@wix/evalforge-types 0.73.0 → 0.75.0

package/build/index.js

@@ -60,6 +60,13 @@ __export(index_exports, {
   BulkImportResultItemSchema: () => BulkImportResultItemSchema,
   BulkImportResultSchema: () => BulkImportResultSchema,
   BulkImportSkillsInputSchema: () => BulkImportSkillsInputSchema,
+  CAPABILITY_NAME_REGEX: () => CAPABILITY_NAME_REGEX,
+  CapabilityContentSchema: () => CapabilityContentSchema,
+  CapabilitySchema: () => CapabilitySchema,
+  CapabilityTypeSchema: () => CapabilityTypeSchema,
+  CapabilityVersionOriginSchema: () => CapabilityVersionOriginSchema,
+  CapabilityVersionSchema: () => CapabilityVersionSchema,
+  CapabilityWithLatestVersionSchema: () => CapabilityWithLatestVersionSchema,
   ClaudeModel: () => ClaudeModel,
   ClaudeModelSchema: () => ClaudeModelSchema,
   CommandExecutionSchema: () => CommandExecutionSchema,
@@ -70,6 +77,8 @@ __export(index_exports, {
   CostAssertionSchema: () => CostAssertionSchema,
   CostConfigSchema: () => CostConfigSchema,
   CreateAgentInputSchema: () => CreateAgentInputSchema,
+  CreateCapabilityInputSchema: () => CreateCapabilityInputSchema,
+  CreateCapabilityVersionInputSchema: () => CreateCapabilityVersionInputSchema,
   CreateEvalRunFolderInputSchema: () => CreateEvalRunFolderInputSchema,
   CreateEvalRunInputSchema: () => CreateEvalRunInputSchema,
   CreateEvalScheduleInputSchema: () => CreateEvalScheduleInputSchema,
@@ -109,6 +118,7 @@ __export(index_exports, {
   FilePresenceTestSchema: () => FilePresenceTestSchema,
   FrequencyType: () => FrequencyType,
   GitHubSourceSchema: () => GitHubSourceSchema,
+  InitialCapabilityVersionInputSchema: () => InitialCapabilityVersionInputSchema,
   InitialVersionInputSchema: () => InitialVersionInputSchema,
   LEGACY_MODEL_ID_MAP: () => LEGACY_MODEL_ID_MAP,
   LLMBreakdownStatsSchema: () => LLMBreakdownStatsSchema,
@@ -185,6 +195,7 @@ __export(index_exports, {
   TriggerSchema: () => TriggerSchema,
   TriggerType: () => TriggerType,
   UpdateAgentInputSchema: () => UpdateAgentInputSchema,
+  UpdateCapabilityInputSchema: () => UpdateCapabilityInputSchema,
   UpdateEvalRunFolderInputSchema: () => UpdateEvalRunFolderInputSchema,
   UpdateEvalScheduleInputSchema: () => UpdateEvalScheduleInputSchema,
   UpdateMcpInputSchema: () => UpdateMcpInputSchema,
@@ -197,12 +208,20 @@ __export(index_exports, {
   UpdateTestScenarioInputSchema: () => UpdateTestScenarioInputSchema,
   UpdateTestSuiteInputSchema: () => UpdateTestSuiteInputSchema,
   VitestTestSchema: () => VitestTestSchema,
+  capabilityToMcp: () => capabilityToMcp,
+  capabilityToRule: () => capabilityToRule,
+  capabilityToSkill: () => capabilityToSkill,
+  capabilityToSkillWithLatestVersion: () => capabilityToSkillWithLatestVersion,
+  capabilityToSubAgent: () => capabilityToSubAgent,
+  capabilityVersionToSkillVersion: () => capabilityVersionToSkillVersion,
   classifyAssertionRef: () => classifyAssertionRef,
   formatTraceEventLine: () => formatTraceEventLine,
   getSystemAssertion: () => getSystemAssertion,
   getSystemAssertions: () => getSystemAssertions,
+  groupCapabilitiesByType: () => groupCapabilitiesByType,
   isAllowedBuildCommandString: () => isAllowedBuildCommandString,
   isSystemAssertionId: () => isSystemAssertionId,
+  isValidCapabilityName: () => isValidCapabilityName,
   isValidSkillFolderName: () => isValidSkillFolderName,
   normalizeBatchAssertionLink: () => normalizeBatchAssertionLink,
   normalizeModelId: () => normalizeModelId,
@@ -564,10 +583,14 @@ var PresetSchema = TenantEntitySchema.extend({
   /** Sub-agent IDs included in this preset */
   subAgentIds: import_zod9.z.array(import_zod9.z.string()).default([]),
   /** Rule IDs included in this preset */
-  ruleIds: import_zod9.z.array(import_zod9.z.string()).default([])
-});
-var atLeastOneEntity = (data) => (data.skillIds?.length ?? 0) > 0 || (data.mcpIds?.length ?? 0) > 0 || (data.subAgentIds?.length ?? 0) > 0 || (data.ruleIds?.length ?? 0) > 0;
-var AT_LEAST_ONE_ENTITY_MESSAGE = "At least one of skillIds, mcpIds, subAgentIds, or ruleIds must be non-empty";
+  ruleIds: import_zod9.z.array(import_zod9.z.string()).default([]),
+  /** Unified capability IDs (replaces skill/mcp/subAgent/ruleIds) */
+  capabilityIds: import_zod9.z.array(import_zod9.z.string()).optional(),
+  /** Map of capabilityId to capabilityVersionId for version pinning */
+  capabilityVersions: import_zod9.z.record(import_zod9.z.string(), import_zod9.z.string()).optional()
+});
+var atLeastOneEntity = (data) => (data.capabilityIds?.length ?? 0) > 0 || (data.skillIds?.length ?? 0) > 0 || (data.mcpIds?.length ?? 0) > 0 || (data.subAgentIds?.length ?? 0) > 0 || (data.ruleIds?.length ?? 0) > 0;
+var AT_LEAST_ONE_ENTITY_MESSAGE = "At least one of capabilityIds, skillIds, mcpIds, subAgentIds, or ruleIds must be non-empty";
 var CreatePresetInputSchema = PresetSchema.omit({
   id: true,
   createdAt: true,
@@ -581,11 +604,179 @@ var UpdatePresetInputSchema = PresetSchema.omit({
   deleted: true
 }).partial();
 
