optiattack-client

OptiAttack Client

The OptiAttack Client (optiattack_client) is a FastAPI-based server package designed to facilitate remote integration between the OptiAttack core engine and external models or systems under test (NUT). It provides a simple decorator and API endpoints for seamless communication and adversarial test execution.

Table of Contents

• Features
• Installation
• Quick Start
• API Endpoints
• Example Usage
• Configuration
• Contributing
• License

Features

• FastAPI server for remote model or NUT integration
• Decorator-based interface for easy wrapping of prediction functions
• Automatic endpoint creation for running, stopping, and querying the NUT
• Supports image-based adversarial attacks
• Easy integration with OptiAttack core

Installation

You can install the OptiAttack Client package using pip:

pip install optiattack-client

Alternatively, if you want to install from the local directory (for development or custom modifications), run:

cd client
pip install -e .

Quick Start

• Wrap your prediction function with the provided decorator:

from optiattack_client import collect_info

@collect_info(host="localhost", port=38000)
def predict(image_array):
    # Your prediction logic here
    return {"predictions": ...}

• Start your script. The FastAPI server will automatically run and expose endpoints for OptiAttack to communicate with.

API Endpoints

The client exposes the following endpoints (default base path: /api/v1):

• GET /api/v1/infoNUT — Get the current state of the NUT
• POST /api/v1/runNUT — Run a prediction on the NUT (expects base64-encoded image)
• POST /api/v1/stopNUT — Stop the NUT
• POST /api/v1/newAction — Apply a new action (mutation) and get prediction

See client/constants.py for all endpoint paths and defaults.

Example Usage

from optiattack_client import collect_info

@collect_info(host="localhost", port=38000)
def predict(image_array):
    # Example: Use your ML model to predict
    # result = model.predict(image_array)
    return {"predictions": [0.1, 0.9]}  # Example output

# The FastAPI server is now running at http://localhost:38000

Configuration

• Host and Port: Set via the collect_info decorator arguments.
• Endpoints: Configurable in client/constants.py.
• Python Version: Requires Python 3.9 or higher.

Contributing

Contributions are welcome! Please open issues or pull requests for bug fixes, new features, or documentation improvements.

License

This package is part of the OptiAttack project and is licensed under the GNU Lesser General Public License v3 (LGPLv3).