Skill

banana-ads

Generates banners, social media ads, display ads, and promotional graphics via Gemini using auto-built prompts, format detection, and aspect ratios like 4:5 Instagram or 16:9 YouTube.

Gemini

Bash

ai-ml

design

npx claudepluginhub agentic-dev3o/devx-plugins --plugin nano-banana-ultimate

Tool Access

This skill is limited to using the following tools:

Bash(uv:*)Bash(ls:*)Read

Preview

**Input:** `$ARGUMENTS` (optional — product/service and ad format)

Supporting Assets

references/prompts.md

SKILL.md

Similar Skills

form-social

Use when asked to design social media graphics, ad creatives, or marketing assets. Examples: "design a LinkedIn post for our launch", "create ad creatives for our campaign", "make an Instagram story", "design a Twitter card", "create a banner ad", "social assets for the product announcement".

tonone-form

form-social

Guides creation of social media graphics, ad creatives, and marketing assets via phased discovery on platform, format, goal, brand, message, followed by confirmation brief.

11 tools

tonone

image

27.7k

Generates, edits, and optimizes marketing images such as blog heroes, social graphics, product mockups, and banners using AI tools like Gemini, Flux, DALL-E, Ideogram and design tools like Figma, Canva.

1 file

marketing-skills

Stats

Parent Repo Stars10

Parent Repo Forks1

Last CommitApr 5, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

Online Advertising Visuals

Input: $ARGUMENTS (optional — product/service and ad format)

If $ARGUMENTS is provided, use it as the campaign brief and skip to prompt building. If empty, ask about the product and target placement.

Requirements

uv installed
GEMINI_API_KEY environment variable set (get one at https://aistudio.google.com/apikey)

Workflow

Understand — Clarify the product or service, the value proposition, and the target ad placement (Instagram feed, YouTube pre-roll, web banner, etc.).
Build prompt — Apply the Prompt Formula below to construct a bold, commercially viable image description.
Configure — Detect the ad format from context and select aspect ratio and resolution from the Ad Format Detection table.
Generate — Run the script. the generation script with the assembled prompt and settings.
Deliver — Report the saved file path. Do NOT read the image file back. Offer to iterate: adjust composition, color palette, or layout.

Prompt Formula

[Product/Service] + [Value Proposition as Action] + [Ad Placement Context] + [Bold Center-Weighted Composition] + [Clean Commercial Photography or Flat Design]

Ad Format Detection

Placement	Aspect Ratio	Resolution
Instagram/Facebook feed	4:5	2K
Instagram story	9:16	2K
YouTube pre-roll	16:9	2K
Web banner	21:9	2K
Square social	1:1	2K

Key Instructions

Negative space for text overlay -- Always leave generous negative space (top, bottom, or side) for headline and CTA placement. Ad visuals are incomplete without copy; design the composition to accommodate it.
No baked-in text -- Do NOT prompt Gemini to render text into the image. Gemini text rendering is unreliable and produces artifacts. Advise the user to add headlines, CTAs, and legal copy in a separate design tool (Figma, Canva, Photoshop).
Bold center-weighted composition -- Place the hero product or key visual element at the center or slightly off-center with clean breathing room around it.
Commercial lighting -- Default to clean, bright studio lighting with soft shadows. Use high-key lighting for upbeat campaigns and moody side lighting for premium or luxury positioning.
Color psychology -- Match the color palette to the campaign intent: warm tones for urgency and sales, cool tones for trust and tech, vibrant saturated hues for youth and energy.

Generation

uv run "${CLAUDE_PLUGIN_ROOT}/scripts/banana.py" \
  --prompt "<detailed prompt following the formula>" \
  --output "<YYYY-MM-DD-HH-MM-SS-descriptive-name>.png" \
  --resolution <2K> \
  --aspect-ratio <ratio from Ad Format Detection table> \
  --model gemini-3-pro-image-preview

Editing (with input images)

When the user provides a product photo or existing visual to modify:

uv run "${CLAUDE_PLUGIN_ROOT}/scripts/banana.py" \
  --prompt "<edit instruction: what to change AND what to keep>" \
  --output "<YYYY-MM-DD-HH-MM-SS-descriptive-name>.png" \
  --input-image "/path/to/source.png" \
  --aspect-ratio <ratio>

For edits, describe both the change and what must stay identical:

"Place this product on a clean bright studio background with soft shadow. Keep the product, its color, and proportions exactly the same."

Up to 14 input images can be passed (repeat --input-image for each).

References

See references/prompts.md for ready-to-use example prompts covering common advertising visual scenarios.