nicepkg/ai-workflow

nano-banana

REQUIRED for all image generation requests.

ソースを見る

使い方

  1. インストールコマンドをコピーし、自分の環境へ導入する前にソースを確認します。
  2. 必要なツール、認証情報、対応環境を原文で確認します。
  3. 入力と成果物を指定し、出力を資料と照合します。
npx skills add https://github.com/nicepkg/ai-workflow --skill nano-banana

開始前の確認

  • 導入にはモデル契約、API枠、ファイルへのアクセス許可は含まれません。
  • スター数や導入数は活動の指標であり、独立した品質・安全性試験ではありません。
リポジトリの原文

見出し、例、コード、表、リンク、参照画像を含む原文を表示しています。

Nano Banana Image Generation

Generate professional images via the Gemini CLI's nanobanana extension.

When to Use This Skill

ALWAYS use this skill when the user:

  • Asks for any image, graphic, illustration, or visual
  • Wants a thumbnail, featured image, or banner
  • Requests icons, diagrams, or patterns
  • Asks to edit, modify, or restore a photo
  • Uses words like: generate, create, make, draw, design, visualize

Do NOT attempt to generate images through any other method.

Before First Use

  1. Verify extension is installed:
bash
   gemini extensions list | grep nanobanana
  1. If missing, install it:
bash
   gemini extensions install https://github.com/gemini-cli-extensions/nanobanana
  1. Verify API key is set:
bash
   [ -n "$GEMINI_API_KEY" ] && echo "API key configured" || echo "Missing GEMINI_API_KEY"

Command Selection

User RequestCommand
"make me a blog header"/generate
"create an app icon"/icon
"draw a flowchart of..."/diagram
"fix this old photo"/restore
"remove the background"/edit
"create a repeating texture"/pattern
"make a comic strip"/story

Available Commands

Note: Always use the --yolo flag to automatically approve all tool actions.

CommandUse Case
gemini --yolo "/generate 'prompt'"Text-to-image generation
gemini --yolo "/edit file.png 'instruction'"Modify existing image
gemini --yolo "/restore old_photo.jpg 'fix scratches'"Repair damaged photos
gemini --yolo "/icon 'description'"App icons, favicons, UI elements
gemini --yolo "/diagram 'description'"Flowcharts, architecture diagrams
gemini --yolo "/pattern 'description'"Seamless textures and patterns
gemini --yolo "/story 'description'"Sequential/narrative images
gemini --yolo "/nanobanana prompt"Natural language interface

Common Options

  • --yolo - Required. Auto-approve all tool actions (no confirmation prompts)
  • --count=N - Generate N variations (1-8)
  • --preview - Auto-open generated images
  • --styles="style1,style2" - Apply artistic styles
  • --format=grid|separate - Output arrangement

Common Sizes

Use CaseDimensionsNotes
YouTube thumbnail1280x720--aspect=16:9
Blog featured image1200x630Social preview friendly
Square social1080x1080Instagram, LinkedIn
Twitter/X header1500x500Wide banner
Vertical story1080x1920--aspect=9:16

Model Selection

Default: gemini-2.5-flash-image (~$0.04/image)

For higher quality (4K, better reasoning):

bash
export NANOBANANA_MODEL=gemini-3-pro-image-preview

Blog Featured Image Examples

bash
# Modern illustration style
gemini --yolo "/generate 'modern flat illustration of developer coding at laptop, purple and blue gradient background, minimalist style, no text' --preview"

# Professional photography style
gemini --yolo "/generate 'professional editorial photo of coffee cup next to laptop on wooden desk, morning sunlight, shallow depth of field, no text' --count=3"

# Tech/abstract
gemini --yolo "/generate 'abstract visualization of neural network connections, dark background with glowing blue nodes, futuristic style' --preview"

Icon Generation

bash
gemini --yolo "/icon 'minimalist app logo for productivity tool' --sizes='64,128,256,512' --type='app-icon' --corners='rounded'"

Diagram Generation

bash
gemini --yolo "/diagram 'user authentication flow with OAuth' --type='flowchart' --style='modern'"

Output Location

All generated images are saved to ./nanobanana-output/ in the current directory.

Presenting Results

After generation completes:

  1. List contents of ./nanobanana-output/ to find generated files
  2. Present the most recent image(s) to the user
  3. Offer to regenerate with variations if needed

Refinements and Iterations

When the user asks for changes:

  • "Try again" / "Give me options": Regenerate with --count=3
  • "Make it more [adjective]": Adjust prompt and regenerate
  • "Edit this one": Use gemini --yolo "/edit nanobanana-output/filename.png 'adjustment'"
  • "Different style": Add --styles="requested_style" to the command

Prompt Tips

  1. Be specific: Include style, mood, colors, composition details
  2. Add "no text": If you don't want text rendered in the image
  3. Reference styles: "editorial photography", "flat illustration", "3D render", "watercolor"
  4. Specify aspect ratio context: "wide banner", "square thumbnail", "vertical story"

Troubleshooting

ProblemSolution
GEMINI_API_KEY not setexport GEMINI_API_KEY="your-key"
Extension not foundRun install command from setup section
Quota exceededWait for reset or switch to flash model
Image generation failedCheck prompt for policy violations, simplify request
Output directory missingWill be created automatically on first run