google-image

Google Image Skill

After calling generate_image, ALWAYS use the Read tool on the returned file_path to display the image inline to the user.

This skill provides the generate_image tool for AI image generation with Google Gemini.

Available Tools

generate_image

Generate an image from a text prompt using Google Gemini.

Use for:

• Creating illustrations or concept art
• Generating visual assets
• Visualizing ideas or concepts
• Creating placeholder images

Parameters:

• prompt (required): Image description (5-2000 chars)
• output_dir (optional): Custom output directory (default: generated-images/)
• aspect_ratio (optional): 1:1 (default), 16:9, 9:16, 4:3, 3:4
• image_size (optional): Output resolution — 1K (default, ~1024px), 2K (~2048px), 4K (ultra HD)
• model (optional): gemini-2.5-flash-image (default, fast) or gemini-3-pro-image-preview (higher quality, reasoning-driven)

Latency: ~5-20 seconds

Workflow

1. Call generate_image with the prompt
2. Parse the JSON response to get file_path
3. ALWAYS use the Read tool on file_path to display the image inline

Examples

Basic Image

Use generate_image with prompt: "A serene mountain lake at sunset with reflections"

Wide Format

Use generate_image with:
  prompt: "Panoramic view of a futuristic city skyline"
  aspect_ratio: "16:9"

High Resolution

Use generate_image with:
  prompt: "Detailed botanical illustration of a rare orchid"
  image_size: "4K"

Custom Output Directory

Use generate_image with:
  prompt: "Logo design for a coffee shop"
  output_dir: "assets/images"

Higher Quality Model

Use generate_image with:
  prompt: "Photorealistic portrait in oil painting style"
  model: "gemini-3-pro-image-preview"

Response Format

Returns JSON:

• file_path: Absolute path to the saved PNG file
• message: Instruction to use Read tool for inline display
• metadata: Model, aspect ratio, image size, prompt, execution time, file size

Configuration

Requires GEMINI_API_KEY environment variable. Get your key at: https://aistudio.google.com/apikey