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. Bearer session tokens remain valid after theft because there is no replay detection, token rotation, or device binding for high-risk actions.
{
"metrics": {
"correct": {
"checks": [
{
"contains": [
"session replay"
]
},
{
"contains": [
"bearer token"
]
},
{
"contains": [
"theft"
]
}
]
},
"hidden": {
"checks": [
{
"contains": [
"rotation"
]
},
{
"contains": [
"device binding"
]
},
{
"contains": [
"step-up"
]
}
]
}
}
}temperature
0
max_tokens
560
timeout (s)
120
type
scored
file
sec-session-replay-token-binding.json