+// src/target/capability.ts
+var import_zod10 = require("zod");
+var CapabilityTypeSchema = import_zod10.z.enum([
+  "SKILL",
+  "SUB_AGENT",
+  "RULE",
+  "MCP"
+]);
+var CAPABILITY_NAME_REGEX = /^[a-z0-9]+(-[a-z0-9]+)*$/;
+function isValidCapabilityName(name) {
+  return typeof name === "string" && name.length > 0 && CAPABILITY_NAME_REGEX.test(name);
+}
+var KEBAB_CASE_MESSAGE2 = "Name must be in kebab-case (lowercase letters, numbers, hyphens only, e.g. my-capability)";
+var CapabilityContentSchema = import_zod10.z.record(import_zod10.z.string(), import_zod10.z.unknown());
+var CapabilityVersionOriginSchema = import_zod10.z.enum(["manual", "pr", "master"]);
+var CapabilitySchema = TenantEntitySchema.extend({
+  capabilityType: CapabilityTypeSchema,
+  source: GitHubSourceSchema.optional()
+});
+var CapabilityVersionSchema = import_zod10.z.object({
+  id: import_zod10.z.string(),
+  projectId: import_zod10.z.string(),
+  capabilityId: import_zod10.z.string(),
+  version: import_zod10.z.string(),
+  origin: CapabilityVersionOriginSchema,
+  source: GitHubSourceSchema.optional(),
+  content: CapabilityContentSchema.optional(),
+  notes: import_zod10.z.string().optional(),
+  createdAt: import_zod10.z.string()
+});
+var CapabilityWithLatestVersionSchema = CapabilitySchema.extend({
+  latestVersion: CapabilityVersionSchema.optional()
+});
+var CapabilityInputBaseSchema = CapabilitySchema.omit({
+  id: true,
+  createdAt: true,
+  updatedAt: true,
+  deleted: true,
+  description: true,
+  source: true
+}).extend({
+  description: import_zod10.z.string().optional(),
+  source: GitHubSourceSchema.optional()
+});
+var InitialCapabilityVersionInputSchema = import_zod10.z.object({
+  content: CapabilityContentSchema.optional(),
+  notes: import_zod10.z.string().optional(),
+  source: GitHubSourceSchema.optional(),
+  version: import_zod10.z.string().optional(),
+  origin: CapabilityVersionOriginSchema.optional()
+});
+var CreateCapabilityInputSchema = CapabilityInputBaseSchema.extend({
+  initialVersion: InitialCapabilityVersionInputSchema.optional()
+}).refine((data) => isValidCapabilityName(data.name), {
+  message: KEBAB_CASE_MESSAGE2,
+  path: ["name"]
+});
+var UpdateCapabilityInputSchema = CapabilityInputBaseSchema.omit({
+  capabilityType: true
+}).partial().refine(
+  (data) => data.name === void 0 || isValidCapabilityName(data.name),
+  { message: KEBAB_CASE_MESSAGE2, path: ["name"] }
+);
+var CreateCapabilityVersionInputSchema = import_zod10.z.object({
+  source: GitHubSourceSchema.optional(),
+  version: import_zod10.z.string().min(1),
+  notes: import_zod10.z.string().optional(),
+  origin: CapabilityVersionOriginSchema.optional(),
+  content: CapabilityContentSchema.optional()
+});
+
+// src/target/capability-converters.ts
+function capabilityToSkill(cap) {
+  return {
+    id: cap.id,
+    projectId: cap.projectId,
+    name: cap.name,
+    description: cap.description,
+    source: cap.source,
+    createdAt: cap.createdAt,
+    updatedAt: cap.updatedAt,
+    deleted: cap.deleted
+  };
+}
+function capabilityVersionToSkillVersion(cv) {
+  const content = cv.content;
+  return {
+    id: cv.id,
+    projectId: cv.projectId,
+    skillId: cv.capabilityId,
+    version: cv.version,
+    origin: cv.origin,
+    source: cv.source,
+    files: content?.files,
+    notes: cv.notes,
+    createdAt: cv.createdAt
+  };
+}
+function capabilityToSkillWithLatestVersion(cap) {
+  const skill = capabilityToSkill(cap);
+  const latestVersion = cap.latestVersion ? capabilityVersionToSkillVersion(cap.latestVersion) : void 0;
+  return { ...skill, latestVersion };
+}
+function capabilityToSubAgent(cap) {
+  const content = cap.latestVersion?.content;
+  return {
+    id: cap.id,
+    projectId: cap.projectId,
+    name: cap.name,
+    description: cap.description,
+    subAgentMd: content?.subAgentMd ?? "",
+    source: cap.source,
+    createdAt: cap.createdAt,
+    updatedAt: cap.updatedAt,
+    deleted: cap.deleted
+  };
+}
+function capabilityToRule(cap) {
+  const content = cap.latestVersion?.content;
+  return {
+    id: cap.id,
+    projectId: cap.projectId,
+    name: cap.name,
+    description: cap.description,
+    ruleType: content?.ruleType ?? "claude-md",
+    content: content?.content ?? "",
+    createdAt: cap.createdAt,
+    updatedAt: cap.updatedAt,
+    deleted: cap.deleted
+  };
+}
+function capabilityToMcp(cap) {
+  const content = cap.latestVersion?.content;
+  return {
+    id: cap.id,
+    projectId: cap.projectId,
+    name: cap.name,
+    description: cap.description,
+    config: content?.config ?? {},
+    createdAt: cap.createdAt,
+    updatedAt: cap.updatedAt,
+    deleted: cap.deleted
+  };
+}
+function groupCapabilitiesByType(capabilities) {
+  const skills = [];
+  const subAgents = [];
+  const rules = [];
+  const mcps = [];
+  for (const cap of capabilities) {
+    switch (cap.capabilityType) {
+      case "SKILL":
+        skills.push(capabilityToSkillWithLatestVersion(cap));
+        break;
+      case "SUB_AGENT":
+        subAgents.push(capabilityToSubAgent(cap));
+        break;
+      case "RULE":
+        rules.push(capabilityToRule(cap));
+        break;
+      case "MCP":
+        mcps.push(capabilityToMcp(cap));
+        break;
+    }
+  }
+  return { skills, subAgents, rules, mcps };
+}
+
 // src/test/index.ts
-var import_zod20 = require("zod");
+var import_zod21 = require("zod");
 
