@mastra/server 1.15.0 → 1.16.0-alpha.0

package/dist/{chunk-M2PIFW7O.cjs → chunk-AOAIJJY6.cjs}

@@ -1,12 +1,16 @@
 'use strict';
 
-var chunkKTKTSTDR_cjs = require('./chunk-KTKTSTDR.cjs');
+var chunkN5RFB7A4_cjs = require('./chunk-N5RFB7A4.cjs');
 var chunkZYXDUS6Q_cjs = require('./chunk-ZYXDUS6Q.cjs');
 var chunkY2BKW5ND_cjs = require('./chunk-Y2BKW5ND.cjs');
 var chunkYJLDMFSE_cjs = require('./chunk-YJLDMFSE.cjs');
 var chunk64ITUOXI_cjs = require('./chunk-64ITUOXI.cjs');
+var agent = require('@mastra/core/agent');
 var error = require('@mastra/core/error');
 var features = require('@mastra/core/features');
+var llm = require('@mastra/core/llm');
+var requestContext = require('@mastra/core/request-context');
+var zod = require('zod');
 
 function assertDatasetsAvailable() {
   if (!features.coreFeatures.has("datasets")) {
@@ -34,8 +38,8 @@ var LIST_DATASETS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets",
   responseType: "json",
-  queryParamSchema: chunkKTKTSTDR_cjs.paginationQuerySchema,
-  responseSchema: chunkKTKTSTDR_cjs.listDatasetsResponseSchema,
+  queryParamSchema: chunkN5RFB7A4_cjs.paginationQuerySchema,
+  responseSchema: chunkN5RFB7A4_cjs.listDatasetsResponseSchema,
   summary: "List all datasets",
   description: "Returns a paginated list of all datasets",
   tags: ["Datasets"],
@@ -61,8 +65,8 @@ var CREATE_DATASET_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "POST",
   path: "/datasets",
   responseType: "json",
-  bodySchema: chunkKTKTSTDR_cjs.createDatasetBodySchema,
-  responseSchema: chunkKTKTSTDR_cjs.datasetResponseSchema,
+  bodySchema: chunkN5RFB7A4_cjs.createDatasetBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.datasetResponseSchema,
   summary: "Create a new dataset",
   description: "Creates a new dataset with the specified name and optional metadata",
   tags: ["Datasets"],
@@ -70,14 +74,25 @@ var CREATE_DATASET_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   handler: async ({ mastra, ...params }) => {
     assertDatasetsAvailable();
     try {
-      const { name, description, metadata, inputSchema, groundTruthSchema, requestContextSchema } = params;
+      const {
+        name,
+        description,
+        metadata,
+        inputSchema,
+        groundTruthSchema,
+        requestContextSchema,
+        targetType,
+        targetIds
+      } = params;
       const ds = await mastra.datasets.create({
         name,
         description,
         metadata,
         inputSchema,
         groundTruthSchema,
-        requestContextSchema
+        requestContextSchema,
+        targetType,
+        targetIds
       });
       const details = await ds.getDetails();
       return details;
@@ -93,8 +108,8 @@ var GET_DATASET_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets/:datasetId",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  responseSchema: chunkKTKTSTDR_cjs.datasetResponseSchema.nullable(),
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  responseSchema: chunkN5RFB7A4_cjs.datasetResponseSchema.nullable(),
   summary: "Get dataset by ID",
   description: "Returns details for a specific dataset",
   tags: ["Datasets"],
@@ -116,9 +131,9 @@ var UPDATE_DATASET_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "PATCH",
   path: "/datasets/:datasetId",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  bodySchema: chunkKTKTSTDR_cjs.updateDatasetBodySchema,
-  responseSchema: chunkKTKTSTDR_cjs.datasetResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  bodySchema: chunkN5RFB7A4_cjs.updateDatasetBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.datasetResponseSchema,
   summary: "Update dataset",
   description: "Updates a dataset with the specified fields",
   tags: ["Datasets"],
@@ -126,7 +141,17 @@ var UPDATE_DATASET_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   handler: async ({ mastra, datasetId, ...params }) => {
     assertDatasetsAvailable();
     try {
-      const { name, description, metadata, inputSchema, groundTruthSchema, requestContextSchema } = params;
+      const {
+        name,
+        description,
+        metadata,
+        inputSchema,
+        groundTruthSchema,
+        requestContextSchema,
+        tags,
+        targetType,
+        targetIds
+      } = params;
       const ds = await mastra.datasets.get({ id: datasetId });
