Deliver cynical, adversarial reviews

Review code for logic bugs via adversarial test generation

Finding logic bugs that escape normal code review by generating adversarial test cases.

• · Code to review (function, module, or file)
• · Domain context (e.g., "payment processor")

• · List of edge cases
• · Generated test cases
• · Severity-ranked findings

• · Playwright (optional, for integration testing)

• · Code syntactically valid
• · Domain context provided

• · LLM generates unrealistic edge cases
• · Flaky tests
• · Coverage gaps not caught

• · Adversarial test generation
• · Domain-aware edge cases
• · Severity ranking