SKILL: Anatomical Errors

Version: 1.0 | Hermes Agent Failure Pattern Library

DESCRIPTION

Mastery of detecting and correcting the most common anatomical failures in AI image generation: extra limbs, merged fingers, missing digits, deformed joints, and asymmetric features. These errors are the primary quality gate for commercial AI imagery. This skill provides detection taxonomies, model-specific fix strategies, and post-processing workflows [^82^].

ERROR TAXONOMY

Hands & Fingers (Highest Failure Rate):

• Extra fingers (6+ digits)
• Merged fingers (fused together like mittens)
• Missing fingers (3 or fewer visible)
• Distorted thumb (wrong position, too long, backward)
• Clubbed or blob hands (no individual fingers)
• Wrist anomalies (too thin, broken angle, extra joints)