AgentSkillsCN

gemini-imagen

使用Google Gemini的图像生成API通过Python CLI生成图像。当需要“生成图像”、“创建图像”、“gemini图像”、“AI图像”,或在Claude Code中需要AI图像生成时使用。

SKILL.md
--- frontmatter
name: gemini-imagen
description: >
  Generate images using Google Gemini's image generation APIs via Python CLI.
  Use for "generate image", "create image", "gemini image", "AI image",
  or when needing AI image generation in Claude Code.
version: 1.0.0
command: /imagen
routing:
  triggers:
    - generate image
    - create image
    - gemini image
    - AI image
    - text to image
  complexity: simple
  category: image-generation

Gemini Imagen

Generate images from text prompts using Google's Gemini APIs. This plugin gives Claude Code the ability to generate images directly.


Quick Start

bash
# Generate an image
python3 ~/.claude/plugins/gemini-imagen/skills/gemini-imagen/scripts/generate_image.py \
  --prompt "A cute cartoon cat" \
  --output cat.png

CRITICAL: Exact Model Names

Use ONLY these exact model strings:

Model StringSpeedBest For
gemini-2.5-flash-imageFast (2-5s)Drafts, iterations
gemini-3-pro-image-previewSlower (5-15s)Quality, text rendering, 2K

Common mistakes:

  • gemini-2.5-flash-preview-05-20 - WRONG (date suffixes are for text models)
  • gemini-2.5-pro-image - WRONG (doesn't exist)
  • gemini-3-flash-image - WRONG (doesn't exist)

Instructions

Step 1: Check API Key

bash
echo "GEMINI_API_KEY is ${GEMINI_API_KEY:+set}"

If not set, tell the user to run /imagen:setup.

Step 2: Install Dependencies

bash
pip install google-genai Pillow

Step 3: Generate Image

bash
python3 ~/.claude/plugins/gemini-imagen/skills/gemini-imagen/scripts/generate_image.py \
  --prompt "YOUR PROMPT HERE" \
  --output /path/to/output.png

Step 4: Verify Output

bash
ls -la /path/to/output.png

Model Selection

Use CaseModelWhy
Iterating on promptsgemini-2.5-flash-imageFast feedback (2-5s)
Final assetgemini-3-pro-image-previewBest quality
Game spritesgemini-2.5-flash-imageMany images, consistent
Text in imagegemini-3-pro-image-previewBetter typography
Batch generationgemini-2.5-flash-imageCost effective

Post-Processing Options

Remove Watermarks (--remove-watermark)

Removes bright pixels from image corners. Very useful for cleaning up generated images.

Background Transparency (--transparent-bg)

Converts solid-color backgrounds to transparent. Great for sprites and icons.

bash
python3 generate_image.py \
  --prompt "Character on gray background" \
  --output char.png \
  --remove-watermark \
  --transparent-bg

Batch Generation

Generate multiple images from a file:

bash
# prompts.txt (one per line)
python3 generate_image.py \
  --batch prompts.txt \
  --output-dir ./images/

Error Handling

ErrorSolution
GEMINI_API_KEY not setRun /imagen:setup
Rate limit (429)Wait 60s, script auto-retries
Content policy (400)Modify prompt
No image in responseAdd more detail to prompt
Pillow not installedRun pip install Pillow

Script Reference

Location: scripts/generate_image.py

ArgumentRequiredDescription
--promptYes*Text prompt
--outputYes*Output file path (.png)
--modelNoModel (default: gemini-3-pro-image-preview)
--remove-watermarkNoRemove corner watermarks
--transparent-bgNoMake background transparent
--bg-colorNoBackground hex color (default: #3a3a3a)
--batchNoPrompts file (one per line)
--output-dirNoDirectory for batch output

*Required unless using --batch

Exit Codes:

  • 0: Success
  • 1: Missing API key
  • 2: Generation failed
  • 3: Invalid arguments

What This Plugin CAN Do

  • Generate images from text prompts
  • Select between fast and quality models
  • Remove watermarks from images
  • Make backgrounds transparent
  • Batch generate multiple images

What This Plugin CANNOT Do

  • Use non-Gemini models (DALL-E, Midjourney, Stable Diffusion)
  • Generate video or audio
  • Bypass content policy restrictions