Nanobanana Skill

This is a solid wrapper around Gemini's image models, specifically tuned for Nanobanana 2 with Google Search grounding and thinking summaries turned on by default. That's the main reason to use it: you get grounded visuals that can pull from live web data, which matters for anything that needs current references, product accuracy, or real-world composition cues. The skill handles both text-to-image generation and multi-reference editing, supports extreme aspect ratios like 1:8 and 21:9, and lets you pick between 512px ideation mode and 4K finals. The prompting guidance is unusually specific and production-focused, pushing you toward art-director-style briefs instead of vague requests. If you're doing marketing assets, product mockups, or anything where factual grounding improves output quality, this is a clean way to route those tasks to Gemini without manual CLI work.

skills.sh installs: 120
GitHub stars: 249
last updated: Mar 9, 2026

View on GitHub

Install

$ npx -y skills add feiskyer/codex-settings --skill nanobanana-skill --agent claude-code

Installs into .claude/skills of the current project.

Files

Nanobanana Image Skill

Use the bundled nanobanana.py to generate or edit images with Gemini image models. The default path targets Nano Banana 2.1 (gemini-nano-banana-2.1) with thinking summaries and Google Search grounding on, because those defaults are the main reason to use this skill instead of a generic image prompt.

Requirements

GEMINI_API_KEY in ~/.nanobanana.env (GEMINI_API_KEY=sk-dummy) or the shell environment.
Python dependencies from requirements.txt.
The executable is nanobanana.py in this same skill directory. Resolve its absolute path once before running it.

The CLI can print --help without credentials. It validates local input files before creating an API client and exits non-zero when the API returns no image.

Defaults

Model gemini-nano-banana-2.1, search grounding on, thinking summaries on at level high, resolution 1K, and no aspect ratio unless the request implies a shape. Do not make the user choose a model, search mode, or thinking mode unless they asked for that level of control.

SKILL.md on GitHub