app.py

import gradio as gr

def detect(text):
    # 模擬防禦邏輯
    malicious_keywords = ["ignore", "system prompt", "secret", "override"]
    if any(key in text.lower() for key in malicious_keywords):
        return "⚠️ 偵測到潛在提示注入攻擊！ (High Risk)"
    return "✅ 輸入語意安全 (Clean)"

demo = gr.Interface(
    fn=detect, 
    inputs=gr.Textbox(label="輸入待測 Prompt", placeholder="例如：Ignore previous instructions..."), 
    outputs=gr.Textbox(label="防禦系統分析結果"),
    title="🛡️ PromptGuard RAG 防禦演示系統",
    description="本系統由 PromptGuard Research 開發，用於偵測檢索增強生成架構中的惡意指令注入。"
)

demo.launch()