AgentSkillsCN

nano-banana-pro-3

通过 OpenRouter API 使用 Gemini 3 Pro 生成或编辑图像。适用于用户说“生成图像”、“创作图片”、“编辑照片”或“制作一张图像”时使用。支持文本转图像、图像编辑以及图像合成等功能。

SKILL.md
--- frontmatter
name: nano-banana-pro-3
description: Generate or edit images using Gemini 3 Pro via OpenRouter API. Use when user says "generate image," "create picture," "edit photo," or "make an image." Supports text-to-image, image editing, and compositing.

Nano Banana Pro 3

Generate or edit images using Gemini 3 Pro Image model via OpenRouter.

Quick Start

Full path: uv run .opencode/skill/nano-banana-pro-3/scripts/gen_image.py

bash
# Show help
gen_image.py --help

# Dry run (no API cost)
gen_image.py --dry-run --prompt "A sunset over mountains"

# Generate image
gen_image.py --prompt "A sunset over mountains"

# Edit existing image
gen_image.py --prompt "Make the sky dramatic" --input-image photo.jpg

Important: Run from user's working directory so images save there.

Workflow

  1. •

    Determine operation type:

    • •Generate new → --prompt only
    • •Edit single image → --prompt + --input-image
    • •Compose multiple → --prompt + multiple --input-image
  2. •

    Use --dry-run first to verify parameters (no API cost)

  3. •

    Run the generation

  4. •

    Report output path to user (do NOT read the image back)

API Key

Set OPENROUTER_API_KEY in .env or environment variable.

bash
echo "OPENROUTER_API_KEY=your-key-here" > .env

Get key at: https://openrouter.ai/keys

Common Failures

ErrorSolution
No API key providedSet OPENROUTER_API_KEY in .env
No image generatedModel returned text; try different prompt
API errors (403, quota)Check credits at openrouter.ai

Output

  • •Default directory: ./EXPORT/
  • •Default filename: yyyy-mm-dd-hh-mm-ss-generated.png

References

  • •All flags & options: See references/options.md
  • •More examples: See references/examples.md