       const result = await ds.update({
         name,
@@ -134,7 +159,10 @@ var UPDATE_DATASET_ROUTE = chunkYJLDMFSE_cjs.createRoute({
         metadata,
         inputSchema,
         groundTruthSchema,
-        requestContextSchema
+        requestContextSchema,
+        tags,
+        targetType,
+        targetIds
       });
       return result;
     } catch (error$1) {
@@ -161,7 +189,7 @@ var DELETE_DATASET_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "DELETE",
   path: "/datasets/:datasetId",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
   responseSchema: chunkZYXDUS6Q_cjs.successResponseSchema,
   summary: "Delete dataset",
   description: "Deletes a dataset and all its items",
@@ -185,9 +213,9 @@ var LIST_ITEMS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets/:datasetId/items",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  queryParamSchema: chunkKTKTSTDR_cjs.listItemsQuerySchema,
-  responseSchema: chunkKTKTSTDR_cjs.listItemsResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  queryParamSchema: chunkN5RFB7A4_cjs.listItemsQuerySchema,
+  responseSchema: chunkN5RFB7A4_cjs.listItemsResponseSchema,
   summary: "List dataset items",
   description: "Returns a paginated list of items in the dataset",
   tags: ["Datasets"],
@@ -219,9 +247,9 @@ var ADD_ITEM_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "POST",
   path: "/datasets/:datasetId/items",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  bodySchema: chunkKTKTSTDR_cjs.addItemBodySchema,
-  responseSchema: chunkKTKTSTDR_cjs.datasetItemResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  bodySchema: chunkN5RFB7A4_cjs.addItemBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.datasetItemResponseSchema,
   summary: "Add item to dataset",
   description: "Adds a new item to the dataset (auto-increments dataset version)",
   tags: ["Datasets"],
@@ -250,8 +278,8 @@ var GET_ITEM_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets/:datasetId/items/:itemId",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetAndItemIdPathParams,
-  responseSchema: chunkKTKTSTDR_cjs.datasetItemResponseSchema.nullable(),
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetAndItemIdPathParams,
+  responseSchema: chunkN5RFB7A4_cjs.datasetItemResponseSchema.nullable(),
   summary: "Get dataset item by ID",
   description: "Returns details for a specific dataset item",
   tags: ["Datasets"],
@@ -277,9 +305,9 @@ var UPDATE_ITEM_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "PATCH",
   path: "/datasets/:datasetId/items/:itemId",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetAndItemIdPathParams,
-  bodySchema: chunkKTKTSTDR_cjs.updateItemBodySchema,
-  responseSchema: chunkKTKTSTDR_cjs.datasetItemResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetAndItemIdPathParams,
+  bodySchema: chunkN5RFB7A4_cjs.updateItemBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.datasetItemResponseSchema,
   summary: "Update dataset item",
   description: "Updates a dataset item (auto-increments dataset version)",
   tags: ["Datasets"],
@@ -312,7 +340,7 @@ var DELETE_ITEM_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "DELETE",
   path: "/datasets/:datasetId/items/:itemId",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetAndItemIdPathParams,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetAndItemIdPathParams,
   responseSchema: chunkZYXDUS6Q_cjs.successResponseSchema,
   summary: "Delete dataset item",
   description: "Deletes a dataset item",
@@ -340,9 +368,9 @@ var LIST_EXPERIMENTS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets/:datasetId/experiments",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  queryParamSchema: chunkKTKTSTDR_cjs.paginationQuerySchema,
-  responseSchema: chunkKTKTSTDR_cjs.listExperimentsResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  queryParamSchema: chunkN5RFB7A4_cjs.paginationQuerySchema,
+  responseSchema: chunkN5RFB7A4_cjs.listExperimentsResponseSchema,
   summary: "List experiments for dataset",
   description: "Returns a paginated list of experiments for the dataset",
   tags: ["Datasets"],
@@ -366,9 +394,9 @@ var TRIGGER_EXPERIMENT_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "POST",
   path: "/datasets/:datasetId/experiments",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  bodySchema: chunkKTKTSTDR_cjs.triggerExperimentBodySchema,
