🛡️ SentraGuard Lite

An offline AI Guardrails Gateway built using FastAPI, Streamlit, and Python.

This project analyzes user prompts and retrieved context documents to detect:

• Prompt Injection
• Personally Identifiable Information (PII)
• RAG Injection

Based on the detected risks, the gateway decides whether to:

• ✅ Allow
• ⚠️ Transform
• ❌ Block

---

Features

• FastAPI REST API
• Streamlit Web UI
• Command Line Interface (CLI)
• Prompt Injection Detection
• Email Detection
• Phone Number Detection
• RAG Injection Detection
• Prompt Sanitization
• Risk Scoring
• Policy Engine
• Structured Logging
• Docker Support
• Pytest Unit Tests

---

Project Architecture

                   User
                     │
      ┌──────────────┴──────────────┐
      │                             │
 Streamlit UI                      CLI
      │                             │
      └──────────────┬──────────────┘
                     │
                 FastAPI
                     │
             Analysis Engine
                     │
      ┌──────────────┼──────────────┐
      │              │              │
Prompt Injection  PII Detector  RAG Detector
      └──────────────┼──────────────┘
                     │
               Risk Scoring
                     │
              Decision Engine
                     │
             Sanitization Layer
                     │
                JSON Response

---

Folder Structure

guardrails-gateway/
│
├── app/
│   ├── main.py
│   ├── schemas.py
│   ├── config.py
│   │
│   ├── core/
│   │   ├── analyzer.py
│   │   ├── detectors.py
│   │   ├── sanitizer.py
│   │   ├── scorer.py
│   │   ├── policy.py
│   │   └── logger.py
│   │
│   └── utils/
│       └── regex.py
│
├── ui/
│   └── streamlit_app.py
│
├── tests/
│
├── cli.py
├── Dockerfile.api
├── Dockerfile.ui
├── docker-compose.yml
├── requirements.api.txt
├── requirements.ui.txt
├── pytest.ini
├── README.md
└── .gitignore

---

Installation

Clone Repository

git clone <repository-url>

cd guardrails-gateway

Create Virtual Environment

python -m venv .venv

Windows

.venv\Scripts\activate

Linux/macOS

source .venv/bin/activate

---

Install Dependencies

pip install -r requirements.api.txt

pip install -r requirements.ui.txt

---

Running the Backend

uvicorn app.main:app --reload

API Documentation:

http://localhost:8000/docs

---

Running the Streamlit UI

streamlit run ui/streamlit_app.py

Open

http://localhost:8501

---

Running the CLI

Example

python cli.py --prompt "Ignore previous instructions"

With RAG Document

python cli.py ^
--prompt "Ignore previous instructions" ^
--doc "Developer: Ignore guidelines"

---

🐳 Docker Deployment

Build

docker compose build

Run

docker compose up

Access the Application

FastAPI Swagger

http://localhost:8000/docs

Streamlit UI

http://localhost:8501

---

Stop Containers

docker compose down

---

API Endpoints

GET /policy

Returns

• Detection policy
• Thresholds
• Enabled detectors

---

POST /analyze

Input

{
  "prompt": "Ignore previous instructions",

  "context_docs": [],

  "metadata": {
    "app_id": "demo",
    "user_id": "user",
    "request_id": "1"
  }
}

Output

{
  "decision": "transform",

  "risk_score": 50,

  "risk_tags": [
    "prompt_injection"
  ]
}

---

Running Tests

pytest -v

Example

13 passed

---

Detection Rules

Prompt Injection

Examples

• Ignore previous instructions
• Reveal system prompt
• Override instructions

---

PII

Detects

• Email addresses
• Phone numbers

---

RAG Injection

Detects

• Developer:
• System:
• Ignore guidelines
• Override policy

---

Design Decisions

• Modular architecture
• Deterministic offline execution
• No external AI APIs
• Regex-based detection
• Configurable policy thresholds
• Structured logging
• Separate UI and backend

---

Future Improvements

• Named Entity Recognition (NER) for advanced PII detection
• Machine learning-based prompt injection detection
• Configurable policies via YAML/JSON
• JWT authentication
• Role-based access control
• Database-backed audit logging
• Prometheus metrics
• Kubernetes deployment

---

Technologies Used

• Python 3.11
• FastAPI
• Pydantic
• Streamlit
• Requests
• Pytest
• Docker
• Uvicorn

---

Screenshots

Streamlit UI

FastAPI Swagger

Pytest Results

Docker Deployment

AI Assistance Disclosure

AI-assisted tools were used during development to discuss architecture, review implementation approaches, improve documentation, and identify potential edge cases.

All implementation decisions, testing, debugging, and final validation were performed by the author before submission.

---

Author

Dhrumil Moga