Text this: SHIELD: System for Harmful Explicit-Content Identification and Evaluation Through LLM-Driven Approach