@sanity/ailf-studio 1.2.1 → 1.3.0

package/dist/index.d.ts

@@ -434,6 +434,14 @@ interface ScoreItem {
     /** Ceiling score — gold-standard docs injected */
     ceilingScore?: number;
 }
+/** Reference to an artifact stored in an external object store (GCS). */
+interface ArtifactRef {
+    store: "gcs";
+    bucket: string;
+    path: string;
+    bytes?: number;
+    entryCount?: number;
+}
 /** A single gap/recommendation from gap analysis */
 interface RecommendationGap {
     affectedTaskIds: string[];
@@ -451,12 +459,41 @@ interface RecommendationsData {
     generatedAt: string;
     totalPotentialLift: number;
 }
+/**
+ * Per-test result stored in reports for drill-down and audit.
+ * Mirrors StoredTestResult from @sanity/ailf-core. See D0029.
+ */
+interface StoredTestResultData {
+    area: string;
+    canonicalDocs?: DocumentRef[];
+    compositeScore?: number;
+    cost?: number;
+    dimensions: {
+        dimension: string;
+        reason: string;
+        score: number;
+    }[];
+    latencyMs?: number;
+    modelId: string;
+    outputFailure?: boolean;
+    responseOutput: string;
+    responseOutputTruncated?: boolean;
+    taskId: string;
+    tokenUsage?: {
+        completion: number;
+        prompt: number;
+        total: number;
+    };
+    variant: "baseline" | "gold";
+}
 /** A single low-scoring grader judgment stored in reports */
 interface JudgmentData {
     /** Docs the task expected the model to use */
     canonicalDocs?: DocumentRef[];
     dimension: string;
     modelId: string;
+    /** True when the model failed to produce output (empty response, API error, refusal) */
+    outputFailure?: boolean;
     reason: string;
     score: number;
     taskId: string;
@@ -497,6 +534,13 @@ interface PerModelData {
 interface SummaryData {
     /** Per-feature agent behavior data (only present when agentic mode ran) */
     agentBehavior?: FeatureAgentBehaviorData[] | null;
+    /** External artifact references — present when pipeline uploads to GCS (D0030) */
+    artifacts?: {
+        testOutputs?: ArtifactRef;
+        renderedPrompts?: ArtifactRef;
+        rawResults?: ArtifactRef;
+        traces?: ArtifactRef;
+    };
     belowCritical: string[];
     /** All Sanity documents used across the entire evaluation */
     documentManifest?: DocumentRef[];
@@ -521,6 +565,8 @@ interface SummaryData {
     /** Gap analysis recommendations (when gap analysis was run) */
     recommendations: null | RecommendationsData;
     scores: ScoreItem[];
+    /** Per-test results with model output and metadata (D0029) */
+    testResults?: StoredTestResultData[] | null;
     timestamp: string;
 }
 /** Shape returned by scoreTimelineQuery */
@@ -732,7 +778,7 @@ declare const scoreTimelineQuery: string;
  *
  * Used by: ReportDetail view
  */
-declare const reportDetailQuery = "\n  *[_type == \"ailf.report\" && reportId == $reportId][0] {\n    _id,\n    reportId,\n    completedAt,\n    durationMs,\n    tag,\n    title,\n    provenance,\n    summary,\n    comparison\n  }\n";
+declare const reportDetailQuery = "\n  *[_type == \"ailf.report\" && reportId == $reportId][0] {\n    _id,\n    reportId,\n    completedAt,\n    durationMs,\n    tag,\n    title,\n    provenance,\n    summary,\n    \"comparison\": comparison {\n      areas,\n      deltas,\n      generatedAt,\n      improved,\n      mismatched,\n      noiseThreshold,\n      noiseThresholdEmpirical,\n      notEvaluated,\n      regressed,\n      unchanged\n    }\n  }\n";
 /**
  * Find all reports that evaluated a specific Sanity document or perspective.
  *