 // src/test/base.ts
-var import_zod10 = require("zod");
+var import_zod11 = require("zod");
 var TestType = /* @__PURE__ */ ((TestType2) => {
   TestType2["LLM"] = "LLM";
   TestType2["TOOL"] = "TOOL";
@@ -598,7 +789,7 @@ var TestType = /* @__PURE__ */ ((TestType2) => {
   TestType2["PLAYWRIGHT_NL"] = "PLAYWRIGHT_NL";
   return TestType2;
 })(TestType || {});
-var TestTypeSchema = import_zod10.z.enum(TestType);
+var TestTypeSchema = import_zod11.z.enum(TestType);
 var TestImportance = /* @__PURE__ */ ((TestImportance2) => {
   TestImportance2["LOW"] = "low";
   TestImportance2["MEDIUM"] = "medium";
@@ -606,153 +797,153 @@ var TestImportance = /* @__PURE__ */ ((TestImportance2) => {
   TestImportance2["CRITICAL"] = "critical";
   return TestImportance2;
 })(TestImportance || {});
-var TestImportanceSchema = import_zod10.z.enum(TestImportance);
-var BaseTestSchema = import_zod10.z.object({
-  id: import_zod10.z.string(),
+var TestImportanceSchema = import_zod11.z.enum(TestImportance);
+var BaseTestSchema = import_zod11.z.object({
+  id: import_zod11.z.string(),
   type: TestTypeSchema,
-  name: import_zod10.z.string().min(3),
-  description: import_zod10.z.string().optional(),
+  name: import_zod11.z.string().min(3),
+  description: import_zod11.z.string().optional(),
   importance: TestImportanceSchema.optional()
 });
 
