README.md

metadata

title: NeuroShield PoC
emoji: 🛡️
colorFrom: blue
colorTo: indigo
sdk: streamlit
sdk_version: 1.32.0
app_file: app.py
pinned: true

🛡️ NeuroShield PoC (Enhanced Edition)

A powerful AI-based moderation assistant built with Streamlit, Hugging Face Transformers, and Groq API. Designed for nuanced, voice-guided responses to online toxicity.

---

🚀 Features

• ✅ 14-label toxicity classification (simulated Jigsaw + extended logic)
• 🧠 Coaching voice personas (compassionate, assertive, reflective, etc.)
• 🔥 Visual indicators (emoji SAFE/UNSAFE + toxicity heatmap)
• 🎚️ Tolerance control for each category
• 🧒 Kids Mode and NSFW Filter
• ✍️ Groq LLM Rewrites using selected tone/strategy

---

🧠 Coaching Voice Selector

Choose from:

• The Boundary Setter
• The Mirror
• The Compassionate Reframer
• The Challenger → Fully extensible via coaching_voices.json

---

🛠️ Setup

pip install -r requirements.txt
streamlit run app.py

---

🧪 Secrets Configuration (Hugging Face Spaces)

Add the following to your repository secrets:

GROQ_API_KEY=your-groq-api-key

---

MIT License © 2025 – Built for ethical AI and safe communication.