feat(testing): add adversarial testing module for mofa-testing (#1459)

[iashutoshyadav]

Summary

This PR introduces a foundational Adversarial Testing Module in mofa-testing to enable systematic security and robustness evaluation of AI agents.

It provides a modular framework to simulate adversarial scenarios aligned with OWASP LLM Top 10 risks.

---

Key Features

1. Adversarial Testing Framework

• Added a new adversarial module with structured components:

  • category.rs – Defines adversarial categories
  • corpus.rs – Contains adversarial test datasets
  • evaluator.rs – Implements evaluation logic
  • report.rs – Generates structured reports
  • tests.rs – Includes unit tests

2. Modular & Extensible Design

• Clean separation of concerns across components
• Designed for easy extension with new evaluators and datasets

3. Integration

• Integrated the module via lib.rs
• Added necessary crate configuration in Cargo.toml

---

Testing

• Includes basic unit tests in tests.rs
• Validates adversarial evaluation flow and reporting functionality

---

Motivation

Current mofa-testing focuses on functional correctness but lacks mechanisms for security validation.

This module lays the groundwork for:

• Prompt injection detection
• Jailbreak attack simulation
• Safety and robustness evaluation of AI agents

---

Scope

This PR implements only the foundational layer.

Advanced evaluators and comprehensive OWASP coverage will be added in future PRs.

---

Related Issue

Closes #1459

[diiviikk5]

You do not need to create mofa testing inside crates there is an existing foundation in the tests directory in root for this , i suggest you check it out first