@wix/evalforge-types 0.72.0 → 0.73.0

package/build/index.js

@@ -21,6 +21,7 @@ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: tru
 var index_exports = {};
 __export(index_exports, {
   AGENT_TYPE_LABELS: () => AGENT_TYPE_LABELS,
+  ALLOWED_BUILD_COMMANDS: () => ALLOWED_BUILD_COMMANDS,
   ALL_AVAILABLE_MODEL_IDS: () => ALL_AVAILABLE_MODEL_IDS,
   AVAILABLE_CLAUDE_MODEL_IDS: () => AVAILABLE_CLAUDE_MODEL_IDS,
   AVAILABLE_OPENAI_MODEL_IDS: () => AVAILABLE_OPENAI_MODEL_IDS,
@@ -54,6 +55,7 @@ __export(index_exports, {
   BatchSummarySchema: () => BatchSummarySchema,
   BuildCheckTestSchema: () => BuildCheckTestSchema,
   BuildPassedAssertionSchema: () => BuildPassedAssertionSchema,
+  BuildPassedCommandStringSchema: () => BuildPassedCommandStringSchema,
   BuildPassedConfigSchema: () => BuildPassedConfigSchema,
   BulkImportResultItemSchema: () => BulkImportResultItemSchema,
   BulkImportResultSchema: () => BulkImportResultSchema,
@@ -81,6 +83,7 @@ __export(index_exports, {
   CreateTemplateInputSchema: () => CreateTemplateInputSchema,
   CreateTestScenarioInputSchema: () => CreateTestScenarioInputSchema,
   CreateTestSuiteInputSchema: () => CreateTestSuiteInputSchema,
+  DEFAULT_BUILD_PASSED_COMMAND: () => DEFAULT_BUILD_PASSED_COMMAND,
   DEFAULT_EVALUATOR_SYSTEM_PROMPT: () => DEFAULT_EVALUATOR_SYSTEM_PROMPT,
   DEFAULT_JUDGE_MODEL: () => DEFAULT_JUDGE_MODEL,
   DiffContentSchema: () => DiffContentSchema,
@@ -198,12 +201,15 @@ __export(index_exports, {
   formatTraceEventLine: () => formatTraceEventLine,
   getSystemAssertion: () => getSystemAssertion,
   getSystemAssertions: () => getSystemAssertions,
+  isAllowedBuildCommandString: () => isAllowedBuildCommandString,
   isSystemAssertionId: () => isSystemAssertionId,
   isValidSkillFolderName: () => isValidSkillFolderName,
   normalizeBatchAssertionLink: () => normalizeBatchAssertionLink,
   normalizeModelId: () => normalizeModelId,
+  parseBuildCommandToArgv: () => parseBuildCommandToArgv,
   parseTraceEventLine: () => parseTraceEventLine,
-  validateAssertionConfig: () => validateAssertionConfig
+  validateAssertionConfig: () => validateAssertionConfig,
+  validateBuildPassedParamsInAssertionLinks: () => validateBuildPassedParamsInAssertionLinks
 });
 module.exports = __toCommonJS(index_exports);
 
@@ -793,11 +799,42 @@ var EnvironmentSchema = import_zod21.z.object({
 });
 
 // src/scenario/test-scenario.ts
-var import_zod23 = require("zod");
+var import_zod24 = require("zod");
 
