AI Security Proxy

An AI and security project that sits in front of a model endpoint and checks prompts for injection, hidden reasoning leakage, and secret exposure.

What it does

• scores prompts for risk
• redacts secrets and tokens
• blocks jailbreak-style requests
• returns a mock AI response for safe prompts
• shows the full flow in a small web UI

Run locally

pip install -r requirements.txt
uvicorn app.main:app --reload

Open http://127.0.0.1:8000.

AWS

You can deploy it with App Runner, ECS Fargate, or Elastic Beanstalk using the Dockerfile.