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 public production GraphQL endpoint exposes full introspection and detailed resolver errors to unauthenticated users.
{
"metrics": {
"correct": {
"checks": [
{
"contains": [
"GraphQL introspection"
]
},
{
"contains": [
"production"
]
},
{
"contains": [
"unauthenticated"
]
}
]
},
"hidden": {
"checks": [
{
"contains": [
"disable introspection"
]
},
{
"contains": [
"detailed errors"
]
},
{
"contains": [
"authorization"
]
}
]
}
}
}temperature
0
max_tokens
460
timeout (s)
120
type
scored
file
sec-graphql-introspection-prod.json