scenario-generator

You can automate AI red-teaming scenario generation by providing user descriptions of your agent to generate tailored test scenarios. It leverages the giskard.checks library to evaluate AI performance within these adversarial suites.

Building test suites for AI agents with giskard.checks is accomplished by assembling individually generated test scenarios into a comprehensive suite. This streamlines the red-teaming process by automatically structuring adversarial evaluations.

Yes, you need the giskard-checks dependency installed to execute the generated AI test scenarios. The Skill relies on this library to operate and evaluate AI performance during red-teaming.

Yes, you can generate a test scenario from a simple description of your AI agent by running the command with your agent and inputs. The Skill outputs the generated scenarios to a YAML file for testing.

Automating AI testing scenarios with giskard.checks requires the giskard-checks dependency to be present for execution. Scenarios are generated based solely on user descriptions, so testing accuracy depends on the detail of that input.