Test fixture
Code transformation while preserving behavior and intent.
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 plan, refactor, tests. Email, Slack, and webhook notifications duplicate recipient filtering, retries, and error handling. Refactor shared fanout behavior while preserving channel specifics.
{
"metrics": {
"intent": {
"checks": [
{
"contains": [
"recipient filtering"
]
},
{
"contains": [
"shared fanout"
]
},
{
"contains": [
"channel specifics"
]
}
]
},
"visible": {
"checks": [
{
"contains": [
"Email"
]
},
{
"contains": [
"Slack"
]
},
{
"contains": [
"webhook"
]
}
]
},
"hidden": {
"checks": [
{
"contains": [
"retry behavior"
]
},
{
"contains": [
"error handling"
]
},
{
"contains": [
"tests"
]
}
]
}
}
}temperature
0
max_tokens
560
timeout (s)
120
type
scored
file
refactor-notification-fanout.json