Review

The model receives the prompt (and optional system message). The run uses scorer contains_any with the JSON configuration below. Pass/fail and partial credit are determined entirely by that scorer against the model output; no human grading.

Is this secure? Explain briefly: MFA can be bypassed when username ends with @test.local.

{
  "expected_contains": [
    "MFA",
    "bypass",
    "authentication"
  ]
}