@sanity/ailf-studio 1.3.1 → 1.6.0

package/dist/index.d.ts

@@ -318,6 +318,8 @@ interface ProvenanceData {
         id: string;
         label: string;
     }[];
+    /** Identity of the pipeline run that produced this report (D0032) */
+    runId: string;
     /** @deprecated Use `promptfooUrls` when available */
     promptfooUrl?: string;
     /** Per-mode Promptfoo share URLs (one per sub-eval) */
@@ -434,13 +436,36 @@ interface ScoreItem {
     /** Ceiling score — gold-standard docs injected */
     ceilingScore?: number;
 }
-/** Reference to an artifact stored in an external object store (GCS). */
+/**
+ * A single row in `ArtifactRef.entries[]`. W0051 adds optional `preview` +
+ * `association` + `truncated` so list-view renderers can consume the
+ * descriptor-extracted preview without fetching the external payload.
+ * Older manifests (pre-W0051) carry only `{ key, bytes }`; readers treat
+ * missing fields as absent data, not as errors.
+ */
+interface ArtifactRefEntry {
+    key: string;
+    bytes: number;
+    association?: Record<string, string | number>;
+    truncated?: boolean;
+    preview?: unknown;
+}
+/** Reference to an artifact stored in an external object store. */
 interface ArtifactRef {
-    store: "gcs";
+    store: "gcs" | "local";
     bucket: string;
     path: string;
     bytes?: number;
     entryCount?: number;
+    /**
+     * Added in W0047 / D0032. Missing on pre-W0047 legacy refs, in which case
+     * dispatchers must treat it as `"bulk"` (the only layout that existed then).
+     */
+    layout?: "bulk" | "per-entry";
+    /** Per-entry index (populated for `layout: "per-entry"` refs only). */
+    entries?: ArtifactRefEntry[];
+    truncated?: boolean;
+    preview?: unknown;
 }
 /** A single gap/recommendation from gap analysis */
 interface RecommendationGap {
@@ -503,16 +528,33 @@ interface JudgmentData {
     score: number;
     taskId: string;
 }
-/** Per-feature agent behavior data — how agents interacted with docs */
+/**
+ * Per-feature agent behavior data — how agents interacted with docs.
+ *
+ * W0051 slimmed the full `searchQueries` / `docSlugsVisited` arrays out
+ * of the Report summary and replaced them with `*Sample` (bounded first-N)
+ * + `*Count` (distinct total). Older reports still carry the full arrays
+ * under the legacy names; both shapes are optional here so the
+ * `AgentBehaviorCard` renders either fluidly.
+ */
 interface FeatureAgentBehaviorData {
     avgDocPagesVisited: number;
     avgNetworkTimeMs: number;
     avgSearchesPerformed: number;
-    docSlugsVisited: string[];
     externalDomains: string[];
     feature: string;
-    searchQueries: string[];
     tasksWithBehaviorData: number;
+    /** W0051 slim: bounded sample of unique search queries (first ~5). */
+    searchQueriesSample?: string[];
+    /** W0051 slim: count of distinct queries in the full traces artifact. */
+    searchQueriesCount?: number;
+    /** W0051 slim: bounded sample of unique doc slugs visited. */
+    docSlugsVisitedSample?: string[];
+    /** W0051 slim: count of distinct slugs in the full traces artifact. */
+    docSlugsVisitedCount?: number;
+    /** Legacy (pre-W0051): the full arrays inlined on the Report. */
+    searchQueries?: string[];
+    docSlugsVisited?: string[];
 }
 /** Overall agent behavior stats (aggregated across all features) */
 interface OverallAgentBehaviorData {
@@ -539,11 +581,14 @@ 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?: {
+    /** External artifact references — present when pipeline uploads to GCS (D0032) */
+    artifactManifest?: {
         testOutputs?: ArtifactRef;
         renderedPrompts?: ArtifactRef;
         rawResults?: ArtifactRef;
+        graderPrompts?: ArtifactRef;
+        taskDefinitions?: ArtifactRef;
+        evalResults?: ArtifactRef;
         traces?: ArtifactRef;
     };
     belowCritical: string[];
@@ -569,6 +614,24 @@ interface SummaryData {
     perModel?: PerModelData[] | null;
     /** Gap analysis recommendations (when gap analysis was run) */
     recommendations: null | RecommendationsData;
+    /**
+     * Slim failure-mode summary (W0051). `topTitles[*]` carry the
+     * `graderJudgments`-era `id = formatEntryKey({mode, category})` so the
+     * FailureModesPanel can resolve each row to its per-category manifest
+     * entry via `useFailureModeArtifact`.
+     */
+    failureModes?: {
+        counts: Record<string, number>;
+        topTitles: {
+            id: string;
+            category: string;
+            severity: "low" | "medium" | "high" | "critical";
+            title: string;
+            count: number;
+        }[];
+        totalJudgments: number;
+        classificationRate: number;
+    } | null;
     scores: ScoreItem[];
     /** Per-test results with model output and metadata (D0029) */
     testResults?: StoredTestResultData[] | null;
@@ -658,7 +721,7 @@ declare function searchTopics(query: string): HelpTopic[];
  *
  * The drawer:
  * - Reads current topic from HelpContext
- * - Renders markdown body via HelpMarkdown
+ * - Renders markdown body via the shared <Markdown> component
  * - Shows "See also" links for related topics
  * - Includes a search bar for topic discovery
  * - Supports back navigation through topic history