Nano Banana Pro Prompt Recommender

What it does

This skill provides an indexed library of 10,000+ image-generation prompt templates optimized for Nano Banana Pro (Gemini) and compatible with other text-to-image models. It helps agents search by use case (product photo, portrait, thumbnail, poster, etc.), present up to three matched prompts with a mandatory sample image, and — when requested — remix a selected template into a custom prompt tailored to user content. The skill also includes tooling guidance for keeping prompt references up to date and efficient search strategies.

When to use it

Use this skill when a user asks to generate or illustrate visual content: "create a thumbnail", "product photo", "illustrate my article", "portrait in X style", or any request for AI image prompts. Trigger on queries mentioning social media images, product marketing, portraits, thumbnails, posters, infographics, or any call for visual assets.

What's included

• Scripts: (referenced) setup/sync guidance to refresh the prompt references from GitHub.
• References: large curated prompt library (references/*.json) — the skill expects these files to exist and be periodically synced.
• Instructions: detailed workflow covering search heuristics, token-optimized loading, presentation rules (max 3 prompts), mandatory sample images, and remix steps for customization.

Compatible agents

Designed for OpenClaw/Claude Code and file-based agent integrations that can download and send images; broadly compatible with agents that can run Node/Python setup and fetch reference JSON files.