183 lines
4.7 KiB
JSON
183 lines
4.7 KiB
JSON
{
|
|
"version": 1,
|
|
"generatedAt": "2026-09-27T10:18:40.511Z",
|
|
"referenceRuns": [
|
|
{
|
|
"runId": "reward-0.4100-p7644rd",
|
|
"status": "fresh",
|
|
"changed": [],
|
|
"capturedAt": "2026-09-26T22:56:38.770Z",
|
|
"capturedBy": "run"
|
|
},
|
|
{
|
|
"runId": "reward-0.4300-a5pdbqx",
|
|
"status": "fresh",
|
|
"changed": [],
|
|
"capturedAt": "2026-09-26T22:56:38.770Z",
|
|
"capturedBy": "run"
|
|
},
|
|
{
|
|
"runId": "reward-0.5100-2JvrM24",
|
|
"status": "fresh",
|
|
"changed": [],
|
|
"capturedAt": "2026-09-26T22:56:38.770Z",
|
|
"capturedBy": "run"
|
|
},
|
|
{
|
|
"runId": "reward-0.5200-DjvdVkm",
|
|
"status": "fresh",
|
|
"changed": [],
|
|
"capturedAt": "2026-09-26T22:56:38.770Z",
|
|
"capturedBy": "run"
|
|
}
|
|
],
|
|
"detectors": [
|
|
{
|
|
"report": "detector-answer-obviousness.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T20:12:26.866Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-broken-dev-env.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T20:11:11.713Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-credential-leakage.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T20:14:05.275Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-cross-task-reference.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T20:14:55.949Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-dimension-misapplication.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T20:33:02.998Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-fact-check-rubric-claims.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T22:50:16.107Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-good-response-defined.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T20:21:06.289Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-good-response-exhaustiveness.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T20:42:44.698Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-offline-verifiability.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T22:42:43.142Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-over-hinting.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T20:24:19.762Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-rubric-clarity.md",
|
|
"status": "fresh",
|
|
"method": "checksums",
|
|
"changed": [],
|
|
"capturedAt": "2026-09-27T09:59:48.629Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-rubric-coverage.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"atomic rubric (tests/atomic-rubric.yaml)"
|
|
],
|
|
"capturedAt": "2026-09-27T09:45:33.363Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-rubric-form.md",
|
|
"status": "fresh",
|
|
"method": "checksums",
|
|
"changed": [],
|
|
"capturedAt": "2026-09-27T09:51:49.533Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-rubric-generality.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T22:51:22.886Z",
|
|
"capturedBy": "stamp"
|
|
},
|
|
{
|
|
"report": "detector-snapshot-leakage.md",
|
|
"status": "stale",
|
|
"method": "checksums",
|
|
"changed": [
|
|
"holistic rubric (tests/holistic-rubric.md)"
|
|
],
|
|
"capturedAt": "2026-09-26T22:52:10.722Z",
|
|
"capturedBy": "stamp"
|
|
}
|
|
]
|
|
}
|