@wix/evalforge-types 0.72.0 → 0.74.0

package/build/index.mjs

@@ -584,11 +584,42 @@ var EnvironmentSchema = z21.object({
 });
 
 // src/scenario/test-scenario.ts
-import { z as z23 } from "zod";
+import { z as z24 } from "zod";
 
 // src/assertion/assertion.ts
+import { z as z23 } from "zod";
+
+// src/assertion/build-passed-command.ts
 import { z as z22 } from "zod";
-var AssertionTypeSchema = z22.enum([
+var ALLOWED_BUILD_COMMANDS = [
+  "yarn build",
+  "npm run build",
+  "pnpm run build",
+  "pnpm build"
+];
+var DEFAULT_BUILD_PASSED_COMMAND = "yarn build";
+var BUILD_COMMAND_ARGV = {
+  "yarn build": ["yarn", "build"],
+  "npm run build": ["npm", "run", "build"],
+  "pnpm run build": ["pnpm", "run", "build"],
+  "pnpm build": ["pnpm", "build"]
+};
+function isAllowedBuildCommandString(command) {
+  const trimmed = command.trim();
+  return ALLOWED_BUILD_COMMANDS.includes(trimmed);
+}
+function parseBuildCommandToArgv(command) {
+  const trimmed = command.trim();
+  if (!(trimmed in BUILD_COMMAND_ARGV)) {
+    return null;
+  }
+  return BUILD_COMMAND_ARGV[trimmed];
+}
+var enumTuple = ALLOWED_BUILD_COMMANDS;
+var BuildPassedCommandStringSchema = z22.enum(enumTuple);
+
+// src/assertion/assertion.ts
+var AssertionTypeSchema = z23.enum([
   "skill_was_called",
   "tool_called_with_param",
   "build_passed",
@@ -597,61 +628,61 @@ var AssertionTypeSchema = z22.enum([
   "llm_judge",
   "api_call"
 ]);
-var AssertionParameterTypeSchema = z22.enum([
+var AssertionParameterTypeSchema = z23.enum([
   "string",
   "number",
   "boolean"
 ]);
-var AssertionParameterSchema = z22.object({
+var AssertionParameterSchema = z23.object({
   /** Parameter name (used as key in params object) */
-  name: z22.string().min(1),
+  name: z23.string().min(1),
   /** Display label for the parameter */
-  label: z22.string().min(1),
+  label: z23.string().min(1),
   /** Parameter type */
   type: AssertionParameterTypeSchema,
   /** Whether this parameter is required */
-  required: z22.boolean(),
+  required: z23.boolean(),
   /** Default value (optional, used when not provided) */
-  defaultValue: z22.union([z22.string(), z22.number(), z22.boolean()]).optional(),
+  defaultValue: z23.union([z23.string(), z23.number(), z23.boolean()]).optional(),
   /** If true, parameter is hidden by default behind "Show advanced options" */
-  advanced: z22.boolean().optional()
+  advanced: z23.boolean().optional()
 });
-var ScenarioAssertionLinkSchema = z22.object({
+var ScenarioAssertionLinkSchema = z23.object({
   /** ID of the system assertion (e.g., 'system:skill_was_called') */
-  assertionId: z22.string(),
+  assertionId: z23.string(),
   /** Parameter values for this assertion in this scenario */
-  params: z22.record(
-    z22.string(),
-    z22.union([z22.string(), z22.number(), z22.boolean(), z22.null()])
+  params: z23.record(
+    z23.string(),
+    z23.union([z23.string(), z23.number(), z23.boolean(), z23.null()])
   ).optional()
 });
-var SkillWasCalledConfigSchema = z22.object({
+var SkillWasCalledConfigSchema = z23.object({
   /** Names of the skills that must have been called */
-  skillNames: z22.array(z22.string().min(1)).min(1)
+  skillNames: z23.array(z23.string().min(1)).min(1)
 });
-var CostConfigSchema = z22.strictObject({
+var CostConfigSchema = z23.strictObject({
   /** Maximum allowed cost in USD */
-  maxCostUsd: z22.number().positive()
+  maxCostUsd: z23.number().positive()
 });
-var ToolCalledWithParamConfigSchema = z22.strictObject({
+var ToolCalledWithParamConfigSchema = z23.strictObject({
   /** Name of the tool that must have been called */
-  toolName: z22.string().min(1),
+  toolName: z23.string().min(1),
   /** JSON string of key-value pairs for expected parameters (substring match). Optional — when omitted, only checks tool presence. */
-  expectedParams: z22.string().min(1).optional(),
+  expectedParams: z23.string().min(1).optional(),
   /** If true, the matching tool call must also have succeeded (step.success === true) */
-  requireSuccess: z22.boolean().optional()
+  requireSuccess: z23.boolean().optional()
 });
-var BuildPassedConfigSchema = z22.strictObject({
-  /** Command to run (default: "yarn build") */
-  command: z22.string().optional(),
+var BuildPassedConfigSchema = z23.strictObject({
+  /** Allowlisted command only (default at runtime: "yarn build") */
+  command: BuildPassedCommandStringSchema.optional(),
   /** Expected exit code (default: 0) */
-  expectedExitCode: z22.number().int().optional()
+  expectedExitCode: z23.number().int().optional()
 });
-var TimeConfigSchema = z22.strictObject({
+var TimeConfigSchema = z23.strictObject({
   /** Maximum allowed duration in milliseconds */
-  maxDurationMs: z22.number().int().positive()
+  maxDurationMs: z23.number().int().positive()
 });
-var LlmJudgeConfigSchema = z22.object({
+var LlmJudgeConfigSchema = z23.object({
   /**
    * Prompt template with placeholders:
    * - {{output}}: agent's final output
@@ -662,65 +693,65 @@ var LlmJudgeConfigSchema = z22.object({
    * - {{trace}}: step-by-step trace of tool calls
    * - Custom parameters defined in the parameters array
    */
-  prompt: z22.string().min(1),
+  prompt: z23.string().min(1),
   /** Minimum score to pass (0-10, default 7) */
-  minScore: z22.number().int().min(0).max(10).optional(),
+  minScore: z23.number().int().min(0).max(10).optional(),
   /** Model for the judge (e.g. claude-3-5-haiku-20241022) */
-  model: z22.string().optional(),
+  model: z23.string().optional(),
   /** Max output tokens */
-  maxTokens: z22.number().int().optional(),
+  maxTokens: z23.number().int().optional(),
   /** Temperature (0-1) */
-  temperature: z22.number().min(0).max(1).optional(),
+  temperature: z23.number().min(0).max(1).optional(),
   /** User-defined parameters for this assertion */
-  parameters: z22.array(AssertionParameterSchema).optional()
+  parameters: z23.array(AssertionParameterSchema).optional()
 });
-var ApiCallConfigSchema = z22.strictObject({
+var ApiCallConfigSchema = z23.strictObject({
   /** URL to call */
-  url: z22.string().min(1),
+  url: z23.string().min(1),
   /** HTTP method (default GET) */
-  method: z22.enum(["GET", "POST"]).optional(),
+  method: z23.enum(["GET", "POST"]).optional(),
   /** Request body (JSON string, for POST requests) */
-  requestBody: z22.string().optional(),
+  requestBody: z23.string().optional(),
   /** Expected JSON response to validate against (subset match — extra fields in actual are OK) */
-  expectedResponse: z22.string().min(1),
+  expectedResponse: z23.string().min(1),
   /** Request headers as JSON string of key-value pairs */
-  requestHeaders: z22.string().optional(),
+  requestHeaders: z23.string().optional(),
   /** Request timeout in milliseconds (default 30000) */
-  timeoutMs: z22.number().int().positive().optional()
+  timeoutMs: z23.number().int().positive().optional()
 });
 var AssertionBaseFields = {
   /** When true, the assertion's pass/fail logic is inverted (NOT operator). */
-  negate: z22.boolean().optional()
+  negate: z23.boolean().optional()
 };
 var SkillWasCalledAssertionSchema = SkillWasCalledConfigSchema.extend({
-  type: z22.literal("skill_was_called"),
+  type: z23.literal("skill_was_called"),
   ...AssertionBaseFields
 });
 var ToolCalledWithParamAssertionSchema = ToolCalledWithParamConfigSchema.extend({
-  type: z22.literal("tool_called_with_param"),
+  type: z23.literal("tool_called_with_param"),
   ...AssertionBaseFields
 });
 var BuildPassedAssertionSchema = BuildPassedConfigSchema.extend({
-  type: z22.literal("build_passed"),
+  type: z23.literal("build_passed"),
   ...AssertionBaseFields
 });
 var CostAssertionSchema = CostConfigSchema.extend({
-  type: z22.literal("cost"),
+  type: z23.literal("cost"),
   ...AssertionBaseFields
 });
 var LlmJudgeAssertionSchema = LlmJudgeConfigSchema.extend({
-  type: z22.literal("llm_judge"),
+  type: z23.literal("llm_judge"),
   ...AssertionBaseFields
 });
 var ApiCallAssertionSchema = ApiCallConfigSchema.extend({
-  type: z22.literal("api_call"),
+  type: z23.literal("api_call"),
   ...AssertionBaseFields
 });
 var TimeAssertionSchema = TimeConfigSchema.extend({
-  type: z22.literal("time_limit"),
+  type: z23.literal("time_limit"),
   ...AssertionBaseFields
 });
-var AssertionSchema = z22.union([
+var AssertionSchema = z23.union([
   SkillWasCalledAssertionSchema,
   ToolCalledWithParamAssertionSchema,
   BuildPassedAssertionSchema,
@@ -729,7 +760,7 @@ var AssertionSchema = z22.union([
   LlmJudgeAssertionSchema,
   ApiCallAssertionSchema
 ]);
-var AssertionConfigSchema = z22.union([
+var AssertionConfigSchema = z23.union([
   LlmJudgeConfigSchema,
   // requires prompt - check first
   SkillWasCalledConfigSchema,
@@ -744,7 +775,7 @@ var AssertionConfigSchema = z22.union([
   // requires maxCostUsd, uses strictObject
   BuildPassedConfigSchema,
   // all optional, uses strictObject to reject unknown keys
-  z22.object({})
+  z23.object({})
   // fallback empty config
 ]);
 function validateAssertionConfig(type, config) {
@@ -768,63 +799,322 @@ function validateAssertionConfig(type, config) {
   }
 }
 