 // src/test/llm.ts
-var import_zod11 = require("zod");
+var import_zod12 = require("zod");
 var LLMTestSchema = BaseTestSchema.extend({
-  type: import_zod11.z.literal("LLM" /* LLM */),
+  type: import_zod12.z.literal("LLM" /* LLM */),
   /** Maximum steps for the LLM to take */
-  maxSteps: import_zod11.z.number().min(1).max(100),
+  maxSteps: import_zod12.z.number().min(1).max(100),
   /** Prompt to send to the evaluator */
-  prompt: import_zod11.z.string().min(1),
+  prompt: import_zod12.z.string().min(1),
   /** ID of the evaluator agent to use */
-  evaluatorId: import_zod11.z.string()
+  evaluatorId: import_zod12.z.string()
 });
 
 // src/test/tool.ts
-var import_zod12 = require("zod");
+var import_zod13 = require("zod");
 var ToolTestSchema = BaseTestSchema.extend({
-  type: import_zod12.z.literal("TOOL" /* TOOL */),
+  type: import_zod13.z.literal("TOOL" /* TOOL */),
   /** Name of the tool that should be called */
-  toolName: import_zod12.z.string().min(3),
+  toolName: import_zod13.z.string().min(3),
   /** Expected arguments for the tool call */
-  args: import_zod12.z.record(import_zod12.z.string(), import_zod12.z.any()),
+  args: import_zod13.z.record(import_zod13.z.string(), import_zod13.z.any()),
   /** Expected content in the tool results */
-  resultsContent: import_zod12.z.string()
+  resultsContent: import_zod13.z.string()
 });
 
 // src/test/site-config.ts
-var import_zod13 = require("zod");
+var import_zod14 = require("zod");
 var SiteConfigTestSchema = BaseTestSchema.extend({
-  type: import_zod13.z.literal("SITE_CONFIG" /* SITE_CONFIG */),
+  type: import_zod14.z.literal("SITE_CONFIG" /* SITE_CONFIG */),
   /** URL to call */
-  url: import_zod13.z.string().url(),
+  url: import_zod14.z.string().url(),
   /** HTTP method */
-  method: import_zod13.z.enum(["GET", "POST"]),
+  method: import_zod14.z.enum(["GET", "POST"]),
   /** Request body (for POST) */
-  body: import_zod13.z.string().optional(),
+  body: import_zod14.z.string().optional(),
   /** Expected HTTP status code */
-  expectedStatusCode: import_zod13.z.number().int().min(100).max(599),
+  expectedStatusCode: import_zod14.z.number().int().min(100).max(599),
   /** Expected response content */
-  expectedResponse: import_zod13.z.string().optional(),
+  expectedResponse: import_zod14.z.string().optional(),
   /** JMESPath expression to extract from response */
-  expectedResponseJMESPath: import_zod13.z.string().optional()
+  expectedResponseJMESPath: import_zod14.z.string().optional()
 });
 
