generated-image-qa

What problem does it solve?

This Skill catches subtle defects in generated images and video start frames before they are used downstream, preventing duplicated objects, bad hands, invented text, and identity drift from slipping into motion.

Core Features & Use Cases

• Reference-aware QA: Checks every salient object against the reference image(s) and the prompt, flagging anything that appears without a source.
• Animation gate: Blocks defective stills from reaching Seedance, Kling, or any other video model where small image errors become obvious.
• Practical review workflow: Counts objects, inspects anatomy, checks text and logo artifacts, and verifies lighting, geometry, and consistency before you spend credits on video.

Quick Start

Use the generated-image-qa skill to inspect the attached image against its references and tell me whether it passes, needs revision, or should be escalated.