+// src/assertion/system-assertions.ts
+var SYSTEM_ASSERTION_IDS = {
+  SKILL_WAS_CALLED: "system:skill_was_called",
+  TOOL_CALLED_WITH_PARAM: "system:tool_called_with_param",
+  BUILD_PASSED: "system:build_passed",
+  TIME_LIMIT: "system:time_limit",
+  COST: "system:cost",
+  LLM_JUDGE: "system:llm_judge",
+  API_CALL: "system:api_call"
+};
+function isSystemAssertionId(id) {
+  return id.startsWith("system:");
+}
+var SYSTEM_ASSERTIONS = {
+  [SYSTEM_ASSERTION_IDS.SKILL_WAS_CALLED]: {
+    id: SYSTEM_ASSERTION_IDS.SKILL_WAS_CALLED,
+    name: "Skill Was Called",
+    description: "Check that one or more skills were invoked during the agent run",
+    type: "skill_was_called",
+    parameters: [
+      {
+        name: "skillNames",
+        label: "Skills",
+        type: "string",
+        required: true
+      },
+      {
+        name: "negate",
+        label: "Negate (NOT operator)",
+        type: "boolean",
+        required: false,
+        defaultValue: false
+      }
+    ]
+  },
+  [SYSTEM_ASSERTION_IDS.TOOL_CALLED_WITH_PARAM]: {
+    id: SYSTEM_ASSERTION_IDS.TOOL_CALLED_WITH_PARAM,
+    name: "Tool Called With Param",
+    description: "Check that a tool was called with expected parameters (tool name is substring matched)",
+    type: "tool_called_with_param",
+    parameters: [
+      {
+        name: "toolName",
+        label: "Tool Name",
+        type: "string",
+        required: true
+      },
+      {
+        name: "expectedParams",
+        label: "Expected Parameters (JSON, substring match)",
+        type: "string",
+        required: false
+      },
+      {
+        name: "requireSuccess",
+        label: "Require Successful Call",
+        type: "boolean",
+        required: false,
+        defaultValue: false,
+        advanced: true
+      },
+      {
+        name: "negate",
+        label: "Negate (NOT operator)",
+        type: "boolean",
+        required: false,
+        defaultValue: false
+      }
+    ]
+  },
+  [SYSTEM_ASSERTION_IDS.BUILD_PASSED]: {
+    id: SYSTEM_ASSERTION_IDS.BUILD_PASSED,
+    name: "Build Passed",
+    description: "Run a build command and verify it exits with expected code",
+    type: "build_passed",
+    parameters: [
+      {
+        name: "command",
+        label: "Build Command",
+        type: "string",
+        required: false,
+        defaultValue: "yarn build"
+      },
+      {
+        name: "expectedExitCode",
+        label: "Expected Exit Code",
+        type: "number",
+        required: false,
+        defaultValue: 0
+      },
+      {
+        name: "maxBuildTime",
+        label: "Max Build Time (ms)",
+        type: "number",
+        required: false,
+        advanced: true
+      },
+      {
+        name: "maxMemory",
+        label: "Max Memory (MB)",
+        type: "number",
+        required: false,
+        advanced: true
+      }
+    ]
+  },
+  [SYSTEM_ASSERTION_IDS.TIME_LIMIT]: {
+    id: SYSTEM_ASSERTION_IDS.TIME_LIMIT,
+    name: "Time Limit",
+    description: "Check that the scenario completed within a maximum duration",
+    type: "time_limit",
+    parameters: [
+      {
+        name: "maxDurationMs",
+        label: "Max Duration (ms)",
+        type: "number",
+        required: true,
+        defaultValue: 3e5
+      }
+    ]
+  },
+  [SYSTEM_ASSERTION_IDS.COST]: {
+    id: SYSTEM_ASSERTION_IDS.COST,
+    name: "Cost",
+    description: "Check that the scenario LLM execution cost stays within a USD threshold",
+    type: "cost",
+    parameters: [
+      {
+        name: "maxCostUsd",
+        label: "Max Cost (USD)",
+        type: "number",
+        required: true,
+        defaultValue: 1
+      }
+    ]
+  },
+  [SYSTEM_ASSERTION_IDS.LLM_JUDGE]: {
+    id: SYSTEM_ASSERTION_IDS.LLM_JUDGE,
+    name: "LLM Judge",
+    description: "LLM evaluates the output and assigns a score (0-10)",
+    type: "llm_judge",
+    parameters: [
+      {
+        name: "prompt",
+        label: "Judge Prompt",
+        type: "string",
+        required: true,
+        defaultValue: "Verify the output meets the acceptance criteria."
+      },
+      {
+        name: "minScore",
+        label: "Minimum Score (0-10)",
+        type: "number",
+        required: false,
+        defaultValue: 7
+      },
+      {
+        name: "model",
+        label: "Model",
+        type: "string",
+        required: false
+      }
+    ]
+  },
+  [SYSTEM_ASSERTION_IDS.API_CALL]: {
+    id: SYSTEM_ASSERTION_IDS.API_CALL,
+    name: "API Call",
+    description: "Call an API endpoint and verify the response contains expected data",
+    type: "api_call",
+    parameters: [
+      {
+        name: "url",
+        label: "URL",
+        type: "string",
+        required: true
+      },
+      {
+        name: "method",
+        label: "HTTP Method",
+        type: "string",
+        required: false,
+        defaultValue: "GET"
+      },
+      {
+        name: "requestBody",
+        label: "Request Body (JSON)",
+        type: "string",
+        required: false
+      },
+      {
+        name: "expectedResponse",
+        label: "Expected Response (JSON)",
+        type: "string",
+        required: true
+      },
+      {
+        name: "requestHeaders",
+        label: "Headers (JSON)",
+        type: "string",
+        required: false,
+        advanced: true
+      },
+      {
+        name: "timeoutMs",
+        label: "Timeout (ms)",
+        type: "number",
+        required: false,
+        defaultValue: 3e4,
+        advanced: true
+      }
+    ]
+  }
+};
+function getSystemAssertions() {
+  return Object.values(SYSTEM_ASSERTIONS);
+}
+function getSystemAssertion(id) {
+  return SYSTEM_ASSERTIONS[id];
+}
+
 // src/scenario/test-scenario.ts
 var MAX_IMAGE_BASE64_LENGTH = 4 * Math.ceil(2 * 1024 * 1024 / 3);
-var TriggerPromptImageSchema = z23.object({
+var TriggerPromptImageSchema = z24.object({
   /** Base64-encoded image data (no data URL prefix) */
-  base64: z23.string().max(MAX_IMAGE_BASE64_LENGTH, "Image exceeds 2 MB size limit"),
+  base64: z24.string().max(MAX_IMAGE_BASE64_LENGTH, "Image exceeds 2 MB size limit"),
   /** MIME type of the image */
-  mediaType: z23.enum(["image/jpeg", "image/png", "image/gif", "image/webp"]),
+  mediaType: z24.enum(["image/jpeg", "image/png", "image/gif", "image/webp"]),
   /** Original filename of the image */
-  name: z23.string()
+  name: z24.string()
 });
-var ExpectedFileSchema = z23.object({
+var ExpectedFileSchema = z24.object({
   /** Relative path where the file should be created */
-  path: z23.string(),
+  path: z24.string(),
   /** Optional expected content */
-  content: z23.string().optional()
+  content: z24.string().optional()
 });
 var TestScenarioSchema = TenantEntitySchema.extend({
   /** The prompt sent to the agent to trigger the task */
-  triggerPrompt: z23.string().min(10),
+  triggerPrompt: z24.string().min(10),
   /** ID of the template to use for this scenario (null = no template) */
-  templateId: z23.string().nullish(),
+  templateId: z24.string().nullish(),
   /** Inline assertions to evaluate for this scenario (legacy) */
-  assertions: z23.array(AssertionSchema).optional(),
+  assertions: z24.array(AssertionSchema).optional(),
   /** IDs of saved assertions to evaluate (from assertions table) - legacy, use assertionLinks */
-  assertionIds: z23.array(z23.string()).optional(),
+  assertionIds: z24.array(z24.string()).optional(),
   /** Linked assertions with per-scenario parameter values */
-  assertionLinks: z23.array(ScenarioAssertionLinkSchema).optional(),
+  assertionLinks: z24.array(ScenarioAssertionLinkSchema).optional(),
   /** Tags for categorisation and filtering */
-  tags: z23.array(z23.string()).optional(),
+  tags: z24.array(z24.string()).optional(),
   /** Base64-encoded images attached to the trigger prompt (max 3) */
-  triggerPromptImages: z23.array(TriggerPromptImageSchema).max(3).optional()
-});
-var CreateTestScenarioInputSchema = TestScenarioSchema.omit({
+  triggerPromptImages: z24.array(TriggerPromptImageSchema).max(3).optional()
+});
+function validateBuildPassedParamsInAssertionLinks(links, ctx) {
+  if (!links) return;
+  for (let i = 0; i < links.length; i++) {
+    const link = links[i];
+    if (link.assertionId !== SYSTEM_ASSERTION_IDS.BUILD_PASSED) continue;
+    const cmd = link.params?.command;
+    if (cmd === void 0 || cmd === null) continue;
+    if (typeof cmd !== "string") {
+      ctx.addIssue({
+        code: z24.ZodIssueCode.custom,
+        message: "build_passed command must be a string",
+        path: ["assertionLinks", i, "params", "command"]
+      });
+      continue;
+    }
+    if (!isAllowedBuildCommandString(cmd)) {
+      ctx.addIssue({
+        code: z24.ZodIssueCode.custom,
+        message: "Invalid build_passed command. Allowed: yarn build, npm run build, pnpm run build, pnpm build",
+        path: ["assertionLinks", i, "params", "command"]
+      });
+    }
+  }
+}
+var TestScenarioCreateBaseSchema = TestScenarioSchema.omit({
   id: true,
   createdAt: true,
   updatedAt: true,
   deleted: true
 });
-var UpdateTestScenarioInputSchema = CreateTestScenarioInputSchema.partial();
+var CreateTestScenarioInputSchema = TestScenarioCreateBaseSchema.superRefine((data, ctx) => {
+  validateBuildPassedParamsInAssertionLinks(data.assertionLinks, ctx);
+});
+var UpdateTestScenarioInputSchema = TestScenarioCreateBaseSchema.partial().superRefine((data, ctx) => {
+  if (data.assertionLinks !== void 0) {
+    validateBuildPassedParamsInAssertionLinks(data.assertionLinks, ctx);
+  }
+});
 