-  responseSchema: chunkKTKTSTDR_cjs.experimentSummaryResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  bodySchema: chunkN5RFB7A4_cjs.triggerExperimentBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.experimentSummaryResponseSchema,
   summary: "Trigger a new experiment",
   description: "Triggers a new experiment on the dataset against the specified target. Returns immediately with pending status; execution happens in background.",
   tags: ["Datasets"],
@@ -376,7 +404,15 @@ var TRIGGER_EXPERIMENT_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   handler: async ({ mastra, datasetId, ...params }) => {
     assertDatasetsAvailable();
     try {
-      const { targetType, targetId, scorerIds, version, maxConcurrency, requestContext } = params;
+      const {
+        targetType,
+        targetId,
+        scorerIds,
+        version,
+        maxConcurrency,
+        requestContext: rawRequestContext
+      } = params;
+      const requestContext$1 = rawRequestContext instanceof requestContext.RequestContext ? rawRequestContext.all : rawRequestContext;
       const ds = await mastra.datasets.get({ id: datasetId });
       const result = await ds.startExperimentAsync({
         targetType,
@@ -384,7 +420,7 @@ var TRIGGER_EXPERIMENT_ROUTE = chunkYJLDMFSE_cjs.createRoute({
         scorers: scorerIds,
         version,
         maxConcurrency,
-        requestContext
+        requestContext: requestContext$1
       });
       return {
         experimentId: result.experimentId,
@@ -408,8 +444,8 @@ var GET_EXPERIMENT_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets/:datasetId/experiments/:experimentId",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetAndExperimentIdPathParams,
-  responseSchema: chunkKTKTSTDR_cjs.experimentResponseSchema.nullable(),
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetAndExperimentIdPathParams,
+  responseSchema: chunkN5RFB7A4_cjs.experimentResponseSchema.nullable(),
   summary: "Get experiment by ID",
   description: "Returns details for a specific experiment",
   tags: ["Datasets"],
@@ -435,9 +471,9 @@ var LIST_EXPERIMENT_RESULTS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets/:datasetId/experiments/:experimentId/results",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetAndExperimentIdPathParams,
-  queryParamSchema: chunkKTKTSTDR_cjs.paginationQuerySchema,
-  responseSchema: chunkKTKTSTDR_cjs.listExperimentResultsResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetAndExperimentIdPathParams,
+  queryParamSchema: chunkN5RFB7A4_cjs.paginationQuerySchema,
+  responseSchema: chunkN5RFB7A4_cjs.listExperimentResultsResponseSchema,
   summary: "List experiment results",
   description: "Returns a paginated list of results for the experiment",
   tags: ["Datasets"],
@@ -464,13 +500,50 @@ var LIST_EXPERIMENT_RESULTS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
     }
   }
 });
+var UPDATE_EXPERIMENT_RESULT_ROUTE = chunkYJLDMFSE_cjs.createRoute({
+  method: "PATCH",
+  path: "/datasets/:datasetId/experiments/:experimentId/results/:resultId",
+  responseType: "json",
+  pathParamSchema: chunkN5RFB7A4_cjs.experimentResultIdPathParams,
+  bodySchema: chunkN5RFB7A4_cjs.updateExperimentResultBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.experimentResultResponseSchema,
+  summary: "Update an experiment result",
+  description: "Updates the status and/or tags on an experiment result",
+  tags: ["Datasets"],
+  requiresAuth: true,
+  handler: async ({ mastra, resultId, experimentId, ...params }) => {
+    assertDatasetsAvailable();
+    try {
+      const storage = mastra.getStorage();
+      if (!storage) {
+        throw new chunk64ITUOXI_cjs.HTTPException(500, { message: "Storage not configured" });
+      }
+      const experimentsStore = await storage.getStore("experiments");
+      if (!experimentsStore) {
+        throw new chunk64ITUOXI_cjs.HTTPException(500, { message: "Experiments storage not available" });
+      }
+      const result = await experimentsStore.updateExperimentResult({
+        id: resultId,
+        experimentId,
+        status: params.status,
+        tags: params.tags
+      });
+      return result;
+    } catch (error$1) {
+      if (error$1 instanceof error.MastraError) {