 // src/assertion/assertion.ts
+var import_zod23 = require("zod");
+
+// src/assertion/build-passed-command.ts
 var import_zod22 = require("zod");
-var AssertionTypeSchema = import_zod22.z.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 = import_zod22.z.enum(enumTuple);
+
+// src/assertion/assertion.ts
+var AssertionTypeSchema = import_zod23.z.enum([
   "skill_was_called",
   "tool_called_with_param",
   "build_passed",
@@ -806,61 +843,61 @@ var AssertionTypeSchema = import_zod22.z.enum([
   "llm_judge",
   "api_call"
 ]);
-var AssertionParameterTypeSchema = import_zod22.z.enum([
+var AssertionParameterTypeSchema = import_zod23.z.enum([
   "string",
   "number",
   "boolean"
 ]);
-var AssertionParameterSchema = import_zod22.z.object({
+var AssertionParameterSchema = import_zod23.z.object({
   /** Parameter name (used as key in params object) */
-  name: import_zod22.z.string().min(1),
+  name: import_zod23.z.string().min(1),
   /** Display label for the parameter */
-  label: import_zod22.z.string().min(1),
+  label: import_zod23.z.string().min(1),
   /** Parameter type */
   type: AssertionParameterTypeSchema,
   /** Whether this parameter is required */
-  required: import_zod22.z.boolean(),
+  required: import_zod23.z.boolean(),
   /** Default value (optional, used when not provided) */
-  defaultValue: import_zod22.z.union([import_zod22.z.string(), import_zod22.z.number(), import_zod22.z.boolean()]).optional(),
+  defaultValue: import_zod23.z.union([import_zod23.z.string(), import_zod23.z.number(), import_zod23.z.boolean()]).optional(),
   /** If true, parameter is hidden by default behind "Show advanced options" */
-  advanced: import_zod22.z.boolean().optional()
+  advanced: import_zod23.z.boolean().optional()
 });
-var ScenarioAssertionLinkSchema = import_zod22.z.object({
+var ScenarioAssertionLinkSchema = import_zod23.z.object({
   /** ID of the system assertion (e.g., 'system:skill_was_called') */
-  assertionId: import_zod22.z.string(),
+  assertionId: import_zod23.z.string(),
   /** Parameter values for this assertion in this scenario */
-  params: import_zod22.z.record(
-    import_zod22.z.string(),
-    import_zod22.z.union([import_zod22.z.string(), import_zod22.z.number(), import_zod22.z.boolean(), import_zod22.z.null()])
+  params: import_zod23.z.record(
+    import_zod23.z.string(),
+    import_zod23.z.union([import_zod23.z.string(), import_zod23.z.number(), import_zod23.z.boolean(), import_zod23.z.null()])
   ).optional()
 });
-var SkillWasCalledConfigSchema = import_zod22.z.object({
+var SkillWasCalledConfigSchema = import_zod23.z.object({
   /** Names of the skills that must have been called */
-  skillNames: import_zod22.z.array(import_zod22.z.string().min(1)).min(1)
+  skillNames: import_zod23.z.array(import_zod23.z.string().min(1)).min(1)
 });
-var CostConfigSchema = import_zod22.z.strictObject({
+var CostConfigSchema = import_zod23.z.strictObject({
   /** Maximum allowed cost in USD */
-  maxCostUsd: import_zod22.z.number().positive()
+  maxCostUsd: import_zod23.z.number().positive()
 });
-var ToolCalledWithParamConfigSchema = import_zod22.z.strictObject({
+var ToolCalledWithParamConfigSchema = import_zod23.z.strictObject({
   /** Name of the tool that must have been called */
-  toolName: import_zod22.z.string().min(1),
+  toolName: import_zod23.z.string().min(1),
   /** JSON string of key-value pairs for expected parameters (substring match). Optional — when omitted, only checks tool presence. */
-  expectedParams: import_zod22.z.string().min(1).optional(),
+  expectedParams: import_zod23.z.string().min(1).optional(),
   /** If true, the matching tool call must also have succeeded (step.success === true) */
-  requireSuccess: import_zod22.z.boolean().optional()
+  requireSuccess: import_zod23.z.boolean().optional()
 });
-var BuildPassedConfigSchema = import_zod22.z.strictObject({
-  /** Command to run (default: "yarn build") */
-  command: import_zod22.z.string().optional(),
+var BuildPassedConfigSchema = import_zod23.z.strictObject({
+  /** Allowlisted command only (default at runtime: "yarn build") */
+  command: BuildPassedCommandStringSchema.optional(),
   /** Expected exit code (default: 0) */
-  expectedExitCode: import_zod22.z.number().int().optional()
+  expectedExitCode: import_zod23.z.number().int().optional()
 });
-var TimeConfigSchema = import_zod22.z.strictObject({
+var TimeConfigSchema = import_zod23.z.strictObject({
   /** Maximum allowed duration in milliseconds */
-  maxDurationMs: import_zod22.z.number().int().positive()
+  maxDurationMs: import_zod23.z.number().int().positive()
 });
-var LlmJudgeConfigSchema = import_zod22.z.object({
+var LlmJudgeConfigSchema = import_zod23.z.object({
   /**
    * Prompt template with placeholders:
    * - {{output}}: agent's final output
@@ -871,65 +908,65 @@ var LlmJudgeConfigSchema = import_zod22.z.object({
    * - {{trace}}: step-by-step trace of tool calls
    * - Custom parameters defined in the parameters array
    */
-  prompt: import_zod22.z.string().min(1),
+  prompt: import_zod23.z.string().min(1),
   /** Minimum score to pass (0-10, default 7) */
-  minScore: import_zod22.z.number().int().min(0).max(10).optional(),
+  minScore: import_zod23.z.number().int().min(0).max(10).optional(),
   /** Model for the judge (e.g. claude-3-5-haiku-20241022) */
-  model: import_zod22.z.string().optional(),
+  model: import_zod23.z.string().optional(),
   /** Max output tokens */
-  maxTokens: import_zod22.z.number().int().optional(),
+  maxTokens: import_zod23.z.number().int().optional(),
   /** Temperature (0-1) */
-  temperature: import_zod22.z.number().min(0).max(1).optional(),
+  temperature: import_zod23.z.number().min(0).max(1).optional(),
   /** User-defined parameters for this assertion */
-  parameters: import_zod22.z.array(AssertionParameterSchema).optional()
+  parameters: import_zod23.z.array(AssertionParameterSchema).optional()
 });
-var ApiCallConfigSchema = import_zod22.z.strictObject({
+var ApiCallConfigSchema = import_zod23.z.strictObject({
   /** URL to call */
-  url: import_zod22.z.string().min(1),
+  url: import_zod23.z.string().min(1),
   /** HTTP method (default GET) */
-  method: import_zod22.z.enum(["GET", "POST"]).optional(),
+  method: import_zod23.z.enum(["GET", "POST"]).optional(),
   /** Request body (JSON string, for POST requests) */
-  requestBody: import_zod22.z.string().optional(),
+  requestBody: import_zod23.z.string().optional(),
   /** Expected JSON response to validate against (subset match — extra fields in actual are OK) */
-  expectedResponse: import_zod22.z.string().min(1),
+  expectedResponse: import_zod23.z.string().min(1),
   /** Request headers as JSON string of key-value pairs */
-  requestHeaders: import_zod22.z.string().optional(),
+  requestHeaders: import_zod23.z.string().optional(),
   /** Request timeout in milliseconds (default 30000) */
-  timeoutMs: import_zod22.z.number().int().positive().optional()
+  timeoutMs: import_zod23.z.number().int().positive().optional()
 });
 var AssertionBaseFields = {
   /** When true, the assertion's pass/fail logic is inverted (NOT operator). */
-  negate: import_zod22.z.boolean().optional()
+  negate: import_zod23.z.boolean().optional()
 };
 var SkillWasCalledAssertionSchema = SkillWasCalledConfigSchema.extend({
-  type: import_zod22.z.literal("skill_was_called"),
+  type: import_zod23.z.literal("skill_was_called"),
   ...AssertionBaseFields
 });
 var ToolCalledWithParamAssertionSchema = ToolCalledWithParamConfigSchema.extend({
-  type: import_zod22.z.literal("tool_called_with_param"),
+  type: import_zod23.z.literal("tool_called_with_param"),
   ...AssertionBaseFields
 });
 var BuildPassedAssertionSchema = BuildPassedConfigSchema.extend({
-  type: import_zod22.z.literal("build_passed"),
+  type: import_zod23.z.literal("build_passed"),
   ...AssertionBaseFields
 });
 var CostAssertionSchema = CostConfigSchema.extend({
-  type: import_zod22.z.literal("cost"),
+  type: import_zod23.z.literal("cost"),
   ...AssertionBaseFields
 });
 var LlmJudgeAssertionSchema = LlmJudgeConfigSchema.extend({
-  type: import_zod22.z.literal("llm_judge"),
+  type: import_zod23.z.literal("llm_judge"),
   ...AssertionBaseFields
 });
 var ApiCallAssertionSchema = ApiCallConfigSchema.extend({
-  type: import_zod22.z.literal("api_call"),
+  type: import_zod23.z.literal("api_call"),
   ...AssertionBaseFields
 });
 var TimeAssertionSchema = TimeConfigSchema.extend({
-  type: import_zod22.z.literal("time_limit"),
+  type: import_zod23.z.literal("time_limit"),
   ...AssertionBaseFields
 });
-var AssertionSchema = import_zod22.z.union([
+var AssertionSchema = import_zod23.z.union([
   SkillWasCalledAssertionSchema,
   ToolCalledWithParamAssertionSchema,
   BuildPassedAssertionSchema,
@@ -938,7 +975,7 @@ var AssertionSchema = import_zod22.z.union([
   LlmJudgeAssertionSchema,
   ApiCallAssertionSchema
 ]);
-var AssertionConfigSchema = import_zod22.z.union([
+var AssertionConfigSchema = import_zod23.z.union([
   LlmJudgeConfigSchema,
   // requires prompt - check first
   SkillWasCalledConfigSchema,
@@ -953,7 +990,7 @@ var AssertionConfigSchema = import_zod22.z.union([
   // requires maxCostUsd, uses strictObject
   BuildPassedConfigSchema,
   // all optional, uses strictObject to reject unknown keys
-  import_zod22.z.object({})
+  import_zod23.z.object({})
   // fallback empty config
 ]);
 function validateAssertionConfig(type, config) {
@@ -977,63 +1014,322 @@ 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 = import_zod23.z.object({
+var TriggerPromptImageSchema = import_zod24.z.object({
   /** Base64-encoded image data (no data URL prefix) */
-  base64: import_zod23.z.string().max(MAX_IMAGE_BASE64_LENGTH, "Image exceeds 2 MB size limit"),
+  base64: import_zod24.z.string().max(MAX_IMAGE_BASE64_LENGTH, "Image exceeds 2 MB size limit"),
   /** MIME type of the image */
-  mediaType: import_zod23.z.enum(["image/jpeg", "image/png", "image/gif", "image/webp"]),
+  mediaType: import_zod24.z.enum(["image/jpeg", "image/png", "image/gif", "image/webp"]),
   /** Original filename of the image */
-  name: import_zod23.z.string()
+  name: import_zod24.z.string()
 });
-var ExpectedFileSchema = import_zod23.z.object({
+var ExpectedFileSchema = import_zod24.z.object({
   /** Relative path where the file should be created */
-  path: import_zod23.z.string(),
+  path: import_zod24.z.string(),
   /** Optional expected content */
-  content: import_zod23.z.string().optional()
+  content: import_zod24.z.string().optional()
 });
 var TestScenarioSchema = TenantEntitySchema.extend({
   /** The prompt sent to the agent to trigger the task */
-  triggerPrompt: import_zod23.z.string().min(10),
+  triggerPrompt: import_zod24.z.string().min(10),
   /** ID of the template to use for this scenario (null = no template) */
-  templateId: import_zod23.z.string().nullish(),
+  templateId: import_zod24.z.string().nullish(),
   /** Inline assertions to evaluate for this scenario (legacy) */
-  assertions: import_zod23.z.array(AssertionSchema).optional(),
+  assertions: import_zod24.z.array(AssertionSchema).optional(),
   /** IDs of saved assertions to evaluate (from assertions table) - legacy, use assertionLinks */
-  assertionIds: import_zod23.z.array(import_zod23.z.string()).optional(),
+  assertionIds: import_zod24.z.array(import_zod24.z.string()).optional(),
   /** Linked assertions with per-scenario parameter values */
-  assertionLinks: import_zod23.z.array(ScenarioAssertionLinkSchema).optional(),
+  assertionLinks: import_zod24.z.array(ScenarioAssertionLinkSchema).optional(),
   /** Tags for categorisation and filtering */
-  tags: import_zod23.z.array(import_zod23.z.string()).optional(),
+  tags: import_zod24.z.array(import_zod24.z.string()).optional(),
   /** Base64-encoded images attached to the trigger prompt (max 3) */
-  triggerPromptImages: import_zod23.z.array(TriggerPromptImageSchema).max(3).optional()
-});
-var CreateTestScenarioInputSchema = TestScenarioSchema.omit({
+  triggerPromptImages: import_zod24.z.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: import_zod24.z.ZodIssueCode.custom,
+        message: "build_passed command must be a string",
+        path: ["assertionLinks", i, "params", "command"]
+      });
+      continue;
+    }
+    if (!isAllowedBuildCommandString(cmd)) {
+      ctx.addIssue({
+        code: import_zod24.z.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
-var import_zod24 = require("zod");
+var import_zod25 = require("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 = import_zod24.z.union([
-  import_zod24.z.string().min(1),
+var BatchAssertionLinkSchema = import_zod25.z.union([
+  import_zod25.z.string().min(1),
   ScenarioAssertionLinkSchema
 ]);
-var BatchScenarioEntrySchema = import_zod24.z.object({
-  name: import_zod24.z.string().min(1, "name: Required"),
-  description: import_zod24.z.string().optional().default(""),
-  triggerPrompt: import_zod24.z.string().min(10, "triggerPrompt: Must be at least 10 characters"),
-  templateId: import_zod24.z.string().nullish(),
-  tags: import_zod24.z.array(import_zod24.z.string()).optional(),
-  assertionLinks: import_zod24.z.array(BatchAssertionLinkSchema).optional()
+var BatchScenarioEntrySchema = import_zod25.z.object({
+  name: import_zod25.z.string().min(1, "name: Required"),
+  description: import_zod25.z.string().optional().default(""),
+  triggerPrompt: import_zod25.z.string().min(10, "triggerPrompt: Must be at least 10 characters"),
+  templateId: import_zod25.z.string().nullish(),
+  tags: import_zod25.z.array(import_zod25.z.string()).optional(),
+  assertionLinks: import_zod25.z.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 = import_zod24.z.object({
-  scenarios: import_zod24.z.array(BatchScenarioEntrySchema).min(1, "scenarios array must contain at least one entry").max(100, "Maximum 100 scenarios per upload")
+var BatchImportPayloadSchema = import_zod25.z.object({
+  scenarios: import_zod25.z.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,
@@ -1055,29 +1351,29 @@ function normalizeBatchAssertionLink(link) {
   }
   return link;
 }
-var BatchResultItemSchema = import_zod24.z.object({
-  index: import_zod24.z.number(),
-  name: import_zod24.z.string(),
-  status: import_zod24.z.enum(["valid", "invalid"]),
-  id: import_zod24.z.string().nullable().optional(),
-  errors: import_zod24.z.array(import_zod24.z.string()).optional()
-});
-var BatchSummarySchema = import_zod24.z.object({
-  total: import_zod24.z.number(),
-  valid: import_zod24.z.number(),
-  invalid: import_zod24.z.number(),
-  created: import_zod24.z.number()
-});
-var BatchImportResponseSchema = import_zod24.z.object({
+var BatchResultItemSchema = import_zod25.z.object({
+  index: import_zod25.z.number(),
+  name: import_zod25.z.string(),
+  status: import_zod25.z.enum(["valid", "invalid"]),
+  id: import_zod25.z.string().nullable().optional(),
+  errors: import_zod25.z.array(import_zod25.z.string()).optional()
+});
+var BatchSummarySchema = import_zod25.z.object({
+  total: import_zod25.z.number(),
+  valid: import_zod25.z.number(),
+  invalid: import_zod25.z.number(),
+  created: import_zod25.z.number()
+});
+var BatchImportResponseSchema = import_zod25.z.object({
   summary: BatchSummarySchema,
-  results: import_zod24.z.array(BatchResultItemSchema)
+  results: import_zod25.z.array(BatchResultItemSchema)
 });
 
