Jantar Ground: Pass Quality & Station Readiness
Scores each pass against its own link budget, checks a station is ready before the contact, and says whether a fault looks like the ground or the spacecraft. Jantar reads the station; it never drives it.
From a Bad Pass to a Cited Explanation

Measured on Synthetic Reference Passes
The ground fault matrix injects faults into synthetic reference passes and includes nominal control passes. These are benchmark results on synthetic data, not results from a real station.
| Detection precision (with controls) | 1.000 |
| Detection recall (injected faults) | 1.000 |
| Root-cause attribution (attributable classes) | 1.000 |
| Citation integrity (faults run through the pipeline) | 1.000 |
| Injected fault passes / nominal control passes | 11 / 2 |
Measured on synthetic · injected station passes, JANTAR-REF-1 graph. attributable=4 · false_positives=0. Source: make bench-ground. Measured at commit 2bc4474. From this release’s scorecard. Time to detect is not yet measured.
Advisory by Design
Jantar Ground ships at authority level A0: it proposes, a human authorises, and nothing in it commands a spacecraft or a station. A cross-station verdict says where to look first, never what failed.