+        throw new chunk64ITUOXI_cjs.HTTPException(getHttpStatusForMastraError(error$1.id), { message: error$1.message });
+      }
+      return chunkY2BKW5ND_cjs.handleError(error$1, "Error updating experiment result");
+    }
+  }
+});
 var COMPARE_EXPERIMENTS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "POST",
   path: "/datasets/:datasetId/compare",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  bodySchema: chunkKTKTSTDR_cjs.compareExperimentsBodySchema,
-  responseSchema: chunkKTKTSTDR_cjs.comparisonResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  bodySchema: chunkN5RFB7A4_cjs.compareExperimentsBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.comparisonResponseSchema,
   summary: "Compare two experiments",
   description: "Compares two experiments to detect score regressions",
   tags: ["Datasets"],
@@ -497,9 +570,9 @@ var LIST_DATASET_VERSIONS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets/:datasetId/versions",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  queryParamSchema: chunkKTKTSTDR_cjs.paginationQuerySchema,
-  responseSchema: chunkKTKTSTDR_cjs.listDatasetVersionsResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  queryParamSchema: chunkN5RFB7A4_cjs.paginationQuerySchema,
+  responseSchema: chunkN5RFB7A4_cjs.listDatasetVersionsResponseSchema,
   summary: "List dataset versions",
   description: "Returns a paginated list of all versions for the dataset",
   tags: ["Datasets"],
@@ -523,8 +596,8 @@ var LIST_ITEM_VERSIONS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets/:datasetId/items/:itemId/history",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetAndItemIdPathParams,
-  responseSchema: chunkKTKTSTDR_cjs.listItemVersionsResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetAndItemIdPathParams,
+  responseSchema: chunkN5RFB7A4_cjs.listItemVersionsResponseSchema,
   summary: "Get item history",
   description: "Returns the full SCD-2 history of the item across all dataset versions",
   tags: ["Datasets"],
@@ -550,8 +623,8 @@ var GET_ITEM_VERSION_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "GET",
   path: "/datasets/:datasetId/items/:itemId/versions/:datasetVersion",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetItemVersionPathParams,
-  responseSchema: chunkKTKTSTDR_cjs.datasetItemResponseSchema.nullable(),
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetItemVersionPathParams,
+  responseSchema: chunkN5RFB7A4_cjs.datasetItemResponseSchema.nullable(),
   summary: "Get item at specific dataset version",
   description: "Returns the item as it existed at a specific dataset version",
   tags: ["Datasets"],
@@ -580,9 +653,9 @@ var BATCH_INSERT_ITEMS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "POST",
   path: "/datasets/:datasetId/items/batch",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  bodySchema: chunkKTKTSTDR_cjs.batchInsertItemsBodySchema,
-  responseSchema: chunkKTKTSTDR_cjs.batchInsertItemsResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  bodySchema: chunkN5RFB7A4_cjs.batchInsertItemsBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.batchInsertItemsResponseSchema,
   summary: "Batch insert items to dataset",
   description: "Adds multiple items to the dataset in a single operation (single version entry)",
   tags: ["Datasets"],
@@ -612,9 +685,9 @@ var BATCH_DELETE_ITEMS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
   method: "DELETE",
   path: "/datasets/:datasetId/items/batch",
   responseType: "json",
-  pathParamSchema: chunkKTKTSTDR_cjs.datasetIdPathParams,
-  bodySchema: chunkKTKTSTDR_cjs.batchDeleteItemsBodySchema,
-  responseSchema: chunkKTKTSTDR_cjs.batchDeleteItemsResponseSchema,
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  bodySchema: chunkN5RFB7A4_cjs.batchDeleteItemsBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.batchDeleteItemsResponseSchema,
   summary: "Batch delete items from dataset",
   description: "Deletes multiple items from the dataset in a single operation (single version entry)",
   tags: ["Datasets"],
@@ -634,14 +707,213 @@ var BATCH_DELETE_ITEMS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
     }
   }
 });
+var GENERATE_ITEMS_SYSTEM_PROMPT = `You are a test data generation expert. Your job is to generate realistic, diverse test data items for an AI agent evaluation dataset.
