Quality Governance

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.

Release score100/100

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 · Review collection semantics2 collections need curation, scope, item-count, or freshness review; collection review rate is 0.2%.Confirm scope, curated status, freshness, and estimated item counts for collection/resource repositories; apply reviewed collection policy entries where appropriate.
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.

Coverage

Category coverage

13/13

All tracked categories are represented.

Coverage

Low-confidence classifications

0.1%

1 projects include low-confidence classification evidence.

Coverage

Collection review load

5.3%

56 collections; 2 need metadata or freshness review.

Coverage

Stale project rate

1%

10 projects have stale sync data.

Coverage

Average Git Score

26

Repository activity score average across the loaded knowledge set.

Coverage

Average Maintenance

38

Maintenance score average across the loaded knowledge set.

Top Quality Issues

What should be reviewed first

No blocking warnings or errors.Info observations may still appear in the JSON quality report.