 // src/scenario/batch-import.ts
-import { z as z24 } from "zod";
+import { z as z25 } from "zod";
 var UUID_REGEX = /^[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$/i;
-var BatchAssertionLinkSchema = z24.union([
-  z24.string().min(1),
+var BatchAssertionLinkSchema = z25.union([
+  z25.string().min(1),
   ScenarioAssertionLinkSchema
 ]);
-var BatchScenarioEntrySchema = z24.object({
-  name: z24.string().min(1, "name: Required"),
-  description: z24.string().optional().default(""),
-  triggerPrompt: z24.string().min(10, "triggerPrompt: Must be at least 10 characters"),
-  templateId: z24.string().nullish(),
-  tags: z24.array(z24.string()).optional(),
-  assertionLinks: z24.array(BatchAssertionLinkSchema).optional()
+var BatchScenarioEntrySchema = z25.object({
+  name: z25.string().min(1, "name: Required"),
+  description: z25.string().optional().default(""),
+  triggerPrompt: z25.string().min(10, "triggerPrompt: Must be at least 10 characters"),
+  templateId: z25.string().nullish(),
+  tags: z25.array(z25.string()).optional(),
+  assertionLinks: z25.array(BatchAssertionLinkSchema).optional()
+}).superRefine((data, ctx) => {
+  if (!data.assertionLinks) return;
+  const objectLinks = data.assertionLinks.filter(
+    (link) => typeof link !== "string"
+  );
+  if (objectLinks.length > 0) {
+    validateBuildPassedParamsInAssertionLinks(objectLinks, ctx);
+  }
 });
-var BatchImportPayloadSchema = z24.object({
-  scenarios: z24.array(BatchScenarioEntrySchema).min(1, "scenarios array must contain at least one entry").max(100, "Maximum 100 scenarios per upload")
+var BatchImportPayloadSchema = z25.object({
+  scenarios: z25.array(BatchScenarioEntrySchema).min(1, "scenarios array must contain at least one entry").max(100, "Maximum 100 scenarios per upload")
 });
 var BATCH_IMPORT_LIMITS = {
   MAX_SCENARIOS: 100,
@@ -846,29 +1136,29 @@ function normalizeBatchAssertionLink(link) {
   }
   return link;
 }
-var BatchResultItemSchema = z24.object({
-  index: z24.number(),
-  name: z24.string(),
-  status: z24.enum(["valid", "invalid"]),
-  id: z24.string().nullable().optional(),
-  errors: z24.array(z24.string()).optional()
-});
-var BatchSummarySchema = z24.object({
-  total: z24.number(),
-  valid: z24.number(),
-  invalid: z24.number(),
-  created: z24.number()
-});
-var BatchImportResponseSchema = z24.object({
+var BatchResultItemSchema = z25.object({
+  index: z25.number(),
+  name: z25.string(),
+  status: z25.enum(["valid", "invalid"]),
+  id: z25.string().nullable().optional(),
+  errors: z25.array(z25.string()).optional()
+});
+var BatchSummarySchema = z25.object({
+  total: z25.number(),
+  valid: z25.number(),
+  invalid: z25.number(),
+  created: z25.number()
+});
+var BatchImportResponseSchema = z25.object({
   summary: BatchSummarySchema,
-  results: z24.array(BatchResultItemSchema)
+  results: z25.array(BatchResultItemSchema)
 });
 
 // src/suite/test-suite.ts
-import { z as z25 } from "zod";
+import { z as z26 } from "zod";
 var TestSuiteSchema = TenantEntitySchema.extend({
   /** IDs of test scenarios in this suite */
-  scenarioIds: z25.array(z25.string())
+  scenarioIds: z26.array(z26.string())
 });
 var CreateTestSuiteInputSchema = TestSuiteSchema.omit({
   id: true,
@@ -879,21 +1169,21 @@ var CreateTestSuiteInputSchema = TestSuiteSchema.omit({
 var UpdateTestSuiteInputSchema = CreateTestSuiteInputSchema.partial();
 