 // src/test/command-execution.ts
-var import_zod14 = require("zod");
+var import_zod15 = require("zod");
 var AllowedCommands = [
   "yarn install --no-immutable && yarn build",
   "npm run build",
   "yarn typecheck"
 ];
 var CommandExecutionTestSchema = BaseTestSchema.extend({
-  type: import_zod14.z.literal("COMMAND_EXECUTION" /* COMMAND_EXECUTION */),
+  type: import_zod15.z.literal("COMMAND_EXECUTION" /* COMMAND_EXECUTION */),
   /** Command to execute (must be in AllowedCommands) */
-  command: import_zod14.z.string().refine((value) => AllowedCommands.includes(value), {
+  command: import_zod15.z.string().refine((value) => AllowedCommands.includes(value), {
     message: `Command must be one of: ${AllowedCommands.join(", ")}`
   }),
   /** Expected exit code (default: 0) */
-  expectedExitCode: import_zod14.z.number().default(0).optional()
+  expectedExitCode: import_zod15.z.number().default(0).optional()
 });
 
 // src/test/file-presence.ts
-var import_zod15 = require("zod");
+var import_zod16 = require("zod");
 var FilePresenceTestSchema = BaseTestSchema.extend({
-  type: import_zod15.z.literal("FILE_PRESENCE" /* FILE_PRESENCE */),
+  type: import_zod16.z.literal("FILE_PRESENCE" /* FILE_PRESENCE */),
   /** Paths to check */
-  paths: import_zod15.z.array(import_zod15.z.string()),
+  paths: import_zod16.z.array(import_zod16.z.string()),
   /** Whether files should exist (true) or not exist (false) */
-  shouldExist: import_zod15.z.boolean()
+  shouldExist: import_zod16.z.boolean()
 });
 
 // src/test/file-content.ts
-var import_zod16 = require("zod");
-var FileContentCheckSchema = import_zod16.z.object({
+var import_zod17 = require("zod");
+var FileContentCheckSchema = import_zod17.z.object({
   /** Strings that must be present in the file */
-  contains: import_zod16.z.array(import_zod16.z.string()).optional(),
+  contains: import_zod17.z.array(import_zod17.z.string()).optional(),
   /** Strings that must NOT be present in the file */
-  notContains: import_zod16.z.array(import_zod16.z.string()).optional(),
+  notContains: import_zod17.z.array(import_zod17.z.string()).optional(),
   /** Regex pattern the content must match */
-  matches: import_zod16.z.string().optional(),
+  matches: import_zod17.z.string().optional(),
   /** JSON path checks for structured content */
-  jsonPath: import_zod16.z.array(
-    import_zod16.z.object({
-      path: import_zod16.z.string(),
-      value: import_zod16.z.unknown()
+  jsonPath: import_zod17.z.array(
+    import_zod17.z.object({
+      path: import_zod17.z.string(),
+      value: import_zod17.z.unknown()
     })
   ).optional(),
   /** Lines that should be added (for diff checking) */
-  added: import_zod16.z.array(import_zod16.z.string()).optional(),
+  added: import_zod17.z.array(import_zod17.z.string()).optional(),
   /** Lines that should be removed (for diff checking) */
-  removed: import_zod16.z.array(import_zod16.z.string()).optional()
+  removed: import_zod17.z.array(import_zod17.z.string()).optional()
 });
 var FileContentTestSchema = BaseTestSchema.extend({
-  type: import_zod16.z.literal("FILE_CONTENT" /* FILE_CONTENT */),
+  type: import_zod17.z.literal("FILE_CONTENT" /* FILE_CONTENT */),
   /** Path to the file to check */
-  path: import_zod16.z.string(),
+  path: import_zod17.z.string(),
   /** Content checks to perform */
   checks: FileContentCheckSchema
 });
 
 // src/test/build-check.ts
-var import_zod17 = require("zod");
+var import_zod18 = require("zod");
 var BuildCheckTestSchema = BaseTestSchema.extend({
-  type: import_zod17.z.literal("BUILD_CHECK" /* BUILD_CHECK */),
+  type: import_zod18.z.literal("BUILD_CHECK" /* BUILD_CHECK */),
   /** Build command to execute */
-  command: import_zod17.z.string(),
+  command: import_zod18.z.string(),
   /** Whether the build should succeed */
-  expectSuccess: import_zod17.z.boolean(),
+  expectSuccess: import_zod18.z.boolean(),
   /** Maximum allowed warnings (optional) */
-  allowedWarnings: import_zod17.z.number().optional(),
+  allowedWarnings: import_zod18.z.number().optional(),
   /** Timeout in milliseconds */
-  timeout: import_zod17.z.number().optional()
+  timeout: import_zod18.z.number().optional()
 });
 
