meshaudit dashboard

Methodology & what this dashboard cannot see (read before trusting a row)

One-sided evidence. This tool only ever sees packets that reached an observer. A repeater dropping packets via loop.detect/flood.max is, by definition, invisible here. A row with no finding means no evidence was found, not that the node is clean — there is no "pass" state in this tool anywhere, deliberately.
Vantage bias. Every rate is normalized by how well-observed that node was, and every ranking is sorted on a confidence-discounted lower bound, not the raw rate — a huge number seen by one observer is weak evidence, not strong evidence. coverage_confidence is shown alongside every rate so nothing is hidden, only reordered.
What this tool structurally cannot see. Radio parameter mismatches produce no data at all, not bad data. Interference thresholds, AGC behavior, and TX/RX turnaround leave no over-the-air trace. This is not an exhaustive audit.
INSUFFICIENT_EVIDENCE is the default, permanent state for most of the mesh — not a placeholder waiting to become something more decisive.
Full report methodology, node-by-node evidence, and CoreScope's own live traces remain available via meshaudit analyze --html and live.mwmesh.com.

Archive health

—

repeater airtime (ms, window)

—

originator airtime (ms, window)

verdict detector

Key	Kind	Name	Confidence	Rank basis (ms/obs-hr, 95% LB)	Point est.	Verdicts

New	Verdict	Detector	Node	Summary	Confidence	Remedy

close ×