Data trust, coverage, and review risk for Git.Top recommendations.
Use this page before treating a project recommendation as high-confidence. Release health and data trust are separate: a deployment can have no blocking quality issues while the corpus still carries review risk.
Warnings and errors reduce this gate score; info observations stay visible without lowering it.
Data trust98/100
Low-confidence classification, collection review load, stale sync, and stale collection metadata reduce this corpus trust score.
Risk levelMEDIUM
Risk summarizes classification confidence, collection review load, and stale data.
Projects1047
d1 / d1_query
Risk Summary
MEDIUM
Risk level is derived from corpus confidence thresholds and may be high even when the release gate has no blocking issues.
Some projects have stale sync data.
Improvement Plan
How to lower data trust risk
P1 · Refresh stale project sync data10 projects are reported with stale sync observations; stale project rate is 1%.Run the production catch-up sync or targeted admin sync for stale high-traffic repositories, then re-run quality and smoke checks.
P1 · Burn down low-confidence classifications1 projects include low-confidence classification signals; rate is 0.1%.Use the review queue to inspect README/topics/evidence, add classification overrides for reviewed projects, and add eval coverage for recurring ambiguities.
P2 · Strengthen alternatives evidence1 popular projects have weak same-category alternatives and 0 popular projects are missing alternatives.Refresh generated alternatives, inspect same-category overlap, and add reviewed alternatives for high-visibility projects when generated matches are adjacent rather than direct substitutes.