+
+You will be given context about the agent being tested \u2014 its purpose, system prompt, and available tools. Use this to generate inputs that thoroughly exercise the agent's capabilities.
+
+Generate test items that:
+1. Are realistic and diverse \u2014 cover edge cases, different complexities, and various scenarios
+2. Match the provided schemas exactly
+3. Include ground truth values when a ground truth schema is provided
+4. Vary in difficulty (easy, medium, hard cases)
+5. Include potential edge cases and tricky inputs
+6. Test different aspects of the agent's capabilities based on its tools and instructions
+
+Return the items as a JSON array.`;
+var GENERATE_ITEMS_ROUTE = chunkYJLDMFSE_cjs.createRoute({
+  method: "POST",
+  path: "/datasets/:datasetId/generate-items",
+  responseType: "json",
+  pathParamSchema: chunkN5RFB7A4_cjs.datasetIdPathParams,
+  bodySchema: chunkN5RFB7A4_cjs.generateItemsBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.generateItemsResponseSchema,
+  summary: "Generate dataset items using AI",
+  description: "Uses an LLM to generate synthetic dataset items based on the dataset schema and a user prompt. Returns generated items for review \u2014 they are NOT automatically added to the dataset.",
+  tags: ["Datasets"],
+  requiresAuth: true,
+  handler: async ({ mastra, datasetId, modelId, prompt, count, agentContext }) => {
+    assertDatasetsAvailable();
+    try {
+      const ds = await mastra.datasets.get({ id: datasetId });
+      const dataset = await ds.getDetails();
+      const model = await llm.resolveModelConfig(modelId, void 0, mastra);
+      const schemaContext = [
+        dataset.inputSchema ? `Input schema:
+${JSON.stringify(dataset.inputSchema, null, 2)}` : null,
+        dataset.groundTruthSchema ? `Ground truth schema:
+${JSON.stringify(dataset.groundTruthSchema, null, 2)}` : null
+      ].filter(Boolean).join("\n\n");
+      const generatorAgent = new agent.Agent({
+        id: "dataset-item-generator",
+        name: "dataset-item-generator",
+        instructions: GENERATE_ITEMS_SYSTEM_PROMPT,
+        model
+      });
+      const itemSchema = zod.z.object({
+        input: zod.z.string().describe("The input data as a JSON string matching the input schema, or a plain text string if no schema"),
+        groundTruth: zod.z.string().optional().describe("The expected output as a JSON string matching the ground truth schema")
+      });
+      const outputSchema = zod.z.object({
+        items: zod.z.array(itemSchema).min(1).max(count)
+      });
+      const agentContextParts = [];
+      if (agentContext?.description) {
+        agentContextParts.push(`Agent description: ${agentContext.description}`);
+      }
+      if (agentContext?.instructions) {
+        agentContextParts.push(`Agent system prompt:
+${agentContext.instructions}`);
+      }
+      if (agentContext?.tools?.length) {
+        agentContextParts.push(`Agent tools: ${agentContext.tools.join(", ")}`);
+      }
+      const agentContextSection = agentContextParts.length > 0 ? agentContextParts.join("\n\n") : null;
+      const userMessage = [
+        `Generate exactly ${count} test items for a dataset named "${dataset.name}".`,
+        dataset.description ? `Dataset description: ${dataset.description}` : null,
+        agentContextSection ? `--- AGENT CONTEXT ---
+${agentContextSection}` : null,
+        schemaContext || null,
+        `User's request: ${prompt}`,
+        `Return exactly ${count} items.`
+      ].filter(Boolean).join("\n\n");
+      const result = await generatorAgent.generate(userMessage, {
+        structuredOutput: { schema: outputSchema }
+      });
+      const generated = await result.object;
+      const items = generated.items.map((item) => {
+        let input = item.input;
+        try {
+          input = JSON.parse(item.input);
+        } catch {
+        }
+        let groundTruth = item.groundTruth;
+        if (item.groundTruth) {
+          try {
+            groundTruth = JSON.parse(item.groundTruth);
+          } catch {
+          }
+        }
+        return { input, groundTruth };
+      });
+      return { items };
+    } catch (error$1) {
+      if (error$1 instanceof error.MastraError) {
+        throw new chunk64ITUOXI_cjs.HTTPException(getHttpStatusForMastraError(error$1.id), { message: error$1.message });
+      }
+      return chunkY2BKW5ND_cjs.handleError(error$1, "Error generating dataset items");
+    }
+  }
+});
+var CLUSTER_FAILURES_SYSTEM_PROMPT = `You are an AI evaluation expert specializing in failure analysis. Given a set of failure items from an AI agent experiment, identify common failure patterns and assign descriptive tags to each item.