 // src/test/vitest.ts
-var import_zod18 = require("zod");
+var import_zod19 = require("zod");
 var VitestTestSchema = BaseTestSchema.extend({
-  type: import_zod18.z.literal("VITEST" /* VITEST */),
+  type: import_zod19.z.literal("VITEST" /* VITEST */),
   /** Test file content */
-  testFile: import_zod18.z.string(),
+  testFile: import_zod19.z.string(),
   /** Name of the test file */
-  testFileName: import_zod18.z.string(),
+  testFileName: import_zod19.z.string(),
   /** Minimum pass rate required (0-100) */
-  minPassRate: import_zod18.z.number().min(0).max(100)
+  minPassRate: import_zod19.z.number().min(0).max(100)
 });
 
 // src/test/playwright-nl.ts
-var import_zod19 = require("zod");
+var import_zod20 = require("zod");
 var PlaywrightNLTestSchema = BaseTestSchema.extend({
-  type: import_zod19.z.literal("PLAYWRIGHT_NL" /* PLAYWRIGHT_NL */),
+  type: import_zod20.z.literal("PLAYWRIGHT_NL" /* PLAYWRIGHT_NL */),
   /** Natural language steps to execute */
-  steps: import_zod19.z.array(import_zod19.z.string()),
+  steps: import_zod20.z.array(import_zod20.z.string()),
   /** Expected outcome description */
-  expectedOutcome: import_zod19.z.string(),
+  expectedOutcome: import_zod20.z.string(),
   /** Timeout in milliseconds */
-  timeout: import_zod19.z.number().optional()
+  timeout: import_zod20.z.number().optional()
 });
 
 // src/test/index.ts
-var TestSchema = import_zod20.z.discriminatedUnion("type", [
+var TestSchema = import_zod21.z.discriminatedUnion("type", [
   LLMTestSchema,
   ToolTestSchema,
   SiteConfigTestSchema,
@@ -765,33 +956,33 @@ var TestSchema = import_zod20.z.discriminatedUnion("type", [
 ]);
 
 // src/scenario/environment.ts
-var import_zod21 = require("zod");
-var LocalProjectConfigSchema = import_zod21.z.object({
+var import_zod22 = require("zod");
+var LocalProjectConfigSchema = import_zod22.z.object({
   /** Template ID to use for the local project */
-  templateId: import_zod21.z.string().optional(),
+  templateId: import_zod22.z.string().optional(),
   /** Files to create in the project */
-  files: import_zod21.z.array(
-    import_zod21.z.object({
-      path: import_zod21.z.string().min(1),
-      content: import_zod21.z.string().min(1)
+  files: import_zod22.z.array(
+    import_zod22.z.object({
+      path: import_zod22.z.string().min(1),
+      content: import_zod22.z.string().min(1)
     })
   ).optional()
 });
-var MetaSiteConfigSchema = import_zod21.z.object({
-  configurations: import_zod21.z.array(
-    import_zod21.z.object({
-      name: import_zod21.z.string().min(1),
-      apiCalls: import_zod21.z.array(
-        import_zod21.z.object({
-          url: import_zod21.z.string().url(),
-          method: import_zod21.z.enum(["POST", "PUT"]),
-          body: import_zod21.z.string()
+var MetaSiteConfigSchema = import_zod22.z.object({
+  configurations: import_zod22.z.array(
+    import_zod22.z.object({
+      name: import_zod22.z.string().min(1),
+      apiCalls: import_zod22.z.array(
+        import_zod22.z.object({
+          url: import_zod22.z.string().url(),
+          method: import_zod22.z.enum(["POST", "PUT"]),
+          body: import_zod22.z.string()
         })
       )
     })
   ).optional()
 });
-var EnvironmentSchema = import_zod21.z.object({
+var EnvironmentSchema = import_zod22.z.object({
   /** Local project configuration */
   localProject: LocalProjectConfigSchema.optional(),
   /** Meta site configuration */
@@ -799,13 +990,13 @@ var EnvironmentSchema = import_zod21.z.object({
 });
 
 // src/scenario/test-scenario.ts
-var import_zod24 = require("zod");
+var import_zod25 = require("zod");
 
