GuardAI Adversarial Security Assessment Platform for AI API

The GuardAI Platform, created by NavInfo Europe, is used to measure the Adversarial and Natural Robustness of a model.

The provided API is used in automation to integrate testing into your AI development pipeline. Specifically, this enables integration with continuous integration and/or deployment systems.

API Guide

In general, using the GuardAI API consists of the following high level steps:

1. Creating API credentials - go to the Profile page:

   

2. Click on the key icon to create a new API key:

   

3. Setting up the API client

   Using the API key created in step 1, create an API client for your chosen language, for example, in Python:

    ...
    config = load_sdk_config()
    auth_token = get_auth_token(config['connection']['host'],
                                config['connection']['api-key'], config['connection']['api-key-id'])
    api_client = get_api_client(config['connection']['host'], auth_token)
    ...

4. Use the api client to make API calls:

    ...
    project_api = ProjectApi(api_client)
    print('Listing projects...')
    projects = project_api.get_projects()
    ...

   See the example code for examples on how to get information on projects, tests and how to start a robustness test. Also consult the GuardAI SDK for guidance on how to integrate new models into the platform.

API Examples

Python examples

API Reference Documentation

HTML Documentation

Python API client documentation

Source code

GitHub