+
+For each cluster you identify, provide:
+- A short, descriptive tag label (2-5 words, lowercase, hyphenated, e.g., "no-tool-usage", "hallucination")
+- A description explaining the common failure pattern
+- The IDs of items that belong to this cluster
+
+Also return a "proposedTags" array mapping each item ID to the tags you recommend, along with a brief "reason" explaining WHY those tags apply to that specific item. The reason should reference concrete evidence from the item's input/output/error.
+
+Guidelines:
+- Create between 1 and 8 clusters depending on the diversity of failures
+- Every item must be assigned to at least one cluster unless there is no clear pattern of failure
+- Focus on the root cause of failures, not surface-level symptoms
+- If items have scores, use low scores as signals for the failure type
+- Be specific about what went wrong
+- IMPORTANT: If existing tags are provided, PREFER reusing them over creating new ones. Only create new tags when no existing tag fits.
+- Items may already have tags \u2014 consider those when assigning new ones and avoid duplicating existing tags on an item.
+- The "reason" field should be 1-2 sentences explaining the specific evidence for each tag assignment.`;
+var CLUSTER_FAILURES_ROUTE = chunkYJLDMFSE_cjs.createRoute({
+  method: "POST",
+  path: "/datasets/cluster-failures",
+  responseType: "json",
+  bodySchema: chunkN5RFB7A4_cjs.clusterFailuresBodySchema,
+  responseSchema: chunkN5RFB7A4_cjs.clusterFailuresResponseSchema,
+  summary: "Cluster experiment failures using AI",
+  description: "Uses an LLM to analyze failure items from an experiment and group them into meaningful failure pattern clusters.",
+  tags: ["Datasets"],
+  requiresAuth: true,
+  handler: async ({ mastra, modelId, items, availableTags, prompt }) => {
+    assertDatasetsAvailable();
+    try {
+      const model = await llm.resolveModelConfig(modelId, void 0, mastra);
+      const clusterAgent = new agent.Agent({
+        id: "failure-cluster-analyzer",
+        name: "failure-cluster-analyzer",
+        instructions: CLUSTER_FAILURES_SYSTEM_PROMPT,
+        model
+      });
+      const outputSchema = zod.z.object({
+        clusters: zod.z.array(
+          zod.z.object({
+            id: zod.z.string(),
+            label: zod.z.string(),
+            description: zod.z.string(),
+            itemIds: zod.z.array(zod.z.string())
+          })
+        ),
+        proposedTags: zod.z.array(
+          zod.z.object({
+            itemId: zod.z.string(),
+            tags: zod.z.array(zod.z.string()),
+            reason: zod.z.string().describe("Brief explanation of why these tags were assigned")
+          })
+        )
+      });
+      const itemSummaries = items.map((item, i) => {
+        const parts = [`Item ${i + 1} (id: ${item.id}):`];
+        if (item.input !== void 0 && item.input !== null) parts.push(`  Input: ${JSON.stringify(item.input)}`);
+        if (item.output !== void 0 && item.output !== null) parts.push(`  Output: ${JSON.stringify(item.output)}`);