 // src/suite/test-suite.ts
-var import_zod25 = require("zod");
+var import_zod26 = require("zod");
 var TestSuiteSchema = TenantEntitySchema.extend({
   /** IDs of test scenarios in this suite */
-  scenarioIds: import_zod25.z.array(import_zod25.z.string())
+  scenarioIds: import_zod26.z.array(import_zod26.z.string())
 });
 var CreateTestSuiteInputSchema = TestSuiteSchema.omit({
   id: true,
@@ -1088,21 +1384,21 @@ var CreateTestSuiteInputSchema = TestSuiteSchema.omit({
 var UpdateTestSuiteInputSchema = CreateTestSuiteInputSchema.partial();
 
 // src/evaluation/metrics.ts
-var import_zod26 = require("zod");
-var TokenUsageSchema = import_zod26.z.object({
-  prompt: import_zod26.z.number(),
-  completion: import_zod26.z.number(),
-  total: import_zod26.z.number()
-});
-var EvalMetricsSchema = import_zod26.z.object({
-  totalAssertions: import_zod26.z.number(),
-  passed: import_zod26.z.number(),
-  failed: import_zod26.z.number(),
-  skipped: import_zod26.z.number(),
-  errors: import_zod26.z.number(),
-  passRate: import_zod26.z.number(),
-  avgDuration: import_zod26.z.number(),
-  totalDuration: import_zod26.z.number()
+var import_zod27 = require("zod");
+var TokenUsageSchema = import_zod27.z.object({
+  prompt: import_zod27.z.number(),
+  completion: import_zod27.z.number(),
+  total: import_zod27.z.number()
+});
+var EvalMetricsSchema = import_zod27.z.object({
+  totalAssertions: import_zod27.z.number(),
+  passed: import_zod27.z.number(),
+  failed: import_zod27.z.number(),
+  skipped: import_zod27.z.number(),
+  errors: import_zod27.z.number(),
+  passRate: import_zod27.z.number(),
+  avgDuration: import_zod27.z.number(),
+  totalDuration: import_zod27.z.number()
 });
 var EvalStatus = /* @__PURE__ */ ((EvalStatus2) => {
   EvalStatus2["PENDING"] = "pending";
@@ -1112,7 +1408,7 @@ var EvalStatus = /* @__PURE__ */ ((EvalStatus2) => {
   EvalStatus2["CANCELLED"] = "cancelled";
   return EvalStatus2;
 })(EvalStatus || {});
-var EvalStatusSchema = import_zod26.z.enum(EvalStatus);
+var EvalStatusSchema = import_zod27.z.enum(EvalStatus);
 var LLMStepType = /* @__PURE__ */ ((LLMStepType2) => {
   LLMStepType2["COMPLETION"] = "completion";
   LLMStepType2["TOOL_USE"] = "tool_use";
@@ -1120,54 +1416,54 @@ var LLMStepType = /* @__PURE__ */ ((LLMStepType2) => {
   LLMStepType2["THINKING"] = "thinking";
   return LLMStepType2;
 })(LLMStepType || {});
-var LLMTraceStepSchema = import_zod26.z.object({
-  id: import_zod26.z.string(),
-  stepNumber: import_zod26.z.number(),
-  type: import_zod26.z.enum(LLMStepType),
-  model: import_zod26.z.string(),
-  provider: import_zod26.z.string(),
-  startedAt: import_zod26.z.string(),
-  durationMs: import_zod26.z.number(),
+var LLMTraceStepSchema = import_zod27.z.object({
+  id: import_zod27.z.string(),
+  stepNumber: import_zod27.z.number(),
+  type: import_zod27.z.enum(LLMStepType),
+  model: import_zod27.z.string(),
+  provider: import_zod27.z.string(),
+  startedAt: import_zod27.z.string(),
+  durationMs: import_zod27.z.number(),
   tokenUsage: TokenUsageSchema,
-  costUsd: import_zod26.z.number(),
-  toolName: import_zod26.z.string().optional(),
-  toolArguments: import_zod26.z.string().optional(),
-  inputPreview: import_zod26.z.string().optional(),
-  outputPreview: import_zod26.z.string().optional(),
-  success: import_zod26.z.boolean(),
-  error: import_zod26.z.string().optional(),
-  turnIndex: import_zod26.z.number().optional()
-});
-var LLMBreakdownStatsSchema = import_zod26.z.object({
-  count: import_zod26.z.number(),
-  durationMs: import_zod26.z.number(),
-  tokens: import_zod26.z.number(),
-  costUsd: import_zod26.z.number()
-});
-var LLMTraceSummarySchema = import_zod26.z.object({
-  totalSteps: import_zod26.z.number(),
-  totalTurns: import_zod26.z.number().optional(),
-  totalDurationMs: import_zod26.z.number(),
+  costUsd: import_zod27.z.number(),
+  toolName: import_zod27.z.string().optional(),
+  toolArguments: import_zod27.z.string().optional(),
+  inputPreview: import_zod27.z.string().optional(),
+  outputPreview: import_zod27.z.string().optional(),
+  success: import_zod27.z.boolean(),
+  error: import_zod27.z.string().optional(),
+  turnIndex: import_zod27.z.number().optional()
+});
+var LLMBreakdownStatsSchema = import_zod27.z.object({
+  count: import_zod27.z.number(),
+  durationMs: import_zod27.z.number(),
+  tokens: import_zod27.z.number(),
+  costUsd: import_zod27.z.number()
+});
+var LLMTraceSummarySchema = import_zod27.z.object({
+  totalSteps: import_zod27.z.number(),
+  totalTurns: import_zod27.z.number().optional(),
+  totalDurationMs: import_zod27.z.number(),
   totalTokens: TokenUsageSchema,
-  totalCostUsd: import_zod26.z.number(),
-  stepTypeBreakdown: import_zod26.z.record(import_zod26.z.string(), LLMBreakdownStatsSchema).optional(),
-  modelBreakdown: import_zod26.z.record(import_zod26.z.string(), LLMBreakdownStatsSchema),
-  modelsUsed: import_zod26.z.array(import_zod26.z.string())
-});
-var LLMTraceSchema = import_zod26.z.object({
-  id: import_zod26.z.string(),
-  steps: import_zod26.z.array(LLMTraceStepSchema),
+  totalCostUsd: import_zod27.z.number(),
+  stepTypeBreakdown: import_zod27.z.record(import_zod27.z.string(), LLMBreakdownStatsSchema).optional(),
+  modelBreakdown: import_zod27.z.record(import_zod27.z.string(), LLMBreakdownStatsSchema),
+  modelsUsed: import_zod27.z.array(import_zod27.z.string())
+});
+var LLMTraceSchema = import_zod27.z.object({
+  id: import_zod27.z.string(),
+  steps: import_zod27.z.array(LLMTraceStepSchema),
   summary: LLMTraceSummarySchema
 });
 
