How does the skill handle failed image generations?

If a 'gross failure' is detected, such as a missing logo or completely wrong colors, the skill automatically retries the generation up to two times, appending specific corrective instructions to the prompt.

Can it detect if a regulatory disclaimer is present?

Yes, it can verify if a requested disclaimer is visible in the designated area, though it does not perform a character-by-character spell check of the text.

Will this skill block me from seeing my images?

It only blocks and retries on gross brand failures. For minor issues like text legibility or composition, it will still show you the image but include advisory warnings.

At what stage of the workflow does this skill run?

It activates as 'Step 5b', sitting specifically between the initial image generation (Step 5) and the final user review (Step 6).

Does it check for exact brand hex codes?

No, LLM vision is not reliable for pixel-level hex matching. The skill focuses on dominant color families to ensure general brand alignment without blocking outputs over minor shade variations.

Ad Image Quality Review

Name: Ad Image Quality Review
Author: jeffrey94

byjeffrey94

•

Design Tools

Evaluates generated ad images for brand compliance and brief alignment using multimodal vision to ensure high-quality visual outputs.

The Quality Review skill acts as an automated QA gatekeeper for visual asset generation within Claude Code. By leveraging multimodal vision, it inspects generated images immediately after creation to identify gross brand violations—such as incorrect color palettes or missing logos—and evaluates how well the output aligns with the original creative brief. The skill features an intelligent decision engine that automatically retries failed generations up to two times with corrective prompting, while providing helpful advisory warnings for minor issues like text legibility or layout balance, ensuring that the final assets presented to users meet professional standards.

Key Features

01Gross brand compliance checking for colors and logo presence

02Automated multimodal visual inspection of generated assets

031 GitHub stars

04Brief alignment evaluation for mood, tone, and concept

05Intelligent auto-retry logic for significant visual failures

06Advisory warnings for fine-grained typography and composition issues

Use Cases

01Automated QA for high-volume display ad and social media asset generation

02Ensuring brand safety and mandatory elements like disclaimers are present

03Filtering out off-concept AI generations before they reach the user review stage

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jeffrey94/jeffrey-skills quality-review

For use in Claude.ai and ChatGPT

Key Features

01Gross brand compliance checking for colors and logo presence

02Automated multimodal visual inspection of generated assets

031 GitHub stars

04Brief alignment evaluation for mood, tone, and concept

05Intelligent auto-retry logic for significant visual failures

06Advisory warnings for fine-grained typography and composition issues

Use Cases

01Automated QA for high-volume display ad and social media asset generation

02Ensuring brand safety and mandatory elements like disclaimers are present

03Filtering out off-concept AI generations before they reach the user review stage

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jeffrey94/jeffrey-skills quality-review

For use in Claude.ai and ChatGPT