メニュー
Generate or edit images with Google's Nano Banana Pro image model through the bundled Gemini API script. Use for text-to-image generation, image-to-image edits, or requests for 1K, 2K, or 4K Gemini image output.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/marcus/marcus-skills --skill nano-banana-proコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
Present technical information, decisions, tradeoffs, and code in chat using Google Developer Documentation Style Guide standards. Use when answering technical questions, presenting architecture/design choices, explaining technical concepts, structuring procedural steps, or formatting technical communication in chat.
Build Python orchestrators that spawn multiple cursor-agent processes in parallel, capture their output, handle timeouts, log results, and verify task completion. Use this skill when automating bulk tasks via cursor-agent from the terminal.
Design concise, professional dashboards for software projects that surface usage, uptime, and status metrics without a “generative UI” look. Use this skill when a user asks for a modern dashboard, product/engineering health view, or status overview for a SaaS or software system.
SOC 職業分類に基づく
SKILL.md を表示中
| name | nano-banana-pro |
| description | Generate or edit images with Google's Nano Banana Pro image model through the bundled Gemini API script. Use for text-to-image generation, image-to-image edits, or requests for 1K, 2K, or 4K Gemini image output. |
Generate or edit images with the bundled script. Resolve the script relative to
this SKILL.md; do not assume a particular agent harness or installed-skills
directory.
Run the non-billing check before the first generation in a session:
uv run /absolute/path/to/nano-banana-pro/scripts/generate_image.py --check
The script reads GEMINI_API_KEY from the environment. If this machine keeps
the key in ~/.secrets, source that file without printing its contents:
source ~/.secrets >/dev/null 2>&1
uv run /absolute/path/to/nano-banana-pro/scripts/generate_image.py --check
Never echo a key. Prefer GEMINI_API_KEY over --api-key, because command-line
arguments may be visible to other processes.
Run from the user's working directory so relative output paths land with the rest of their work:
uv run /absolute/path/to/nano-banana-pro/scripts/generate_image.py \
--prompt "A serene Japanese garden with cherry blossoms" \
--filename "2026-07-25-09-30-00-japanese-garden.png" \
--resolution 1K
Use the filename format yyyy-mm-dd-hh-mm-ss-description.png. Keep a new
filename for every iteration.
Start at 1K. Move to 2K when normal-resolution output is requested and to
4K only for a requested or approved final high-resolution result. Dimensions
map as follows:
1K2K4KPass the source image with --input-image:
uv run /absolute/path/to/nano-banana-pro/scripts/generate_image.py \
--input-image "/absolute/path/to/original.jpg" \
--prompt "Change only the sky to dramatic storm clouds; preserve the subject, crop, lighting, text, and all other details." \
--filename "2026-07-25-09-35-00-dramatic-sky.png" \
--resolution 2K
For precise edits, name the single change and explicitly list what must remain unchanged. Reuse the same original input while iterating unless the user asks to build on an earlier generated result.
The script creates parent directories as needed, writes a PNG, and prints its absolute path. Inspect the image before handing it off when visual quality is part of the request.
Common failures:
No API key provided: load GEMINI_API_KEY.Error loading input image: verify the source path and permissions.