 // src/evaluation/eval-result.ts
-var import_zod30 = require("zod");
+var import_zod31 = require("zod");
 
 // src/evaluation/eval-run.ts
-var import_zod28 = require("zod");
+var import_zod29 = require("zod");
 
 // src/evaluation/live-trace.ts
-var import_zod27 = require("zod");
+var import_zod28 = require("zod");
 var LiveTraceEventType = /* @__PURE__ */ ((LiveTraceEventType2) => {
   LiveTraceEventType2["THINKING"] = "thinking";
   LiveTraceEventType2["TOOL_USE"] = "tool_use";
@@ -1181,37 +1477,37 @@ var LiveTraceEventType = /* @__PURE__ */ ((LiveTraceEventType2) => {
   LiveTraceEventType2["USER"] = "user";
   return LiveTraceEventType2;
 })(LiveTraceEventType || {});
-var LiveTraceEventSchema = import_zod27.z.object({
+var LiveTraceEventSchema = import_zod28.z.object({
   /** The evaluation run ID */
-  evalRunId: import_zod27.z.string(),
+  evalRunId: import_zod28.z.string(),
   /** The scenario ID being executed */
-  scenarioId: import_zod27.z.string(),
+  scenarioId: import_zod28.z.string(),
   /** The scenario name for display */
-  scenarioName: import_zod27.z.string(),
+  scenarioName: import_zod28.z.string(),
   /** The target ID (skill, agent, etc.) */
-  targetId: import_zod27.z.string(),
+  targetId: import_zod28.z.string(),
   /** The target name for display */
-  targetName: import_zod27.z.string(),
+  targetName: import_zod28.z.string(),
   /** Step number in the current scenario execution */
-  stepNumber: import_zod27.z.number(),
+  stepNumber: import_zod28.z.number(),
   /** Type of trace event */
-  type: import_zod27.z.enum(LiveTraceEventType),
+  type: import_zod28.z.enum(LiveTraceEventType),
   /** Tool name if this is a tool_use event */
-  toolName: import_zod27.z.string().optional(),
+  toolName: import_zod28.z.string().optional(),
   /** Tool arguments preview (truncated JSON) */
-  toolArgs: import_zod27.z.string().optional(),
+  toolArgs: import_zod28.z.string().optional(),
   /** Output preview (truncated text) */
-  outputPreview: import_zod27.z.string().optional(),
+  outputPreview: import_zod28.z.string().optional(),
   /** File path for file operations */
-  filePath: import_zod27.z.string().optional(),
+  filePath: import_zod28.z.string().optional(),
   /** Elapsed time in milliseconds for progress events */
-  elapsedMs: import_zod27.z.number().optional(),
+  elapsedMs: import_zod28.z.number().optional(),
   /** Thinking/reasoning text from Claude */
-  thinking: import_zod27.z.string().optional(),
+  thinking: import_zod28.z.string().optional(),
   /** Timestamp when this event occurred */
-  timestamp: import_zod27.z.string(),
+  timestamp: import_zod28.z.string(),
   /** Whether this is the final event for this scenario */
-  isComplete: import_zod27.z.boolean()
+  isComplete: import_zod28.z.boolean()
 });
 var TRACE_EVENT_PREFIX = "TRACE_EVENT:";
 function parseTraceEventLine(line) {
@@ -1240,40 +1536,40 @@ var TriggerType = /* @__PURE__ */ ((TriggerType2) => {
   TriggerType2["SCHEDULED"] = "SCHEDULED";
   return TriggerType2;
 })(TriggerType || {});
-var TriggerMetadataSchema = import_zod28.z.object({
-  version: import_zod28.z.string().optional(),
-  resourceUpdated: import_zod28.z.array(import_zod28.z.string()).optional(),
-  scheduleId: import_zod28.z.string().optional()
+var TriggerMetadataSchema = import_zod29.z.object({
+  version: import_zod29.z.string().optional(),
+  resourceUpdated: import_zod29.z.array(import_zod29.z.string()).optional(),
+  scheduleId: import_zod29.z.string().optional()
 });
-var TriggerSchema = import_zod28.z.object({
-  id: import_zod28.z.string(),
+var TriggerSchema = import_zod29.z.object({
+  id: import_zod29.z.string(),
   metadata: TriggerMetadataSchema.optional(),
-  type: import_zod28.z.nativeEnum(TriggerType)
+  type: import_zod29.z.nativeEnum(TriggerType)
 });
-var DiffLineTypeSchema = import_zod28.z.enum(["added", "removed", "unchanged"]);
-var DiffLineSchema = import_zod28.z.object({
+var DiffLineTypeSchema = import_zod29.z.enum(["added", "removed", "unchanged"]);
+var DiffLineSchema = import_zod29.z.object({
   type: DiffLineTypeSchema,
-  content: import_zod28.z.string(),
-  lineNumber: import_zod28.z.number()
-});
-var DiffContentSchema = import_zod28.z.object({
-  path: import_zod28.z.string(),
-  expected: import_zod28.z.string(),
-  actual: import_zod28.z.string(),
-  diffLines: import_zod28.z.array(DiffLineSchema),
-  renamedFrom: import_zod28.z.string().optional(),
+  content: import_zod29.z.string(),
+  lineNumber: import_zod29.z.number()
+});
+var DiffContentSchema = import_zod29.z.object({
+  path: import_zod29.z.string(),
+  expected: import_zod29.z.string(),
+  actual: import_zod29.z.string(),
+  diffLines: import_zod29.z.array(DiffLineSchema),
+  renamedFrom: import_zod29.z.string().optional(),
   /** Whether this file is an infrastructure/config file (e.g. .claude/settings.json, .mcp.json) */
-  isInfrastructure: import_zod28.z.boolean().optional()
+  isInfrastructure: import_zod29.z.boolean().optional()
 });
-var CommandExecutionSchema = import_zod28.z.object({
-  command: import_zod28.z.string(),
-  exitCode: import_zod28.z.number(),
-  output: import_zod28.z.string().optional(),
-  duration: import_zod28.z.number()
+var CommandExecutionSchema = import_zod29.z.object({
+  command: import_zod29.z.string(),
+  exitCode: import_zod29.z.number(),
+  output: import_zod29.z.string().optional(),
+  duration: import_zod29.z.number()
 });
-var FileModificationSchema = import_zod28.z.object({
-  path: import_zod28.z.string(),
-  action: import_zod28.z.enum(["created", "modified", "deleted"])
+var FileModificationSchema = import_zod29.z.object({
+  path: import_zod29.z.string(),
+  action: import_zod29.z.enum(["created", "modified", "deleted"])
 });
 var TemplateFileStatus = /* @__PURE__ */ ((TemplateFileStatus2) => {
   TemplateFileStatus2["NEW"] = "new";
@@ -1281,62 +1577,62 @@ var TemplateFileStatus = /* @__PURE__ */ ((TemplateFileStatus2) => {
   TemplateFileStatus2["UNCHANGED"] = "unchanged";
   return TemplateFileStatus2;
 })(TemplateFileStatus || {});
-var TemplateFileSchema = import_zod28.z.object({
+var TemplateFileSchema = import_zod29.z.object({
   /** Relative path within the template */
-  path: import_zod28.z.string(),
+  path: import_zod29.z.string(),
   /** Full file content after execution */
-  content: import_zod28.z.string(),
+  content: import_zod29.z.string(),
   /** File status (new, modified, unchanged) */
-  status: import_zod28.z.enum(["new", "modified", "unchanged"]),
+  status: import_zod29.z.enum(["new", "modified", "unchanged"]),
   /** Whether this file is an infrastructure/config file (e.g. .claude/settings.json, .mcp.json) */
-  isInfrastructure: import_zod28.z.boolean().optional()
+  isInfrastructure: import_zod29.z.boolean().optional()
 });
-var ApiCallSchema = import_zod28.z.object({
-  endpoint: import_zod28.z.string(),
-  tokensUsed: import_zod28.z.number(),
-  duration: import_zod28.z.number()
+var ApiCallSchema = import_zod29.z.object({
+  endpoint: import_zod29.z.string(),
+  tokensUsed: import_zod29.z.number(),
+  duration: import_zod29.z.number()
 });
-var ExecutionTraceSchema = import_zod28.z.object({
-  commands: import_zod28.z.array(CommandExecutionSchema),
-  filesModified: import_zod28.z.array(FileModificationSchema),
-  apiCalls: import_zod28.z.array(ApiCallSchema),
-  totalDuration: import_zod28.z.number()
+var ExecutionTraceSchema = import_zod29.z.object({
+  commands: import_zod29.z.array(CommandExecutionSchema),
+  filesModified: import_zod29.z.array(FileModificationSchema),
+  apiCalls: import_zod29.z.array(ApiCallSchema),
+  totalDuration: import_zod29.z.number()
 });
-var RunAnalysisFindingSchema = import_zod28.z.object({
-  category: import_zod28.z.enum([
+var RunAnalysisFindingSchema = import_zod29.z.object({
+  category: import_zod29.z.enum([
     "failure_pattern",
     "cost_waste",
     "flakiness",
     "inefficiency",