 // src/evaluation/metrics.ts
-import { z as z26 } from "zod";
-var TokenUsageSchema = z26.object({
-  prompt: z26.number(),
-  completion: z26.number(),
-  total: z26.number()
-});
-var EvalMetricsSchema = z26.object({
-  totalAssertions: z26.number(),
-  passed: z26.number(),
-  failed: z26.number(),
-  skipped: z26.number(),
-  errors: z26.number(),
-  passRate: z26.number(),
-  avgDuration: z26.number(),
-  totalDuration: z26.number()
+import { z as z27 } from "zod";
+var TokenUsageSchema = z27.object({
+  prompt: z27.number(),
+  completion: z27.number(),
+  total: z27.number()
+});
+var EvalMetricsSchema = z27.object({
+  totalAssertions: z27.number(),
+  passed: z27.number(),
+  failed: z27.number(),
+  skipped: z27.number(),
+  errors: z27.number(),
+  passRate: z27.number(),
+  avgDuration: z27.number(),
+  totalDuration: z27.number()
 });
 var EvalStatus = /* @__PURE__ */ ((EvalStatus2) => {
   EvalStatus2["PENDING"] = "pending";
@@ -903,7 +1193,7 @@ var EvalStatus = /* @__PURE__ */ ((EvalStatus2) => {
   EvalStatus2["CANCELLED"] = "cancelled";
   return EvalStatus2;
 })(EvalStatus || {});
-var EvalStatusSchema = z26.enum(EvalStatus);
+var EvalStatusSchema = z27.enum(EvalStatus);
 var LLMStepType = /* @__PURE__ */ ((LLMStepType2) => {
   LLMStepType2["COMPLETION"] = "completion";
   LLMStepType2["TOOL_USE"] = "tool_use";
@@ -911,54 +1201,54 @@ var LLMStepType = /* @__PURE__ */ ((LLMStepType2) => {
   LLMStepType2["THINKING"] = "thinking";
   return LLMStepType2;
 })(LLMStepType || {});
-var LLMTraceStepSchema = z26.object({
-  id: z26.string(),
-  stepNumber: z26.number(),
-  type: z26.enum(LLMStepType),
-  model: z26.string(),
-  provider: z26.string(),
-  startedAt: z26.string(),
-  durationMs: z26.number(),
+var LLMTraceStepSchema = z27.object({
+  id: z27.string(),
+  stepNumber: z27.number(),
+  type: z27.enum(LLMStepType),
+  model: z27.string(),
+  provider: z27.string(),
+  startedAt: z27.string(),
+  durationMs: z27.number(),
   tokenUsage: TokenUsageSchema,
-  costUsd: z26.number(),
-  toolName: z26.string().optional(),
-  toolArguments: z26.string().optional(),
-  inputPreview: z26.string().optional(),
-  outputPreview: z26.string().optional(),
-  success: z26.boolean(),
-  error: z26.string().optional(),
-  turnIndex: z26.number().optional()
-});
-var LLMBreakdownStatsSchema = z26.object({
-  count: z26.number(),
-  durationMs: z26.number(),
-  tokens: z26.number(),
-  costUsd: z26.number()
-});
-var LLMTraceSummarySchema = z26.object({
-  totalSteps: z26.number(),
-  totalTurns: z26.number().optional(),
-  totalDurationMs: z26.number(),
+  costUsd: z27.number(),
+  toolName: z27.string().optional(),
+  toolArguments: z27.string().optional(),
+  inputPreview: z27.string().optional(),
+  outputPreview: z27.string().optional(),
+  success: z27.boolean(),
+  error: z27.string().optional(),
+  turnIndex: z27.number().optional()
+});
+var LLMBreakdownStatsSchema = z27.object({
+  count: z27.number(),
+  durationMs: z27.number(),
+  tokens: z27.number(),
+  costUsd: z27.number()
+});
+var LLMTraceSummarySchema = z27.object({
+  totalSteps: z27.number(),
+  totalTurns: z27.number().optional(),
+  totalDurationMs: z27.number(),
   totalTokens: TokenUsageSchema,
-  totalCostUsd: z26.number(),
-  stepTypeBreakdown: z26.record(z26.string(), LLMBreakdownStatsSchema).optional(),
-  modelBreakdown: z26.record(z26.string(), LLMBreakdownStatsSchema),
-  modelsUsed: z26.array(z26.string())
-});
-var LLMTraceSchema = z26.object({
-  id: z26.string(),
-  steps: z26.array(LLMTraceStepSchema),
+  totalCostUsd: z27.number(),
+  stepTypeBreakdown: z27.record(z27.string(), LLMBreakdownStatsSchema).optional(),
+  modelBreakdown: z27.record(z27.string(), LLMBreakdownStatsSchema),
+  modelsUsed: z27.array(z27.string())
+});
+var LLMTraceSchema = z27.object({
+  id: z27.string(),
+  steps: z27.array(LLMTraceStepSchema),
   summary: LLMTraceSummarySchema
 });
 
 // src/evaluation/eval-result.ts
-import { z as z30 } from "zod";
+import { z as z31 } from "zod";
 
 // src/evaluation/eval-run.ts
-import { z as z28 } from "zod";
+import { z as z29 } from "zod";
 
