avatar requests |
| team-group-photo | Individual styled portraits composited into group scenes |
| generate-icon | Platform icons (favicon, iOS, Android, PWA, desktop) with auto sizing |
| edit-image | Inpainting and outpainting with masks |
| upscale-image | 2x/4x upscaling via Vertex AI |
| segment-image | Object identification and extraction |
| optimize-images | Batch compress PNGs/JPEGs for web using sharp |
| generate-svg | Vector graphics, logos, and icons |
| section-dividers | Transparent decorative dividers for web sections |
| deck-creator | Complete presentation decks with consistent visual style |
| ask-gemini | Text + image queries for design critique, prompt refinement, spatial analysis |
# Generate an image with art style
"Generate a mountain landscape in watercolor style"
# Generate a video from text
"Generate a video of ocean waves crashing on rocks"
# Animate an image into video
"Turn this image into a video with gentle motion"
# Create styled avatar from photo
"Create a stylized avatar portrait from my headshot"
# Ask Gemini to refine a prompt
"Ask Gemini to write a better prompt for: a futuristic city at sunset"
# Create presentation slides
"Create a pitch deck for my startup"
# Generate app icons for all platforms
"Generate a favicon for my website with a lightning bolt"- Nano Banana Pro: Google's latest image generation model with thinking capabilities
- Veo 3.1: Text-to-video and image-to-video with native audio generation
- 169 art styles: Curated style library with AI-generated tile references for visual adherence
- Multi-image input: Up to 14 reference images for character consistency and scene composition
- Spatial reasoning: Superior visual understanding for design feedback
- Dynamic docs: Fetch latest API docs via llms.txt
MIT