Test fixture
Grounded answers under adversarial or missing-context prompts.
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 answer, supported_facts, uncertainty. In Node.js, if you require() the same module path twice in a process, is the module code executed twice? What does the second require() return? Avoid claiming require() re-executes the module on every call.
{
"metrics": {
"accuracy": {
"checks": [
{
"contains": [
"cached"
]
},
{
"contains": [
"executed once"
]
},
{
"contains": [
"require.cache"
]
}
]
},
"fabricated": {
"checks": [
{
"contains": [
"module is cached"
]
},
{
"contains": [
"not re-executed"
]
},
{
"contains": [
"returns cached exports"
]
}
]
}
}
}temperature
0
max_tokens
460
timeout (s)
120
type
scored
file
halluc-behav-module-cache.json