 // src/evaluation/live-trace.ts
-import { z as z27 } from "zod";
+import { z as z28 } from "zod";
 var LiveTraceEventType = /* @__PURE__ */ ((LiveTraceEventType2) => {
   LiveTraceEventType2["THINKING"] = "thinking";
   LiveTraceEventType2["TOOL_USE"] = "tool_use";
@@ -972,37 +1262,37 @@ var LiveTraceEventType = /* @__PURE__ */ ((LiveTraceEventType2) => {
   LiveTraceEventType2["USER"] = "user";
   return LiveTraceEventType2;
 })(LiveTraceEventType || {});
-var LiveTraceEventSchema = z27.object({
+var LiveTraceEventSchema = z28.object({
   /** The evaluation run ID */
-  evalRunId: z27.string(),
+  evalRunId: z28.string(),
   /** The scenario ID being executed */
-  scenarioId: z27.string(),
+  scenarioId: z28.string(),
   /** The scenario name for display */
-  scenarioName: z27.string(),
+  scenarioName: z28.string(),
   /** The target ID (skill, agent, etc.) */
-  targetId: z27.string(),
+  targetId: z28.string(),
   /** The target name for display */
-  targetName: z27.string(),
+  targetName: z28.string(),
   /** Step number in the current scenario execution */
-  stepNumber: z27.number(),
+  stepNumber: z28.number(),
   /** Type of trace event */
-  type: z27.enum(LiveTraceEventType),
+  type: z28.enum(LiveTraceEventType),
   /** Tool name if this is a tool_use event */
-  toolName: z27.string().optional(),
+  toolName: z28.string().optional(),
   /** Tool arguments preview (truncated JSON) */
-  toolArgs: z27.string().optional(),
+  toolArgs: z28.string().optional(),
   /** Output preview (truncated text) */
-  outputPreview: z27.string().optional(),
+  outputPreview: z28.string().optional(),
   /** File path for file operations */
-  filePath: z27.string().optional(),
+  filePath: z28.string().optional(),
   /** Elapsed time in milliseconds for progress events */
-  elapsedMs: z27.number().optional(),
+  elapsedMs: z28.number().optional(),
   /** Thinking/reasoning text from Claude */
-  thinking: z27.string().optional(),
+  thinking: z28.string().optional(),
   /** Timestamp when this event occurred */
-  timestamp: z27.string(),
+  timestamp: z28.string(),
   /** Whether this is the final event for this scenario */
-  isComplete: z27.boolean()
+  isComplete: z28.boolean()
 });
 var TRACE_EVENT_PREFIX = "TRACE_EVENT:";
 function parseTraceEventLine(line) {
@@ -1031,40 +1321,40 @@ var TriggerType = /* @__PURE__ */ ((TriggerType2) => {
   TriggerType2["SCHEDULED"] = "SCHEDULED";
   return TriggerType2;
 })(TriggerType || {});
-var TriggerMetadataSchema = z28.object({
-  version: z28.string().optional(),
-  resourceUpdated: z28.array(z28.string()).optional(),
-  scheduleId: z28.string().optional()
+var TriggerMetadataSchema = z29.object({
+  version: z29.string().optional(),
+  resourceUpdated: z29.array(z29.string()).optional(),
+  scheduleId: z29.string().optional()
 });
-var TriggerSchema = z28.object({
-  id: z28.string(),
+var TriggerSchema = z29.object({
+  id: z29.string(),
   metadata: TriggerMetadataSchema.optional(),
-  type: z28.nativeEnum(TriggerType)
+  type: z29.nativeEnum(TriggerType)
 });
-var DiffLineTypeSchema = z28.enum(["added", "removed", "unchanged"]);
-var DiffLineSchema = z28.object({
+var DiffLineTypeSchema = z29.enum(["added", "removed", "unchanged"]);
+var DiffLineSchema = z29.object({
   type: DiffLineTypeSchema,
-  content: z28.string(),
-  lineNumber: z28.number()
-});
-var DiffContentSchema = z28.object({
-  path: z28.string(),
-  expected: z28.string(),
-  actual: z28.string(),
-  diffLines: z28.array(DiffLineSchema),
-  renamedFrom: z28.string().optional(),
+  content: z29.string(),
+  lineNumber: z29.number()
+});
+var DiffContentSchema = z29.object({
+  path: z29.string(),
+  expected: z29.string(),
+  actual: z29.string(),
+  diffLines: z29.array(DiffLineSchema),
+  renamedFrom: z29.string().optional(),
   /** Whether this file is an infrastructure/config file (e.g. .claude/settings.json, .mcp.json) */
-  isInfrastructure: z28.boolean().optional()
+  isInfrastructure: z29.boolean().optional()
 });
-var CommandExecutionSchema = z28.object({
-  command: z28.string(),
-  exitCode: z28.number(),
-  output: z28.string().optional(),
-  duration: z28.number()
+var CommandExecutionSchema = z29.object({
+  command: z29.string(),
+  exitCode: z29.number(),
+  output: z29.string().optional(),
+  duration: z29.number()
 });
-var FileModificationSchema = z28.object({
-  path: z28.string(),
-  action: z28.enum(["created", "modified", "deleted"])
+var FileModificationSchema = z29.object({
+  path: z29.string(),
+  action: z29.enum(["created", "modified", "deleted"])
 });
 var TemplateFileStatus = /* @__PURE__ */ ((TemplateFileStatus2) => {
   TemplateFileStatus2["NEW"] = "new";
@@ -1072,62 +1362,62 @@ var TemplateFileStatus = /* @__PURE__ */ ((TemplateFileStatus2) => {
   TemplateFileStatus2["UNCHANGED"] = "unchanged";
   return TemplateFileStatus2;
 })(TemplateFileStatus || {});
-var TemplateFileSchema = z28.object({
+var TemplateFileSchema = z29.object({
   /** Relative path within the template */
-  path: z28.string(),
+  path: z29.string(),
   /** Full file content after execution */
-  content: z28.string(),
+  content: z29.string(),
   /** File status (new, modified, unchanged) */
-  status: z28.enum(["new", "modified", "unchanged"]),
+  status: z29.enum(["new", "modified", "unchanged"]),
   /** Whether this file is an infrastructure/config file (e.g. .claude/settings.json, .mcp.json) */
-  isInfrastructure: z28.boolean().optional()
+  isInfrastructure: z29.boolean().optional()
 });
-var ApiCallSchema = z28.object({
-  endpoint: z28.string(),
-  tokensUsed: z28.number(),
-  duration: z28.number()
+var ApiCallSchema = z29.object({
+  endpoint: z29.string(),
+  tokensUsed: z29.number(),
+  duration: z29.number()
 });
-var ExecutionTraceSchema = z28.object({
-  commands: z28.array(CommandExecutionSchema),
-  filesModified: z28.array(FileModificationSchema),
-  apiCalls: z28.array(ApiCallSchema),
-  totalDuration: z28.number()
+var ExecutionTraceSchema = z29.object({
+  commands: z29.array(CommandExecutionSchema),
+  filesModified: z29.array(FileModificationSchema),
+  apiCalls: z29.array(ApiCallSchema),
+  totalDuration: z29.number()
 });
-var RunAnalysisFindingSchema = z28.object({
-  category: z28.enum([
+var RunAnalysisFindingSchema = z29.object({
+  category: z29.enum([
     "failure_pattern",
     "cost_waste",
     "flakiness",
     "inefficiency",
     "positive"
   ]),
-  severity: z28.enum(["high", "medium", "low"]),
-  description: z28.string(),
-  affectedScenarios: z28.array(z28.string()),
-  recommendation: z28.string().optional()
+  severity: z29.enum(["high", "medium", "low"]),
+  description: z29.string(),
+  affectedScenarios: z29.array(z29.string()),
+  recommendation: z29.string().optional()
 });
-var RunAnalysisSchema = z28.object({
-  generatedAt: z28.string(),
-  summary: z28.string(),
-  findings: z28.array(RunAnalysisFindingSchema)
+var RunAnalysisSchema = z29.object({
+  generatedAt: z29.string(),
+  summary: z29.string(),
+  findings: z29.array(RunAnalysisFindingSchema)
 });
 var EvalRunSchema = TenantEntitySchema.extend({
   /** Agent ID for this run */
-  agentId: z28.string().optional(),
+  agentId: z29.string().optional(),
   /** Preset ID that originated this run (optional) */
-  presetId: z28.string().optional(),
+  presetId: z29.string().optional(),
   /** Skill IDs for this run */
-  skillIds: z28.array(z28.string()).optional(),
+  skillIds: z29.array(z29.string()).optional(),
   /** Map of skillId to skillVersionId for this run */
-  skillVersions: z28.record(z28.string(), z28.string()).optional(),
+  skillVersions: z29.record(z29.string(), z29.string()).optional(),
   /** Scenario IDs to run (always present — resolved server-side from tags when needed) */
-  scenarioIds: z28.array(z28.string()),
+  scenarioIds: z29.array(z29.string()),
   /** Current status */
   status: EvalStatusSchema,
   /** Progress percentage (0-100) */
-  progress: z28.number(),
+  progress: z29.number(),
   /** Results for each scenario/target combination (lazy to break eval-result ↔ eval-run cycle) */
-  results: z28.array(z28.lazy(() => EvalRunResultSchema)),
+  results: z29.array(z29.lazy(() => EvalRunResultSchema)),
   /** Aggregated metrics across all results */
   aggregateMetrics: EvalMetricsSchema,
   /** Aggregated LLM trace summary */
@@ -1135,41 +1425,41 @@ var EvalRunSchema = TenantEntitySchema.extend({
   /** What triggered this run */
   trigger: TriggerSchema.optional(),
   /** When the run started (set when evaluation is triggered) */
-  startedAt: z28.string().optional(),
+  startedAt: z29.string().optional(),
   /** When the run completed */
-  completedAt: z28.string().optional(),
+  completedAt: z29.string().optional(),
   /** Live trace events captured during execution (for playback on results page) */
-  liveTraceEvents: z28.array(LiveTraceEventSchema).optional(),
+  liveTraceEvents: z29.array(LiveTraceEventSchema).optional(),
   /** Remote job ID for tracking execution in Dev Machines */
-  jobId: z28.string().optional(),
+  jobId: z29.string().optional(),
   /** Remote job status from the Dev Machine API (PENDING, RUNNING, COMPLETED, FAILED, CANCELLED) */
-  jobStatus: z28.string().optional(),
+  jobStatus: z29.string().optional(),
   /** Remote job error message if the job failed */
-  jobError: z28.string().optional(),
+  jobError: z29.string().optional(),
   /** Timestamp of the last job status check */
-  jobStatusCheckedAt: z28.string().optional(),
+  jobStatusCheckedAt: z29.string().optional(),
   /** MCP server IDs to enable for this run (optional) */
-  mcpIds: z28.array(z28.string()).optional(),
+  mcpIds: z29.array(z29.string()).optional(),
   /** Sub-agent IDs to enable for this run (optional) */
-  subAgentIds: z28.array(z28.string()).optional(),
+  subAgentIds: z29.array(z29.string()).optional(),
   /** Rule IDs to enable for this run (optional) */
-  ruleIds: z28.array(z28.string()).optional(),
+  ruleIds: z29.array(z29.string()).optional(),
   /** Tags used to select scenarios for this run (for traceability) */
-  tags: z28.array(z28.string()).optional(),
+  tags: z29.array(z29.string()).optional(),
   /** How many times each scenario is executed within this eval run. Default: 1. Max: 20. */
-  runsPerScenario: z28.number().int().min(1).max(20).optional(),
+  runsPerScenario: z29.number().int().min(1).max(20).optional(),
   /** Snapshot of agent configuration captured at run creation time */
-  agentSnapshot: z28.object({
-    name: z28.string().optional(),
+  agentSnapshot: z29.object({
+    name: z29.string().optional(),
     agentType: AgentTypeSchema.optional(),
     runCommand: AgentRunCommandSchema.optional(),
-    systemPrompt: z28.string().nullable().optional(),
+    systemPrompt: z29.string().nullable().optional(),
     modelConfig: ModelConfigSchema.optional()
   }).optional(),
   /** UUID linking all runs in a comparison group */
-  comparisonGroupId: z28.string().optional(),
+  comparisonGroupId: z29.string().optional(),
   /** Human-readable label for this variant (e.g., "MCP: Wix Stores") */
-  comparisonLabel: z28.string().optional(),
+  comparisonLabel: z29.string().optional(),
   /** LLM-generated analysis of the completed run */
   runAnalysis: RunAnalysisSchema.optional()
 });
@@ -1187,60 +1477,60 @@ var CreateEvalRunInputSchema = EvalRunSchema.omit({
   agentSnapshot: true
 }).extend({
   /** Optional on input — backend resolves from tags when not provided */
-  scenarioIds: z28.array(z28.string()).optional()
+  scenarioIds: z29.array(z29.string()).optional()
 }).refine(
   (data) => data.scenarioIds && data.scenarioIds.length > 0 || data.tags && data.tags.length > 0,
   { message: "Either scenarioIds or tags must be provided" }
 );
-var EvaluationProgressSchema = z28.object({
-  runId: z28.string(),
-  targetId: z28.string(),
-  totalScenarios: z28.number(),
-  completedScenarios: z28.number(),
-  scenarioProgress: z28.array(
-    z28.object({
-      scenarioId: z28.string(),
-      currentStep: z28.string(),
-      error: z28.string().optional()
+var EvaluationProgressSchema = z29.object({
+  runId: z29.string(),
+  targetId: z29.string(),
+  totalScenarios: z29.number(),
+  completedScenarios: z29.number(),
+  scenarioProgress: z29.array(
+    z29.object({
+      scenarioId: z29.string(),
+      currentStep: z29.string(),
+      error: z29.string().optional()
     })
   ),
-  createdAt: z28.number()
-});
-var EvaluationLogSchema = z28.object({
-  runId: z28.string(),
-  scenarioId: z28.string(),
-  log: z28.object({
-    level: z28.enum(["info", "error", "debug"]),
-    message: z28.string().optional(),
-    args: z28.array(z28.any()).optional(),
-    error: z28.string().optional()
+  createdAt: z29.number()
+});
+var EvaluationLogSchema = z29.object({
+  runId: z29.string(),
+  scenarioId: z29.string(),
+  log: z29.object({
+    level: z29.enum(["info", "error", "debug"]),
+    message: z29.string().optional(),
+    args: z29.array(z29.any()).optional(),
+    error: z29.string().optional()
   })
 });
 var LLM_TIMEOUT = 12e4;
 