     "positive"
   ]),
-  severity: import_zod28.z.enum(["high", "medium", "low"]),
-  description: import_zod28.z.string(),
-  affectedScenarios: import_zod28.z.array(import_zod28.z.string()),
-  recommendation: import_zod28.z.string().optional()
+  severity: import_zod29.z.enum(["high", "medium", "low"]),
+  description: import_zod29.z.string(),
+  affectedScenarios: import_zod29.z.array(import_zod29.z.string()),
+  recommendation: import_zod29.z.string().optional()
 });
-var RunAnalysisSchema = import_zod28.z.object({
-  generatedAt: import_zod28.z.string(),
-  summary: import_zod28.z.string(),
-  findings: import_zod28.z.array(RunAnalysisFindingSchema)
+var RunAnalysisSchema = import_zod29.z.object({
+  generatedAt: import_zod29.z.string(),
+  summary: import_zod29.z.string(),
+  findings: import_zod29.z.array(RunAnalysisFindingSchema)
 });
 var EvalRunSchema = TenantEntitySchema.extend({
   /** Agent ID for this run */
-  agentId: import_zod28.z.string().optional(),
+  agentId: import_zod29.z.string().optional(),
   /** Preset ID that originated this run (optional) */
-  presetId: import_zod28.z.string().optional(),
+  presetId: import_zod29.z.string().optional(),
   /** Skill IDs for this run */
-  skillIds: import_zod28.z.array(import_zod28.z.string()).optional(),
+  skillIds: import_zod29.z.array(import_zod29.z.string()).optional(),
   /** Map of skillId to skillVersionId for this run */
-  skillVersions: import_zod28.z.record(import_zod28.z.string(), import_zod28.z.string()).optional(),
+  skillVersions: import_zod29.z.record(import_zod29.z.string(), import_zod29.z.string()).optional(),
   /** Scenario IDs to run (always present — resolved server-side from tags when needed) */
-  scenarioIds: import_zod28.z.array(import_zod28.z.string()),
+  scenarioIds: import_zod29.z.array(import_zod29.z.string()),
   /** Current status */
   status: EvalStatusSchema,
   /** Progress percentage (0-100) */
-  progress: import_zod28.z.number(),
+  progress: import_zod29.z.number(),
   /** Results for each scenario/target combination (lazy to break eval-result ↔ eval-run cycle) */
-  results: import_zod28.z.array(import_zod28.z.lazy(() => EvalRunResultSchema)),
+  results: import_zod29.z.array(import_zod29.z.lazy(() => EvalRunResultSchema)),
   /** Aggregated metrics across all results */
   aggregateMetrics: EvalMetricsSchema,
   /** Aggregated LLM trace summary */
@@ -1344,41 +1640,41 @@ var EvalRunSchema = TenantEntitySchema.extend({
   /** What triggered this run */
   trigger: TriggerSchema.optional(),
   /** When the run started (set when evaluation is triggered) */
-  startedAt: import_zod28.z.string().optional(),
+  startedAt: import_zod29.z.string().optional(),
   /** When the run completed */
-  completedAt: import_zod28.z.string().optional(),
+  completedAt: import_zod29.z.string().optional(),
   /** Live trace events captured during execution (for playback on results page) */
-  liveTraceEvents: import_zod28.z.array(LiveTraceEventSchema).optional(),
+  liveTraceEvents: import_zod29.z.array(LiveTraceEventSchema).optional(),
   /** Remote job ID for tracking execution in Dev Machines */
-  jobId: import_zod28.z.string().optional(),
+  jobId: import_zod29.z.string().optional(),
   /** Remote job status from the Dev Machine API (PENDING, RUNNING, COMPLETED, FAILED, CANCELLED) */
-  jobStatus: import_zod28.z.string().optional(),
+  jobStatus: import_zod29.z.string().optional(),
   /** Remote job error message if the job failed */
-  jobError: import_zod28.z.string().optional(),
+  jobError: import_zod29.z.string().optional(),
   /** Timestamp of the last job status check */
-  jobStatusCheckedAt: import_zod28.z.string().optional(),
+  jobStatusCheckedAt: import_zod29.z.string().optional(),
   /** MCP server IDs to enable for this run (optional) */
-  mcpIds: import_zod28.z.array(import_zod28.z.string()).optional(),
+  mcpIds: import_zod29.z.array(import_zod29.z.string()).optional(),
   /** Sub-agent IDs to enable for this run (optional) */
-  subAgentIds: import_zod28.z.array(import_zod28.z.string()).optional(),
+  subAgentIds: import_zod29.z.array(import_zod29.z.string()).optional(),
   /** Rule IDs to enable for this run (optional) */
-  ruleIds: import_zod28.z.array(import_zod28.z.string()).optional(),
+  ruleIds: import_zod29.z.array(import_zod29.z.string()).optional(),
   /** Tags used to select scenarios for this run (for traceability) */
-  tags: import_zod28.z.array(import_zod28.z.string()).optional(),
+  tags: import_zod29.z.array(import_zod29.z.string()).optional(),
   /** How many times each scenario is executed within this eval run. Default: 1. Max: 20. */
-  runsPerScenario: import_zod28.z.number().int().min(1).max(20).optional(),
+  runsPerScenario: import_zod29.z.number().int().min(1).max(20).optional(),
   /** Snapshot of agent configuration captured at run creation time */
-  agentSnapshot: import_zod28.z.object({
-    name: import_zod28.z.string().optional(),
+  agentSnapshot: import_zod29.z.object({
+    name: import_zod29.z.string().optional(),
     agentType: AgentTypeSchema.optional(),
     runCommand: AgentRunCommandSchema.optional(),
-    systemPrompt: import_zod28.z.string().nullable().optional(),
+    systemPrompt: import_zod29.z.string().nullable().optional(),
     modelConfig: ModelConfigSchema.optional()
   }).optional(),
   /** UUID linking all runs in a comparison group */
-  comparisonGroupId: import_zod28.z.string().optional(),
+  comparisonGroupId: import_zod29.z.string().optional(),
   /** Human-readable label for this variant (e.g., "MCP: Wix Stores") */
-  comparisonLabel: import_zod28.z.string().optional(),
+  comparisonLabel: import_zod29.z.string().optional(),
   /** LLM-generated analysis of the completed run */
   runAnalysis: RunAnalysisSchema.optional()
 });
@@ -1396,60 +1692,60 @@ var CreateEvalRunInputSchema = EvalRunSchema.omit({
   agentSnapshot: true
 }).extend({
   /** Optional on input — backend resolves from tags when not provided */
-  scenarioIds: import_zod28.z.array(import_zod28.z.string()).optional()
+  scenarioIds: import_zod29.z.array(import_zod29.z.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 = import_zod28.z.object({
-  runId: import_zod28.z.string(),
-  targetId: import_zod28.z.string(),
-  totalScenarios: import_zod28.z.number(),
-  completedScenarios: import_zod28.z.number(),
-  scenarioProgress: import_zod28.z.array(
-    import_zod28.z.object({
-      scenarioId: import_zod28.z.string(),
-      currentStep: import_zod28.z.string(),
-      error: import_zod28.z.string().optional()
+var EvaluationProgressSchema = import_zod29.z.object({
+  runId: import_zod29.z.string(),
+  targetId: import_zod29.z.string(),
+  totalScenarios: import_zod29.z.number(),
+  completedScenarios: import_zod29.z.number(),
+  scenarioProgress: import_zod29.z.array(
+    import_zod29.z.object({
+      scenarioId: import_zod29.z.string(),
+      currentStep: import_zod29.z.string(),
+      error: import_zod29.z.string().optional()
     })
   ),
-  createdAt: import_zod28.z.number()
-});
-var EvaluationLogSchema = import_zod28.z.object({
-  runId: import_zod28.z.string(),
-  scenarioId: import_zod28.z.string(),
-  log: import_zod28.z.object({
-    level: import_zod28.z.enum(["info", "error", "debug"]),
-    message: import_zod28.z.string().optional(),
-    args: import_zod28.z.array(import_zod28.z.any()).optional(),
-    error: import_zod28.z.string().optional()
+  createdAt: import_zod29.z.number()
+});
+var EvaluationLogSchema = import_zod29.z.object({
+  runId: import_zod29.z.string(),
+  scenarioId: import_zod29.z.string(),
+  log: import_zod29.z.object({
+    level: import_zod29.z.enum(["info", "error", "debug"]),
+    message: import_zod29.z.string().optional(),
+    args: import_zod29.z.array(import_zod29.z.any()).optional(),
+    error: import_zod29.z.string().optional()
   })
 });
 var LLM_TIMEOUT = 12e4;
 
