CL-L3-REPAIR-RESCUE-RATE
VALID Reproducible from six published aggregate repair rows; individual test-set preservation is unverifiable.
- Metric
repair_rescue_count- Numerator
2- Denominator
6- Unit
matched failed implementation repairs- Analysis rule
Count repair_outcome=pass among the six matched repair rows. Treat the registered per-test pass-set preservation condition as unverified; aggregate test counts are only a separate diagnostic.- Publication status
STATUS_ONLY- Evidence status
Reproducible from six published aggregate repair rows; individual test-set preservation is unverifiable.- Interpretation
Two of six repairs rescued the official hidden-suite outcome; the three-rescue threshold was not met. The result does not show a causal repair benefit or verify per-test no-regression.- Reported values
{"per_test_pass_set_preservation":"unverifiable","repairs":6,"rescued":2,"threshold_met":false,"threshold_rescues":3}
Round: l3-repair