 // src/evaluation/conversation.ts
-import { z as z29 } from "zod";
-var TextBlockSchema = z29.object({
-  type: z29.literal("text"),
-  text: z29.string()
-});
-var ThinkingBlockSchema = z29.object({
-  type: z29.literal("thinking"),
-  thinking: z29.string()
-});
-var ToolUseBlockSchema = z29.object({
-  type: z29.literal("tool_use"),
-  toolName: z29.string(),
-  toolId: z29.string(),
-  input: z29.unknown()
-});
-var ToolResultBlockSchema = z29.object({
-  type: z29.literal("tool_result"),
-  toolUseId: z29.string(),
-  content: z29.string(),
-  isError: z29.boolean().optional()
-});
-var ConversationBlockSchema = z29.discriminatedUnion("type", [
+import { z as z30 } from "zod";
+var TextBlockSchema = z30.object({
+  type: z30.literal("text"),
+  text: z30.string()
+});
+var ThinkingBlockSchema = z30.object({
+  type: z30.literal("thinking"),
+  thinking: z30.string()
+});
+var ToolUseBlockSchema = z30.object({
+  type: z30.literal("tool_use"),
+  toolName: z30.string(),
+  toolId: z30.string(),
+  input: z30.unknown()
+});
+var ToolResultBlockSchema = z30.object({
+  type: z30.literal("tool_result"),
+  toolUseId: z30.string(),
+  content: z30.string(),
+  isError: z30.boolean().optional()
+});
+var ConversationBlockSchema = z30.discriminatedUnion("type", [
   TextBlockSchema,
   ThinkingBlockSchema,
   ToolUseBlockSchema,
@@ -1251,18 +1541,18 @@ var ConversationMessageRoles = [
   "user",
   "system"
 ];
-var ConversationMessageSchema = z29.object({
-  role: z29.enum(ConversationMessageRoles),
-  content: z29.array(ConversationBlockSchema),
-  timestamp: z29.string()
+var ConversationMessageSchema = z30.object({
+  role: z30.enum(ConversationMessageRoles),
+  content: z30.array(ConversationBlockSchema),
+  timestamp: z30.string()
 });
-var ScenarioConversationSchema = z29.object({
-  id: z29.string(),
-  projectId: z29.string(),
-  evalRunId: z29.string(),
-  resultId: z29.string(),
-  messages: z29.array(ConversationMessageSchema),
-  createdAt: z29.string()
+var ScenarioConversationSchema = z30.object({
+  id: z30.string(),
+  projectId: z30.string(),
+  evalRunId: z30.string(),
+  resultId: z30.string(),
+  messages: z30.array(ConversationMessageSchema),
+  createdAt: z30.string()
 });
 
 // src/evaluation/eval-result.ts
@@ -1273,98 +1563,98 @@ var AssertionResultStatus = /* @__PURE__ */ ((AssertionResultStatus2) => {
   AssertionResultStatus2["ERROR"] = "error";
   return AssertionResultStatus2;
 })(AssertionResultStatus || {});
-var AssertionResultSchema = z30.object({
-  id: z30.string(),
-  assertionId: z30.string(),
-  assertionType: z30.string(),
-  assertionName: z30.string(),
-  status: z30.enum(AssertionResultStatus),
-  message: z30.string().optional(),
-  expected: z30.string().optional(),
-  actual: z30.string().optional(),
-  duration: z30.number().optional(),
-  details: z30.record(z30.string(), z30.unknown()).optional(),
-  llmTraceSteps: z30.array(LLMTraceStepSchema).optional()
-});
-var EvalRunResultSchema = z30.object({
-  id: z30.string(),
-  targetId: z30.string(),
-  targetName: z30.string().optional(),
+var AssertionResultSchema = z31.object({
+  id: z31.string(),
+  assertionId: z31.string(),
+  assertionType: z31.string(),
+  assertionName: z31.string(),
+  status: z31.enum(AssertionResultStatus),
+  message: z31.string().optional(),
+  expected: z31.string().optional(),
+  actual: z31.string().optional(),
+  duration: z31.number().optional(),
+  details: z31.record(z31.string(), z31.unknown()).optional(),
+  llmTraceSteps: z31.array(LLMTraceStepSchema).optional()
+});
+var EvalRunResultSchema = z31.object({
+  id: z31.string(),
+  targetId: z31.string(),
+  targetName: z31.string().optional(),
   /** SkillVersion ID used for this evaluation (for version tracking) */
-  skillVersionId: z30.string().optional(),
+  skillVersionId: z31.string().optional(),
   /** SkillVersion semver string (e.g., "1.0.0", "1.2.3") for display */
-  skillVersion: z30.string().optional(),
-  scenarioId: z30.string(),
-  scenarioName: z30.string(),
+  skillVersion: z31.string().optional(),
+  scenarioId: z31.string(),
+  scenarioName: z31.string(),
   /** Snapshot of the trigger prompt used during the run (prevents stale display after edits) */
-  triggerPrompt: z30.string().optional(),
+  triggerPrompt: z31.string().optional(),
   modelConfig: ModelConfigSchema.optional(),
-  assertionResults: z30.array(AssertionResultSchema),
+  assertionResults: z31.array(AssertionResultSchema),
   metrics: EvalMetricsSchema.optional(),
-  passed: z30.number(),
-  failed: z30.number(),
-  passRate: z30.number(),
-  duration: z30.number(),
-  outputText: z30.string().optional(),
-  files: z30.array(ExpectedFileSchema).optional(),
-  fileDiffs: z30.array(DiffContentSchema).optional(),
+  passed: z31.number(),
+  failed: z31.number(),
+  passRate: z31.number(),
+  duration: z31.number(),
+  outputText: z31.string().optional(),
+  files: z31.array(ExpectedFileSchema).optional(),
+  fileDiffs: z31.array(DiffContentSchema).optional(),
   /** Full template files after execution with status indicators */
-  templateFiles: z30.array(TemplateFileSchema).optional(),
-  startedAt: z30.string().optional(),
-  completedAt: z30.string().optional(),
+  templateFiles: z31.array(TemplateFileSchema).optional(),
+  startedAt: z31.string().optional(),
+  completedAt: z31.string().optional(),
   llmTrace: LLMTraceSchema.optional(),
   /** Full conversation messages (only present in transit; stripped before DB storage) */
-  conversation: z30.array(ConversationMessageSchema).optional(),
+  conversation: z31.array(ConversationMessageSchema).optional(),
   /** 0-based iteration index when a scenario is run multiple times within a single eval run */
-  iterationIndex: z30.number().int().min(0).optional()
-});
-var PromptResultSchema = z30.object({
-  text: z30.string(),
-  files: z30.array(z30.unknown()).optional(),
-  finishReason: z30.string().optional(),
-  reasoning: z30.string().optional(),
-  reasoningDetails: z30.unknown().optional(),
-  toolCalls: z30.array(z30.unknown()).optional(),
-  toolResults: z30.array(z30.unknown()).optional(),
-  warnings: z30.array(z30.unknown()).optional(),
-  sources: z30.array(z30.unknown()).optional(),
-  steps: z30.array(z30.unknown()),
-  generationTimeMs: z30.number(),
-  prompt: z30.string(),
-  systemPrompt: z30.string(),
-  usage: z30.object({
-    totalTokens: z30.number().optional(),
-    totalMicrocentsSpent: z30.number().optional()
+  iterationIndex: z31.number().int().min(0).optional()
+});
+var PromptResultSchema = z31.object({
+  text: z31.string(),
+  files: z31.array(z31.unknown()).optional(),
+  finishReason: z31.string().optional(),
+  reasoning: z31.string().optional(),
+  reasoningDetails: z31.unknown().optional(),
+  toolCalls: z31.array(z31.unknown()).optional(),
+  toolResults: z31.array(z31.unknown()).optional(),
+  warnings: z31.array(z31.unknown()).optional(),
+  sources: z31.array(z31.unknown()).optional(),
+  steps: z31.array(z31.unknown()),
+  generationTimeMs: z31.number(),
+  prompt: z31.string(),
+  systemPrompt: z31.string(),
+  usage: z31.object({
+    totalTokens: z31.number().optional(),
+    totalMicrocentsSpent: z31.number().optional()
   })
 });
-var EvaluationResultSchema = z30.object({
-  id: z30.string(),
-  runId: z30.string(),
-  timestamp: z30.number(),
+var EvaluationResultSchema = z31.object({
+  id: z31.string(),
+  runId: z31.string(),
+  timestamp: z31.number(),
   promptResult: PromptResultSchema,
-  testResults: z30.array(z30.unknown()),
-  tags: z30.array(z30.string()).optional(),
-  feedback: z30.string().optional(),
-  score: z30.number(),
-  suiteId: z30.string().optional()
-});
-var LeanEvaluationResultSchema = z30.object({
-  id: z30.string(),
-  runId: z30.string(),
-  timestamp: z30.number(),
-  tags: z30.array(z30.string()).optional(),
-  scenarioId: z30.string(),
-  scenarioVersion: z30.number().optional(),
-  targetId: z30.string(),
-  targetVersion: z30.number().optional(),
-  suiteId: z30.string().optional(),
-  score: z30.number(),
-  time: z30.number().optional(),
-  microcentsSpent: z30.number().optional()
+  testResults: z31.array(z31.unknown()),
+  tags: z31.array(z31.string()).optional(),
+  feedback: z31.string().optional(),
+  score: z31.number(),
+  suiteId: z31.string().optional()
+});
+var LeanEvaluationResultSchema = z31.object({
+  id: z31.string(),
+  runId: z31.string(),
+  timestamp: z31.number(),
+  tags: z31.array(z31.string()).optional(),
+  scenarioId: z31.string(),
+  scenarioVersion: z31.number().optional(),
+  targetId: z31.string(),
+  targetVersion: z31.number().optional(),
+  suiteId: z31.string().optional(),
+  score: z31.number(),
+  time: z31.number().optional(),
+  microcentsSpent: z31.number().optional()
 });
 