 // src/evaluation/conversation.ts
-var import_zod29 = require("zod");
-var TextBlockSchema = import_zod29.z.object({
-  type: import_zod29.z.literal("text"),
-  text: import_zod29.z.string()
-});
-var ThinkingBlockSchema = import_zod29.z.object({
-  type: import_zod29.z.literal("thinking"),
-  thinking: import_zod29.z.string()
-});
-var ToolUseBlockSchema = import_zod29.z.object({
-  type: import_zod29.z.literal("tool_use"),
-  toolName: import_zod29.z.string(),
-  toolId: import_zod29.z.string(),
-  input: import_zod29.z.unknown()
-});
-var ToolResultBlockSchema = import_zod29.z.object({
-  type: import_zod29.z.literal("tool_result"),
-  toolUseId: import_zod29.z.string(),
-  content: import_zod29.z.string(),
-  isError: import_zod29.z.boolean().optional()
-});
-var ConversationBlockSchema = import_zod29.z.discriminatedUnion("type", [
+var import_zod30 = require("zod");
+var TextBlockSchema = import_zod30.z.object({
+  type: import_zod30.z.literal("text"),
+  text: import_zod30.z.string()
+});
+var ThinkingBlockSchema = import_zod30.z.object({
+  type: import_zod30.z.literal("thinking"),
+  thinking: import_zod30.z.string()
+});
+var ToolUseBlockSchema = import_zod30.z.object({
+  type: import_zod30.z.literal("tool_use"),
+  toolName: import_zod30.z.string(),
+  toolId: import_zod30.z.string(),
+  input: import_zod30.z.unknown()
+});
+var ToolResultBlockSchema = import_zod30.z.object({
+  type: import_zod30.z.literal("tool_result"),
+  toolUseId: import_zod30.z.string(),
+  content: import_zod30.z.string(),
+  isError: import_zod30.z.boolean().optional()
+});
+var ConversationBlockSchema = import_zod30.z.discriminatedUnion("type", [
   TextBlockSchema,
   ThinkingBlockSchema,
   ToolUseBlockSchema,
@@ -1460,18 +1756,18 @@ var ConversationMessageRoles = [
   "user",
   "system"
 ];
-var ConversationMessageSchema = import_zod29.z.object({
-  role: import_zod29.z.enum(ConversationMessageRoles),
-  content: import_zod29.z.array(ConversationBlockSchema),
-  timestamp: import_zod29.z.string()
+var ConversationMessageSchema = import_zod30.z.object({
+  role: import_zod30.z.enum(ConversationMessageRoles),
+  content: import_zod30.z.array(ConversationBlockSchema),
+  timestamp: import_zod30.z.string()
 });
-var ScenarioConversationSchema = import_zod29.z.object({
-  id: import_zod29.z.string(),
-  projectId: import_zod29.z.string(),
-  evalRunId: import_zod29.z.string(),
-  resultId: import_zod29.z.string(),
-  messages: import_zod29.z.array(ConversationMessageSchema),
-  createdAt: import_zod29.z.string()
+var ScenarioConversationSchema = import_zod30.z.object({
+  id: import_zod30.z.string(),
+  projectId: import_zod30.z.string(),
+  evalRunId: import_zod30.z.string(),
+  resultId: import_zod30.z.string(),
+  messages: import_zod30.z.array(ConversationMessageSchema),
+  createdAt: import_zod30.z.string()
 });
 
