QualityCheckResult
Properties
Section titled “Properties”avgScore
Section titled “avgScore”avgScore: number;Mean relevance score across all results.
backgroundReformulation?
Section titled “backgroundReformulation?”optional backgroundReformulation?: boolean;Whether background reformulation was triggered (voice agents).
coverageEstimate
Section titled “coverageEstimate”coverageEstimate: number;0-1 estimate of how well results cover the query.
estimatedTokens
Section titled “estimatedTokens”estimatedTokens: number;Estimated prompt-token cost of the result set (~chars/4). Retrieval pays for itself in tokens — track this to catch a retriever that is “accurate” only by flooding the context window.
quality
Section titled “quality”quality: "high" | "medium" | "low";Quality bucket: high, medium, or low.
reformulated
Section titled “reformulated”reformulated: boolean;Whether the query was reformulated and re-retrieved.
reformulatedQuery?
Section titled “reformulatedQuery?”optional reformulatedQuery?: string;The reformulated query string, if reformulation occurred.
topScore
Section titled “topScore”topScore: number;Highest relevance score in the result set.