Test fixture
Bug fixes, edge conditions, and minimal patch accuracy.
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 diagnosis, fix, tests. A refactor moved idempotency key generation outside the retry loop, causing different user actions to share the same key. Identify the regression and fix.
{
"metrics": {
"repro": {
"checks": [
{
"contains": [
"refactor"
]
},
{
"contains": [
"different user actions"
]
},
{
"contains": [
"same key"
]
},
{
"contains": [
"retry"
]
}
]
},
"hidden": {
"checks": [
{
"contains": [
"per operation"
]
},
{
"contains": [
"generate inside"
]
},
{
"contains": [
"scope"
]
}
]
},
"diagnose": {
"checks": [
{
"contains": [
"idempotency key"
]
},
{
"contains": [
"reuse"
]
},
{
"contains": [
"regression"
]
}
]
}
}
}temperature
0
max_tokens
420
timeout (s)
120
type
scored
file
debug-idempotency-key-reuse-v2.json