 // src/evaluation/eval-run-folder.ts
-import { z as z31 } from "zod";
+import { z as z32 } from "zod";
 var EvalRunFolderSchema = TenantEntitySchema.extend({});
 var CreateEvalRunFolderInputSchema = EvalRunFolderSchema.omit({
   id: true,
@@ -1378,26 +1668,26 @@ var UpdateEvalRunFolderInputSchema = EvalRunFolderSchema.omit({
   updatedAt: true,
   deleted: true
 }).partial();
-var EvalRunFolderMembershipSchema = z31.object({
-  folderId: z31.string(),
-  evalRunId: z31.string(),
-  projectId: z31.string(),
-  createdAt: z31.string()
+var EvalRunFolderMembershipSchema = z32.object({
+  folderId: z32.string(),
+  evalRunId: z32.string(),
+  projectId: z32.string(),
+  createdAt: z32.string()
 });
 
 // src/project/project.ts
-import { z as z32 } from "zod";
+import { z as z33 } from "zod";
 var ProjectSchema = BaseEntitySchema.extend({
-  appId: z32.string().optional().describe("The ID of the app in Dev Center"),
-  scenarioTags: z32.array(z32.string()).optional().describe("Project-level tag vocabulary for scenarios"),
+  appId: z33.string().optional().describe("The ID of the app in Dev Center"),
+  scenarioTags: z33.array(z33.string()).optional().describe("Project-level tag vocabulary for scenarios"),
   /** Per-project Wix auth token (write-only — never returned in GET responses). null = clear. */
-  wixAuthToken: z32.string().nullable().optional().describe("Wix auth token for CLI/MCP authentication (encrypted at rest)"),
+  wixAuthToken: z33.string().nullable().optional().describe("Wix auth token for CLI/MCP authentication (encrypted at rest)"),
   /** Per-project Base44 auth file content (write-only — never returned in GET responses). null = clear. */
-  base44AuthFile: z32.string().nullable().optional().describe("Base64-encoded Base44 auth file content (encrypted at rest)"),
+  base44AuthFile: z33.string().nullable().optional().describe("Base64-encoded Base44 auth file content (encrypted at rest)"),
   /** Resolved at runtime from the encrypted Wix auth token */
-  wixAuthEmail: z32.string().optional().describe("Email associated with the Wix auth token (resolved at runtime)"),
+  wixAuthEmail: z33.string().optional().describe("Email associated with the Wix auth token (resolved at runtime)"),
   /** Resolved at runtime from the encrypted Base44 auth file */
-  base44AuthEmail: z32.string().optional().describe("Email from the Base44 auth file (resolved at runtime)")
+  base44AuthEmail: z33.string().optional().describe("Email from the Base44 auth file (resolved at runtime)")
 });
 var CreateProjectInputSchema = ProjectSchema.omit({
   id: true,
@@ -1406,6 +1696,10 @@ var CreateProjectInputSchema = ProjectSchema.omit({
   deleted: true,
   wixAuthEmail: true,
   base44AuthEmail: true
+}).extend({
+  appId: z33.string().describe(
+    "Required: The ID of the app in Dev Center for credential scoping"
+  )
 });
 var UpdateProjectInputSchema = CreateProjectInputSchema.partial();
 
@@ -1423,7 +1717,7 @@ var CreateTemplateInputSchema = TemplateSchema.omit({
 var UpdateTemplateInputSchema = CreateTemplateInputSchema.partial();
 
