creative-production-generative-polish

Composite generative image polish layers with deterministic text, chart, and logo layers for publish-safe marketing assets.

634|124|Updated Feb 10, 2026
One-click install
npx skills add https://github.com/AIDotNet/OpenCowork --skill creative-production-generative-polish
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: creative-production-generative-polish
Source: https://github.com/AIDotNet/OpenCowork/tree/main/resources/extensions/creative-production/skills/creative-production-generative-polish
Command: npx skills add https://github.com/AIDotNet/OpenCowork --skill creative-production-generative-polish

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Generative image tools often rewrite approved copy, distort charts, or break brand specs when polishing marketing visuals. This Skill keeps text, data, logos, dimensions, and filenames under deterministic control while using ImageGen only for backgrounds, textures, lighting, and scene polish.

Core Features & Use Cases

  • Layered composition workflow: Build exact copy, charts, CTAs, and safe zones in Python, SVG, or HTML, then generate only the visual finish with ImageGen and re-composite deterministically.
  • Publish-safe guardrails: Prevents generative tools from producing final claims, pricing, logos, charts, or fake product UI, and enforces validation of dimensions, text exactness, and provenance before handoff.
  • Format pack and social export support: Produces exact-size local variants with manifests and review boards, or routes Canva-based masters through Canva resize tooling for platform copies.
  • Use Case: You have an approved launch card with exact headline and chart data. Use this Skill to generate a polished background with ImageGen, overlay the deterministic text and chart layers, and export fixed-size social variants with a manifest and review board.

Quick Start

Ask the agent to apply generative polish to your selected marketing asset while keeping the exact text, chart data, logo placement, and dimensions unchanged.

Frequently Asked Questions about creative-production-generative-polish

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I polish a marketing image without changing the text or charts?

Build the exact copy, chart, logo, and safe zones deterministically in Python, SVG, or HTML first. Then use ImageGen only for backgrounds, lighting, and texture, and composite the deterministic layers back on top before exporting final assets.

What should AI image generation not be used for in marketing assets?

ImageGen should not produce final legal, pricing, benchmark, or product-capability claims, nor render text, numbers, charts, logos, or source lines for publish-bound work. Those elements must be rebuilt deterministically after generation.

Can I use Canva designs with generative polish workflows?

Yes. If the master asset lives in Canva, use the Canva resize tooling to create platform copies while preserving the original design. The workflow returns PNG export links and Canva edit links, leaving the source design unchanged.

When should I use generative polish instead of explorer skills?

Use explorer skills like offer-explorer or ads-explorer when you still need direction options or variations. Use generative polish only after a direction is selected and you need publish-bound exactness for text, charts, logos, and dimensions.

Why does my ImageGen output contain wrong text or numbers?

Generative image models cannot reliably render exact text or data. Treat any text or charts in ImageGen output as non-final, and rebuild those elements deterministically in Python, SVG, or HTML during the composition step.