What problem does it solve?
Marketers often see a high-performing image ad and want to reproduce its structure for their own brand, but manually deconstructing layout, typography, and style into a reusable prompt is slow and inconsistent. This Skill turns any existing ad image into a validated, parameterizable prompt template added to a shared 37-template library.
Core Features & Use Cases
- 10-phase cloning workflow: Visual analysis, draft prompt, generate-with-reference, iterate (capped at 4 rounds), generalize into placeholders, cross-brand test, cross-model validation, and documentation.
- Dual-backend validation: Routes through ChatGPT Image 2 (gpt-image-2) for typography/UI-mimicry ads or Nano Banana for photoreal/lifestyle/multi-reference ads, with model notes recorded for both.
- Aspect ratio mapping: Maps source ratios like 4:5 or 1.91:1 to the Arcads-supported 1:1, 16:9, and 9:16 outputs with documented fallbacks.
- Use Case: You find a competitor's fake Slack-thread ad that converts well. Provide the image, and the Skill analyzes it, generates matching variants via gpt-image-2, generalizes the prompt with brand placeholders, and appends it as template T40+ to the shared prompt library.
Quick Start
Clone this ad image into a reusable prompt template and validate it with the Arcads image generators.
