quality gate Viatgecio Way · passing
| module | lines | branch |
|---|---|---|
| coderag core | 72.0% | 31.2% |
| coderag.api | 91.5% | 50.0% |
| coderag.generation | 90.7% | 82.1% |
| coderag.ingest | 92.7% | 92.3% |
| coderag.llm | 76.9% | 78.6% |
| coderag.observability | 94.7% | 50.0% |
| coderag.retrieval | 72.2% | 54.5% |
| coderag.stores | 51.6% | 30.0% |
| evals (eval harness) | 65.4% | 59.3% |
The full scan report is published as codebaserag-report.json.
~/homelab/sonarqube/analyze.sh -r codebaserag-report.json codebaserag python3 scripts/sonar-quality.pyREADME
Rendered from the scan of 2026-10-07T14:35:27+00:00 by the internal SonarQube
(LAN-only dashboard). Page, badges and JSON are regenerated after each scan.
The eval-gated core (ingest, retrieval, evals, stores-in-memory) carries the coverage number;
the live-service adapters (pgvector, qdrant, ollama, litellm) sit low by design — they need a
running backend and are exercised through contract tests, not unit tests.