Skip to content

A Streamlit based chat UI integrating Google Gemini and OpenAI. Powered by Prompt Security, it provides real-time redaction of sensitive data, secrets detection, and prompt injection blocking. Includes live security metrics and an adversarial prompt trigger menu for stress-testing LLM guardrails, including basic support for file/OCR scanning.

Notifications You must be signed in to change notification settings

simlocker/spooky-ai

Repository files navigation

Spooky AI 👻 - Homegrown App

A secure AI chat interface featuring Prompt Security (a SentinelOne company) integration. Both API and AI Gateway (reverse proxy) methods are supported.

** This app needs either a Google Gemini (Free Tier!) and/or OpenAI API key to work **

Setup Instructions

  1. Clone the repo

    git clone https://github.com/simlocker/spooky-ai.git
    cd spooky-ai
    
  2. Configure Keys and other Data

    Copy the example environment file and populate the required fields:

    cp .env.example .env
    nano .env
    # Open .env and paste your API Keys (Gemini, OpenAI, Prompt Security) and other information.
    
Variable Default Description
DEMO_USER_EMAIL user@example.com Change this depending on your Prompt Security policies defined for Homegrown Apps.
PS_APP_ID - Your Prompt Security App Id (api key)
PS_GATEWAY_URL https://******.prompt.security Change this to your Prompt Security base URL
GEMINI_FREE_API_KEY - Your Google AI Studio API key
OPENAI_API_KEY - Your OpenAI API key

(Please note the app won't work unless there's at least ONE API key present for one of the LLMs.)

  1. Run with Docker

    docker compose up -d --build
    
  2. Access your Instance

    Open your browser to http://your-ip-address:8501




More detailed information coming soon....
Gastón Z - 2026

About

A Streamlit based chat UI integrating Google Gemini and OpenAI. Powered by Prompt Security, it provides real-time redaction of sensitive data, secrets detection, and prompt injection blocking. Includes live security metrics and an adversarial prompt trigger menu for stress-testing LLM guardrails, including basic support for file/OCR scanning.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published