Test fixture
Secure code changes, vulnerability recognition, and safe defaults.
The model receives the prompt (and optional system message). The run uses scorer rubric_json_metrics with the JSON configuration below. Pass/fail and partial credit are determined entirely by that scorer against the model output; no human grading.
Return JSON only with keys verdict, risk, mitigation. A service encrypts customer exports with one long-lived KMS key and has no rotation or key usage audit process.
{
"metrics": {
"correct": {
"checks": [
{
"contains": [
"KMS key"
]
},
{
"contains": [
"rotation"
]
},
{
"contains": [
"audit"
]
}
]
},
"hidden": {
"checks": [
{
"contains": [
"envelope encryption"
]
},
{
"contains": [
"key policy"
]
},
{
"contains": [
"usage logs"
]
}
]
}
}
}temperature
0
max_tokens
560
timeout (s)
120
type
scored
file
sec-kms-key-rotation-policy.json