 // src/evaluation/eval-result.ts
@@ -1482,98 +1778,98 @@ var AssertionResultStatus = /* @__PURE__ */ ((AssertionResultStatus2) => {
   AssertionResultStatus2["ERROR"] = "error";
   return AssertionResultStatus2;
 })(AssertionResultStatus || {});
-var AssertionResultSchema = import_zod30.z.object({
-  id: import_zod30.z.string(),
-  assertionId: import_zod30.z.string(),
-  assertionType: import_zod30.z.string(),
-  assertionName: import_zod30.z.string(),
-  status: import_zod30.z.enum(AssertionResultStatus),
-  message: import_zod30.z.string().optional(),
-  expected: import_zod30.z.string().optional(),
-  actual: import_zod30.z.string().optional(),
-  duration: import_zod30.z.number().optional(),
-  details: import_zod30.z.record(import_zod30.z.string(), import_zod30.z.unknown()).optional(),
-  llmTraceSteps: import_zod30.z.array(LLMTraceStepSchema).optional()
-});
-var EvalRunResultSchema = import_zod30.z.object({
-  id: import_zod30.z.string(),
-  targetId: import_zod30.z.string(),
-  targetName: import_zod30.z.string().optional(),
+var AssertionResultSchema = import_zod31.z.object({
+  id: import_zod31.z.string(),
+  assertionId: import_zod31.z.string(),
+  assertionType: import_zod31.z.string(),
+  assertionName: import_zod31.z.string(),
+  status: import_zod31.z.enum(AssertionResultStatus),
+  message: import_zod31.z.string().optional(),
+  expected: import_zod31.z.string().optional(),
+  actual: import_zod31.z.string().optional(),
+  duration: import_zod31.z.number().optional(),
+  details: import_zod31.z.record(import_zod31.z.string(), import_zod31.z.unknown()).optional(),
+  llmTraceSteps: import_zod31.z.array(LLMTraceStepSchema).optional()
+});
+var EvalRunResultSchema = import_zod31.z.object({
+  id: import_zod31.z.string(),
+  targetId: import_zod31.z.string(),
+  targetName: import_zod31.z.string().optional(),
   /** SkillVersion ID used for this evaluation (for version tracking) */
-  skillVersionId: import_zod30.z.string().optional(),
+  skillVersionId: import_zod31.z.string().optional(),
   /** SkillVersion semver string (e.g., "1.0.0", "1.2.3") for display */
-  skillVersion: import_zod30.z.string().optional(),
-  scenarioId: import_zod30.z.string(),
-  scenarioName: import_zod30.z.string(),
+  skillVersion: import_zod31.z.string().optional(),
+  scenarioId: import_zod31.z.string(),
+  scenarioName: import_zod31.z.string(),
   /** Snapshot of the trigger prompt used during the run (prevents stale display after edits) */
-  triggerPrompt: import_zod30.z.string().optional(),
+  triggerPrompt: import_zod31.z.string().optional(),
   modelConfig: ModelConfigSchema.optional(),
-  assertionResults: import_zod30.z.array(AssertionResultSchema),
+  assertionResults: import_zod31.z.array(AssertionResultSchema),
   metrics: EvalMetricsSchema.optional(),
-  passed: import_zod30.z.number(),
-  failed: import_zod30.z.number(),
-  passRate: import_zod30.z.number(),
-  duration: import_zod30.z.number(),
-  outputText: import_zod30.z.string().optional(),
-  files: import_zod30.z.array(ExpectedFileSchema).optional(),
-  fileDiffs: import_zod30.z.array(DiffContentSchema).optional(),
+  passed: import_zod31.z.number(),
+  failed: import_zod31.z.number(),
+  passRate: import_zod31.z.number(),
+  duration: import_zod31.z.number(),
+  outputText: import_zod31.z.string().optional(),
+  files: import_zod31.z.array(ExpectedFileSchema).optional(),
+  fileDiffs: import_zod31.z.array(DiffContentSchema).optional(),
   /** Full template files after execution with status indicators */
-  templateFiles: import_zod30.z.array(TemplateFileSchema).optional(),
-  startedAt: import_zod30.z.string().optional(),
-  completedAt: import_zod30.z.string().optional(),
+  templateFiles: import_zod31.z.array(TemplateFileSchema).optional(),
+  startedAt: import_zod31.z.string().optional(),
+  completedAt: import_zod31.z.string().optional(),
   llmTrace: LLMTraceSchema.optional(),
   /** Full conversation messages (only present in transit; stripped before DB storage) */
-  conversation: import_zod30.z.array(ConversationMessageSchema).optional(),
+  conversation: import_zod31.z.array(ConversationMessageSchema).optional(),
   /** 0-based iteration index when a scenario is run multiple times within a single eval run */
-  iterationIndex: import_zod30.z.number().int().min(0).optional()
-});
-var PromptResultSchema = import_zod30.z.object({
-  text: import_zod30.z.string(),
-  files: import_zod30.z.array(import_zod30.z.unknown()).optional(),
-  finishReason: import_zod30.z.string().optional(),
-  reasoning: import_zod30.z.string().optional(),
-  reasoningDetails: import_zod30.z.unknown().optional(),
-  toolCalls: import_zod30.z.array(import_zod30.z.unknown()).optional(),
-  toolResults: import_zod30.z.array(import_zod30.z.unknown()).optional(),
-  warnings: import_zod30.z.array(import_zod30.z.unknown()).optional(),
-  sources: import_zod30.z.array(import_zod30.z.unknown()).optional(),
-  steps: import_zod30.z.array(import_zod30.z.unknown()),
-  generationTimeMs: import_zod30.z.number(),
-  prompt: import_zod30.z.string(),
-  systemPrompt: import_zod30.z.string(),
-  usage: import_zod30.z.object({
-    totalTokens: import_zod30.z.number().optional(),
-    totalMicrocentsSpent: import_zod30.z.number().optional()
+  iterationIndex: import_zod31.z.number().int().min(0).optional()
+});
+var PromptResultSchema = import_zod31.z.object({
+  text: import_zod31.z.string(),
+  files: import_zod31.z.array(import_zod31.z.unknown()).optional(),
+  finishReason: import_zod31.z.string().optional(),
+  reasoning: import_zod31.z.string().optional(),
+  reasoningDetails: import_zod31.z.unknown().optional(),
+  toolCalls: import_zod31.z.array(import_zod31.z.unknown()).optional(),
+  toolResults: import_zod31.z.array(import_zod31.z.unknown()).optional(),
+  warnings: import_zod31.z.array(import_zod31.z.unknown()).optional(),
+  sources: import_zod31.z.array(import_zod31.z.unknown()).optional(),
+  steps: import_zod31.z.array(import_zod31.z.unknown()),
+  generationTimeMs: import_zod31.z.number(),
+  prompt: import_zod31.z.string(),
+  systemPrompt: import_zod31.z.string(),
+  usage: import_zod31.z.object({
+    totalTokens: import_zod31.z.number().optional(),
+    totalMicrocentsSpent: import_zod31.z.number().optional()
   })
 });
-var EvaluationResultSchema = import_zod30.z.object({
-  id: import_zod30.z.string(),
-  runId: import_zod30.z.string(),
-  timestamp: import_zod30.z.number(),
+var EvaluationResultSchema = import_zod31.z.object({
+  id: import_zod31.z.string(),
+  runId: import_zod31.z.string(),
+  timestamp: import_zod31.z.number(),
   promptResult: PromptResultSchema,
-  testResults: import_zod30.z.array(import_zod30.z.unknown()),
-  tags: import_zod30.z.array(import_zod30.z.string()).optional(),
-  feedback: import_zod30.z.string().optional(),
-  score: import_zod30.z.number(),
-  suiteId: import_zod30.z.string().optional()
-});
-var LeanEvaluationResultSchema = import_zod30.z.object({
-  id: import_zod30.z.string(),
-  runId: import_zod30.z.string(),
-  timestamp: import_zod30.z.number(),
-  tags: import_zod30.z.array(import_zod30.z.string()).optional(),
-  scenarioId: import_zod30.z.string(),
-  scenarioVersion: import_zod30.z.number().optional(),
-  targetId: import_zod30.z.string(),
-  targetVersion: import_zod30.z.number().optional(),
-  suiteId: import_zod30.z.string().optional(),
-  score: import_zod30.z.number(),
-  time: import_zod30.z.number().optional(),
-  microcentsSpent: import_zod30.z.number().optional()
+  testResults: import_zod31.z.array(import_zod31.z.unknown()),
+  tags: import_zod31.z.array(import_zod31.z.string()).optional(),
+  feedback: import_zod31.z.string().optional(),
+  score: import_zod31.z.number(),
+  suiteId: import_zod31.z.string().optional()
+});
+var LeanEvaluationResultSchema = import_zod31.z.object({
+  id: import_zod31.z.string(),
+  runId: import_zod31.z.string(),
+  timestamp: import_zod31.z.number(),
+  tags: import_zod31.z.array(import_zod31.z.string()).optional(),
+  scenarioId: import_zod31.z.string(),
+  scenarioVersion: import_zod31.z.number().optional(),
+  targetId: import_zod31.z.string(),
+  targetVersion: import_zod31.z.number().optional(),
+  suiteId: import_zod31.z.string().optional(),
+  score: import_zod31.z.number(),
+  time: import_zod31.z.number().optional(),
+  microcentsSpent: import_zod31.z.number().optional()
 });
 
 // src/evaluation/eval-run-folder.ts
-var import_zod31 = require("zod");
+var import_zod32 = require("zod");
 var EvalRunFolderSchema = TenantEntitySchema.extend({});
 var CreateEvalRunFolderInputSchema = EvalRunFolderSchema.omit({
   id: true,
@@ -1587,26 +1883,26 @@ var UpdateEvalRunFolderInputSchema = EvalRunFolderSchema.omit({
   updatedAt: true,
   deleted: true
 }).partial();
-var EvalRunFolderMembershipSchema = import_zod31.z.object({
-  folderId: import_zod31.z.string(),
-  evalRunId: import_zod31.z.string(),
-  projectId: import_zod31.z.string(),
-  createdAt: import_zod31.z.string()
+var EvalRunFolderMembershipSchema = import_zod32.z.object({
+  folderId: import_zod32.z.string(),
+  evalRunId: import_zod32.z.string(),
+  projectId: import_zod32.z.string(),
+  createdAt: import_zod32.z.string()
 });
 