 // src/schedule/eval-schedule.ts
-import { z as z33 } from "zod";
+import { z as z34 } from "zod";
 var FrequencyType = /* @__PURE__ */ ((FrequencyType2) => {
   FrequencyType2["DAILY"] = "daily";
   FrequencyType2["WEEKDAY"] = "weekday";
@@ -1433,29 +1727,29 @@ var FrequencyType = /* @__PURE__ */ ((FrequencyType2) => {
 })(FrequencyType || {});
 var EvalScheduleSchema = TenantEntitySchema.extend({
   /** Whether the schedule is active */
-  enabled: z33.boolean(),
+  enabled: z34.boolean(),
   /** Test suite to run */
-  suiteId: z33.string(),
+  suiteId: z34.string(),
   /** Preset that provides agent + entities for this schedule */
-  presetId: z33.string(),
+  presetId: z34.string(),
   /** How often to run */
-  frequencyType: z33.nativeEnum(FrequencyType),
+  frequencyType: z34.nativeEnum(FrequencyType),
   /** Time of day in 24h format (HH:MM), hours 00-23, minutes 00-59 */
-  timeOfDay: z33.string().regex(/^([01]\d|2[0-3]):[0-5]\d$/),
+  timeOfDay: z34.string().regex(/^([01]\d|2[0-3]):[0-5]\d$/),
   /** Day of week (0=Sun, 6=Sat) for weekly schedules */
-  dayOfWeek: z33.number().min(0).max(6).optional(),
+  dayOfWeek: z34.number().min(0).max(6).optional(),
   /** Day of month (1-31) for monthly schedules */
-  dayOfMonth: z33.number().min(1).max(31).optional(),
+  dayOfMonth: z34.number().min(1).max(31).optional(),
   /** IANA timezone (e.g., 'America/New_York') */
-  timezone: z33.string(),
+  timezone: z34.string(),
   /** ID of the last eval run created by this schedule */
-  lastRunId: z33.string().optional(),
+  lastRunId: z34.string().optional(),
   /** Denormalized status of the last run */
-  lastRunStatus: z33.string().optional(),
+  lastRunStatus: z34.string().optional(),
   /** ISO timestamp of the last run */
-  lastRunAt: z33.string().optional(),
+  lastRunAt: z34.string().optional(),
   /** Next scheduled run time in UTC (pre-computed for efficient querying, set by backend) */
-  nextRunAt: z33.string().optional()
+  nextRunAt: z34.string().optional()
 });
 function isValidTimezone(tz) {
   try {
@@ -1468,14 +1762,14 @@ function isValidTimezone(tz) {
 function validateScheduleFields(data, ctx, options) {
   if (data.frequencyType === "weekly" /* WEEKLY */ && data.dayOfWeek == null) {
     ctx.addIssue({
-      code: z33.ZodIssueCode.custom,
+      code: z34.ZodIssueCode.custom,
       message: "dayOfWeek is required for weekly schedules",
       path: ["dayOfWeek"]
     });
   }
   if (data.frequencyType === "monthly" /* MONTHLY */ && data.dayOfMonth == null) {
     ctx.addIssue({
-      code: z33.ZodIssueCode.custom,
+      code: z34.ZodIssueCode.custom,
       message: "dayOfMonth is required for monthly schedules",
       path: ["dayOfMonth"]
     });
@@ -1483,7 +1777,7 @@ function validateScheduleFields(data, ctx, options) {
   const shouldValidateTz = options.partial ? data.timezone !== void 0 : true;
   if (shouldValidateTz && !isValidTimezone(data.timezone)) {
     ctx.addIssue({
-      code: z33.ZodIssueCode.custom,
+      code: z34.ZodIssueCode.custom,
       message: "Invalid IANA timezone",
       path: ["timezone"]
     });
@@ -1506,228 +1800,9 @@ var CreateEvalScheduleInputSchema = BaseCreateScheduleSchema.superRefine((data,
 var UpdateEvalScheduleInputSchema = BaseCreateScheduleSchema.partial().superRefine((data, ctx) => {
   validateScheduleFields(data, ctx, { partial: true });
 });
-
-// src/assertion/system-assertions.ts
-var SYSTEM_ASSERTION_IDS = {
-  SKILL_WAS_CALLED: "system:skill_was_called",
-  TOOL_CALLED_WITH_PARAM: "system:tool_called_with_param",
-  BUILD_PASSED: "system:build_passed",
-  TIME_LIMIT: "system:time_limit",
-  COST: "system:cost",
-  LLM_JUDGE: "system:llm_judge",
-  API_CALL: "system:api_call"
-};
-function isSystemAssertionId(id) {
-  return id.startsWith("system:");
-}
-var SYSTEM_ASSERTIONS = {
-  [SYSTEM_ASSERTION_IDS.SKILL_WAS_CALLED]: {
-    id: SYSTEM_ASSERTION_IDS.SKILL_WAS_CALLED,
-    name: "Skill Was Called",
-    description: "Check that one or more skills were invoked during the agent run",
-    type: "skill_was_called",
-    parameters: [
-      {
-        name: "skillNames",
-        label: "Skills",
-        type: "string",
-        required: true
-      },
-      {
-        name: "negate",
-        label: "Negate (NOT operator)",
-        type: "boolean",
-        required: false,
-        defaultValue: false
-      }
-    ]
-  },
-  [SYSTEM_ASSERTION_IDS.TOOL_CALLED_WITH_PARAM]: {
-    id: SYSTEM_ASSERTION_IDS.TOOL_CALLED_WITH_PARAM,
-    name: "Tool Called With Param",
-    description: "Check that a tool was called with expected parameters (tool name is substring matched)",
-    type: "tool_called_with_param",
-    parameters: [
-      {
-        name: "toolName",
-        label: "Tool Name",
-        type: "string",
-        required: true
-      },
-      {
-        name: "expectedParams",
-        label: "Expected Parameters (JSON, substring match)",
-        type: "string",
-        required: false
-      },
-      {
-        name: "requireSuccess",
-        label: "Require Successful Call",
-        type: "boolean",
-        required: false,
-        defaultValue: false,
-        advanced: true
-      },
-      {
-        name: "negate",
-        label: "Negate (NOT operator)",
-        type: "boolean",
-        required: false,
-        defaultValue: false
-      }
-    ]
-  },
-  [SYSTEM_ASSERTION_IDS.BUILD_PASSED]: {
-    id: SYSTEM_ASSERTION_IDS.BUILD_PASSED,
-    name: "Build Passed",
-    description: "Run a build command and verify it exits with expected code",
-    type: "build_passed",
-    parameters: [
-      {
-        name: "command",
-        label: "Build Command",
-        type: "string",
-        required: false,
-        defaultValue: "yarn build"
-      },
-      {
-        name: "expectedExitCode",
-        label: "Expected Exit Code",
-        type: "number",
-        required: false,
-        defaultValue: 0
-      },
-      {
-        name: "maxBuildTime",
-        label: "Max Build Time (ms)",
-        type: "number",
-        required: false,
-        advanced: true
-      },
-      {
-        name: "maxMemory",
-        label: "Max Memory (MB)",
-        type: "number",
-        required: false,
-        advanced: true
-      }
-    ]
-  },
-  [SYSTEM_ASSERTION_IDS.TIME_LIMIT]: {
-    id: SYSTEM_ASSERTION_IDS.TIME_LIMIT,
-    name: "Time Limit",
-    description: "Check that the scenario completed within a maximum duration",
-    type: "time_limit",
-    parameters: [
-      {
-        name: "maxDurationMs",
-        label: "Max Duration (ms)",
-        type: "number",
-        required: true,
-        defaultValue: 3e5
-      }
-    ]
-  },
-  [SYSTEM_ASSERTION_IDS.COST]: {
-    id: SYSTEM_ASSERTION_IDS.COST,
-    name: "Cost",
-    description: "Check that the scenario LLM execution cost stays within a USD threshold",
-    type: "cost",
-    parameters: [
-      {
-        name: "maxCostUsd",
-        label: "Max Cost (USD)",
-        type: "number",
-        required: true,
-        defaultValue: 1
-      }
-    ]
-  },
-  [SYSTEM_ASSERTION_IDS.LLM_JUDGE]: {
-    id: SYSTEM_ASSERTION_IDS.LLM_JUDGE,
-    name: "LLM Judge",
-    description: "LLM evaluates the output and assigns a score (0-10)",
-    type: "llm_judge",
-    parameters: [
-      {
-        name: "prompt",
-        label: "Judge Prompt",
-        type: "string",
-        required: true,
-        defaultValue: "Verify the output meets the acceptance criteria."
-      },
-      {
-        name: "minScore",
-        label: "Minimum Score (0-10)",
-        type: "number",
-        required: false,
-        defaultValue: 7
-      },
-      {
-        name: "model",
-        label: "Model",
-        type: "string",
-        required: false
-      }
-    ]
-  },
-  [SYSTEM_ASSERTION_IDS.API_CALL]: {
-    id: SYSTEM_ASSERTION_IDS.API_CALL,
-    name: "API Call",
-    description: "Call an API endpoint and verify the response contains expected data",
-    type: "api_call",
-    parameters: [
-      {
-        name: "url",
-        label: "URL",
-        type: "string",
-        required: true
-      },
-      {
-        name: "method",
-        label: "HTTP Method",
-        type: "string",
-        required: false,
-        defaultValue: "GET"
-      },
-      {
-        name: "requestBody",
-        label: "Request Body (JSON)",
-        type: "string",
-        required: false
-      },
-      {
-        name: "expectedResponse",
-        label: "Expected Response (JSON)",
-        type: "string",
-        required: true
-      },
-      {
-        name: "requestHeaders",
-        label: "Headers (JSON)",
-        type: "string",
-        required: false,
-        advanced: true
-      },
-      {
-        name: "timeoutMs",
-        label: "Timeout (ms)",
-        type: "number",
-        required: false,
-        defaultValue: 3e4,
-        advanced: true
-      }
-    ]
-  }
-};
-function getSystemAssertions() {
-  return Object.values(SYSTEM_ASSERTIONS);
-}
-function getSystemAssertion(id) {
-  return SYSTEM_ASSERTIONS[id];
-}
 export {
   AGENT_TYPE_LABELS,
+  ALLOWED_BUILD_COMMANDS,
   ALL_AVAILABLE_MODEL_IDS,
   AVAILABLE_CLAUDE_MODEL_IDS,
   AVAILABLE_OPENAI_MODEL_IDS,
@@ -1761,6 +1836,7 @@ export {
   BatchSummarySchema,
   BuildCheckTestSchema,
   BuildPassedAssertionSchema,
+  BuildPassedCommandStringSchema,
   BuildPassedConfigSchema,
   BulkImportResultItemSchema,
   BulkImportResultSchema,
@@ -1788,6 +1864,7 @@ export {
   CreateTemplateInputSchema,
   CreateTestScenarioInputSchema,
   CreateTestSuiteInputSchema,
+  DEFAULT_BUILD_PASSED_COMMAND,
   DEFAULT_EVALUATOR_SYSTEM_PROMPT,
   DEFAULT_JUDGE_MODEL,
   DiffContentSchema,
@@ -1905,11 +1982,14 @@ export {
   formatTraceEventLine,
   getSystemAssertion,
   getSystemAssertions,
+  isAllowedBuildCommandString,
   isSystemAssertionId,
   isValidSkillFolderName,
   normalizeBatchAssertionLink,
   normalizeModelId,
+  parseBuildCommandToArgv,
   parseTraceEventLine,
-  validateAssertionConfig
+  validateAssertionConfig,
+  validateBuildPassedParamsInAssertionLinks
 };
 //# sourceMappingURL=index.mjs.map