Test fixture
Arithmetic, symbolic steps, and structured problem solving.
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, evidence, constraints. Users receive duplicate welcome emails only when signup retries after a timeout. The database has one user row but the email log has two sends. Identify the likely root cause.
{
"metrics": {
"accuracy": {
"checks": [
{
"contains": [
"side effect"
]
},
{
"contains": [
"retry"
]
},
{
"contains": [
"idempotency"
]
}
]
},
"evidence": {
"checks": [
{
"contains": [
"one user row"
]
},
{
"contains": [
"two sends"
]
},
{
"contains": [
"timeout"
]
}
]
},
"constraint": {
"checks": [
{
"contains": [
"welcome email"
]
},
{
"contains": [
"dedupe"
]
},
{
"contains": [
"idempotency key"
]
}
]
}
}
}temperature
0
max_tokens
500
timeout (s)
120
type
scored
file
reason-root-email-duplication.json