 // src/project/project.ts
-var import_zod32 = require("zod");
+var import_zod33 = require("zod");
 var ProjectSchema = BaseEntitySchema.extend({
-  appId: import_zod32.z.string().optional().describe("The ID of the app in Dev Center"),
-  scenarioTags: import_zod32.z.array(import_zod32.z.string()).optional().describe("Project-level tag vocabulary for scenarios"),
+  appId: import_zod33.z.string().optional().describe("The ID of the app in Dev Center"),
+  scenarioTags: import_zod33.z.array(import_zod33.z.string()).optional().describe("Project-level tag vocabulary for scenarios"),
   /** Per-project Wix auth token (write-only — never returned in GET responses). null = clear. */
-  wixAuthToken: import_zod32.z.string().nullable().optional().describe("Wix auth token for CLI/MCP authentication (encrypted at rest)"),
+  wixAuthToken: import_zod33.z.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: import_zod32.z.string().nullable().optional().describe("Base64-encoded Base44 auth file content (encrypted at rest)"),
+  base44AuthFile: import_zod33.z.string().nullable().optional().describe("Base64-encoded Base44 auth file content (encrypted at rest)"),
   /** Resolved at runtime from the encrypted Wix auth token */
-  wixAuthEmail: import_zod32.z.string().optional().describe("Email associated with the Wix auth token (resolved at runtime)"),
+  wixAuthEmail: import_zod33.z.string().optional().describe("Email associated with the Wix auth token (resolved at runtime)"),
   /** Resolved at runtime from the encrypted Base44 auth file */
-  base44AuthEmail: import_zod32.z.string().optional().describe("Email from the Base44 auth file (resolved at runtime)")
+  base44AuthEmail: import_zod33.z.string().optional().describe("Email from the Base44 auth file (resolved at runtime)")
 });
 var CreateProjectInputSchema = ProjectSchema.omit({
   id: true,
@@ -1632,7 +1928,7 @@ var CreateTemplateInputSchema = TemplateSchema.omit({
 var UpdateTemplateInputSchema = CreateTemplateInputSchema.partial();
 
 // src/schedule/eval-schedule.ts
-var import_zod33 = require("zod");
+var import_zod34 = require("zod");
 var FrequencyType = /* @__PURE__ */ ((FrequencyType2) => {
   FrequencyType2["DAILY"] = "daily";
   FrequencyType2["WEEKDAY"] = "weekday";
@@ -1642,29 +1938,29 @@ var FrequencyType = /* @__PURE__ */ ((FrequencyType2) => {
 })(FrequencyType || {});
 var EvalScheduleSchema = TenantEntitySchema.extend({
   /** Whether the schedule is active */
-  enabled: import_zod33.z.boolean(),
+  enabled: import_zod34.z.boolean(),
   /** Test suite to run */
-  suiteId: import_zod33.z.string(),
+  suiteId: import_zod34.z.string(),
   /** Preset that provides agent + entities for this schedule */
-  presetId: import_zod33.z.string(),
+  presetId: import_zod34.z.string(),
   /** How often to run */
-  frequencyType: import_zod33.z.nativeEnum(FrequencyType),
+  frequencyType: import_zod34.z.nativeEnum(FrequencyType),
   /** Time of day in 24h format (HH:MM), hours 00-23, minutes 00-59 */
-  timeOfDay: import_zod33.z.string().regex(/^([01]\d|2[0-3]):[0-5]\d$/),
+  timeOfDay: import_zod34.z.string().regex(/^([01]\d|2[0-3]):[0-5]\d$/),
   /** Day of week (0=Sun, 6=Sat) for weekly schedules */
-  dayOfWeek: import_zod33.z.number().min(0).max(6).optional(),
+  dayOfWeek: import_zod34.z.number().min(0).max(6).optional(),
   /** Day of month (1-31) for monthly schedules */
-  dayOfMonth: import_zod33.z.number().min(1).max(31).optional(),
+  dayOfMonth: import_zod34.z.number().min(1).max(31).optional(),
   /** IANA timezone (e.g., 'America/New_York') */
-  timezone: import_zod33.z.string(),
+  timezone: import_zod34.z.string(),
   /** ID of the last eval run created by this schedule */
-  lastRunId: import_zod33.z.string().optional(),
+  lastRunId: import_zod34.z.string().optional(),
   /** Denormalized status of the last run */
-  lastRunStatus: import_zod33.z.string().optional(),
+  lastRunStatus: import_zod34.z.string().optional(),
   /** ISO timestamp of the last run */
-  lastRunAt: import_zod33.z.string().optional(),
+  lastRunAt: import_zod34.z.string().optional(),
   /** Next scheduled run time in UTC (pre-computed for efficient querying, set by backend) */
-  nextRunAt: import_zod33.z.string().optional()
+  nextRunAt: import_zod34.z.string().optional()
 });
 function isValidTimezone(tz) {
   try {
@@ -1677,14 +1973,14 @@ function isValidTimezone(tz) {
 function validateScheduleFields(data, ctx, options) {
   if (data.frequencyType === "weekly" /* WEEKLY */ && data.dayOfWeek == null) {
     ctx.addIssue({
-      code: import_zod33.z.ZodIssueCode.custom,
+      code: import_zod34.z.ZodIssueCode.custom,
       message: "dayOfWeek is required for weekly schedules",
       path: ["dayOfWeek"]
     });
   }
   if (data.frequencyType === "monthly" /* MONTHLY */ && data.dayOfMonth == null) {
     ctx.addIssue({
-      code: import_zod33.z.ZodIssueCode.custom,
+      code: import_zod34.z.ZodIssueCode.custom,
       message: "dayOfMonth is required for monthly schedules",
       path: ["dayOfMonth"]
     });
@@ -1692,7 +1988,7 @@ function validateScheduleFields(data, ctx, options) {
   const shouldValidateTz = options.partial ? data.timezone !== void 0 : true;
   if (shouldValidateTz && !isValidTimezone(data.timezone)) {
     ctx.addIssue({
-      code: import_zod33.z.ZodIssueCode.custom,
+      code: import_zod34.z.ZodIssueCode.custom,
       message: "Invalid IANA timezone",
       path: ["timezone"]
     });
@@ -1715,229 +2011,10 @@ 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];
-}
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   AGENT_TYPE_LABELS,
+  ALLOWED_BUILD_COMMANDS,
   ALL_AVAILABLE_MODEL_IDS,
   AVAILABLE_CLAUDE_MODEL_IDS,
   AVAILABLE_OPENAI_MODEL_IDS,
@@ -1971,6 +2048,7 @@ function getSystemAssertion(id) {
   BatchSummarySchema,
   BuildCheckTestSchema,
   BuildPassedAssertionSchema,
+  BuildPassedCommandStringSchema,
   BuildPassedConfigSchema,
   BulkImportResultItemSchema,
   BulkImportResultSchema,
@@ -1998,6 +2076,7 @@ function getSystemAssertion(id) {
   CreateTemplateInputSchema,
   CreateTestScenarioInputSchema,
   CreateTestSuiteInputSchema,
+  DEFAULT_BUILD_PASSED_COMMAND,
   DEFAULT_EVALUATOR_SYSTEM_PROMPT,
   DEFAULT_JUDGE_MODEL,
   DiffContentSchema,
@@ -2115,11 +2194,14 @@ function getSystemAssertion(id) {
   formatTraceEventLine,
   getSystemAssertion,
   getSystemAssertions,
+  isAllowedBuildCommandString,
   isSystemAssertionId,
   isValidSkillFolderName,
   normalizeBatchAssertionLink,
   normalizeModelId,
+  parseBuildCommandToArgv,
   parseTraceEventLine,
-  validateAssertionConfig
+  validateAssertionConfig,
+  validateBuildPassedParamsInAssertionLinks
 });
 //# sourceMappingURL=index.js.map