+        if (item.error !== void 0 && item.error !== null) {
+          parts.push(`  Error: ${typeof item.error === "string" ? item.error : JSON.stringify(item.error)}`);
+        }
+        if (item.scores !== void 0 && item.scores !== null) {
+          parts.push(`  Scores: ${JSON.stringify(item.scores)}`);
+        }
+        if (item.existingTags && item.existingTags.length > 0) {
+          parts.push(`  Existing tags: ${item.existingTags.join(", ")}`);
+        }
+        return parts.join("\n");
+      });
+      let userMessage = `Analyze these ${items.length} failure items and group them into clusters of common failure patterns:
+
+${itemSummaries.join("\n\n")}`;
+      if (availableTags && availableTags.length > 0) {
+        userMessage += `
+
+Existing tag vocabulary (prefer reusing these): ${availableTags.join(", ")}`;
+      }
+      if (prompt) {
+        userMessage += `
+
+Additional instructions from the reviewer: ${prompt}`;
+      }
+      userMessage += `
+
+Return both "clusters" (grouping items by pattern) and "proposedTags" (a list mapping each item ID to the tag labels you recommend, with a "reason" explaining why). For proposedTags, only include NEW tags to add \u2014 do not repeat tags the item already has.`;
+      const result = await clusterAgent.generate(userMessage, {
+        structuredOutput: { schema: outputSchema }
+      });
+      const generated = await result.object;
+      return { clusters: generated.clusters, proposedTags: generated.proposedTags ?? [] };
+    } catch (error$1) {
+      if (error$1 instanceof error.MastraError) {
+        throw new chunk64ITUOXI_cjs.HTTPException(getHttpStatusForMastraError(error$1.id), { message: error$1.message });
+      }
+      return chunkY2BKW5ND_cjs.handleError(error$1, "Error clustering failures");
+    }
+  }
+});
 
 exports.ADD_ITEM_ROUTE = ADD_ITEM_ROUTE;
 exports.BATCH_DELETE_ITEMS_ROUTE = BATCH_DELETE_ITEMS_ROUTE;
 exports.BATCH_INSERT_ITEMS_ROUTE = BATCH_INSERT_ITEMS_ROUTE;
+exports.CLUSTER_FAILURES_ROUTE = CLUSTER_FAILURES_ROUTE;
 exports.COMPARE_EXPERIMENTS_ROUTE = COMPARE_EXPERIMENTS_ROUTE;
 exports.CREATE_DATASET_ROUTE = CREATE_DATASET_ROUTE;
 exports.DELETE_DATASET_ROUTE = DELETE_DATASET_ROUTE;
 exports.DELETE_ITEM_ROUTE = DELETE_ITEM_ROUTE;
+exports.GENERATE_ITEMS_ROUTE = GENERATE_ITEMS_ROUTE;
 exports.GET_DATASET_ROUTE = GET_DATASET_ROUTE;
 exports.GET_EXPERIMENT_ROUTE = GET_EXPERIMENT_ROUTE;
 exports.GET_ITEM_ROUTE = GET_ITEM_ROUTE;
@@ -654,6 +926,7 @@ exports.LIST_ITEMS_ROUTE = LIST_ITEMS_ROUTE;
 exports.LIST_ITEM_VERSIONS_ROUTE = LIST_ITEM_VERSIONS_ROUTE;
 exports.TRIGGER_EXPERIMENT_ROUTE = TRIGGER_EXPERIMENT_ROUTE;
 exports.UPDATE_DATASET_ROUTE = UPDATE_DATASET_ROUTE;
+exports.UPDATE_EXPERIMENT_RESULT_ROUTE = UPDATE_EXPERIMENT_RESULT_ROUTE;
 exports.UPDATE_ITEM_ROUTE = UPDATE_ITEM_ROUTE;
-//# sourceMappingURL=chunk-M2PIFW7O.cjs.map
-//# sourceMappingURL=chunk-M2PIFW7O.cjs.map
+//# sourceMappingURL=chunk-AOAIJJY6.cjs.map
+//# sourceMappingURL=chunk-AOAIJJY6.cjs.map