 // src/assertion/assertion.ts
-var import_zod23 = require("zod");
+var import_zod24 = require("zod");
 
 // src/assertion/build-passed-command.ts
-var import_zod22 = require("zod");
+var import_zod23 = require("zod");
 var ALLOWED_BUILD_COMMANDS = [
   "yarn build",
   "npm run build",
@@ -831,10 +1022,10 @@ function parseBuildCommandToArgv(command) {
   return BUILD_COMMAND_ARGV[trimmed];
 }
 var enumTuple = ALLOWED_BUILD_COMMANDS;
-var BuildPassedCommandStringSchema = import_zod22.z.enum(enumTuple);
+var BuildPassedCommandStringSchema = import_zod23.z.enum(enumTuple);
 
 // src/assertion/assertion.ts
-var AssertionTypeSchema = import_zod23.z.enum([
+var AssertionTypeSchema = import_zod24.z.enum([
   "skill_was_called",
   "tool_called_with_param",
   "build_passed",
@@ -843,61 +1034,61 @@ var AssertionTypeSchema = import_zod23.z.enum([
   "llm_judge",
   "api_call"
 ]);
-var AssertionParameterTypeSchema = import_zod23.z.enum([
+var AssertionParameterTypeSchema = import_zod24.z.enum([
   "string",
   "number",
   "boolean"
 ]);
-var AssertionParameterSchema = import_zod23.z.object({
+var AssertionParameterSchema = import_zod24.z.object({
   /** Parameter name (used as key in params object) */
-  name: import_zod23.z.string().min(1),
+  name: import_zod24.z.string().min(1),
   /** Display label for the parameter */
-  label: import_zod23.z.string().min(1),
+  label: import_zod24.z.string().min(1),
   /** Parameter type */
   type: AssertionParameterTypeSchema,
   /** Whether this parameter is required */
-  required: import_zod23.z.boolean(),
+  required: import_zod24.z.boolean(),
   /** Default value (optional, used when not provided) */
-  defaultValue: import_zod23.z.union([import_zod23.z.string(), import_zod23.z.number(), import_zod23.z.boolean()]).optional(),
+  defaultValue: import_zod24.z.union([import_zod24.z.string(), import_zod24.z.number(), import_zod24.z.boolean()]).optional(),
   /** If true, parameter is hidden by default behind "Show advanced options" */
-  advanced: import_zod23.z.boolean().optional()
+  advanced: import_zod24.z.boolean().optional()
 });
-var ScenarioAssertionLinkSchema = import_zod23.z.object({
+var ScenarioAssertionLinkSchema = import_zod24.z.object({
   /** ID of the system assertion (e.g., 'system:skill_was_called') */
-  assertionId: import_zod23.z.string(),
+  assertionId: import_zod24.z.string(),
   /** Parameter values for this assertion in this scenario */
-  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()])
+  params: import_zod24.z.record(
+    import_zod24.z.string(),
+    import_zod24.z.union([import_zod24.z.string(), import_zod24.z.number(), import_zod24.z.boolean(), import_zod24.z.null()])
   ).optional()
 });
-var SkillWasCalledConfigSchema = import_zod23.z.object({
+var SkillWasCalledConfigSchema = import_zod24.z.object({
   /** Names of the skills that must have been called */
-  skillNames: import_zod23.z.array(import_zod23.z.string().min(1)).min(1)
+  skillNames: import_zod24.z.array(import_zod24.z.string().min(1)).min(1)
 });
-var CostConfigSchema = import_zod23.z.strictObject({
+var CostConfigSchema = import_zod24.z.strictObject({
   /** Maximum allowed cost in USD */
-  maxCostUsd: import_zod23.z.number().positive()
+  maxCostUsd: import_zod24.z.number().positive()
 });
-var ToolCalledWithParamConfigSchema = import_zod23.z.strictObject({
+var ToolCalledWithParamConfigSchema = import_zod24.z.strictObject({
   /** Name of the tool that must have been called */
-  toolName: import_zod23.z.string().min(1),
+  toolName: import_zod24.z.string().min(1),
   /** JSON string of key-value pairs for expected parameters (substring match). Optional — when omitted, only checks tool presence. */
-  expectedParams: import_zod23.z.string().min(1).optional(),
+  expectedParams: import_zod24.z.string().min(1).optional(),
   /** If true, the matching tool call must also have succeeded (step.success === true) */
-  requireSuccess: import_zod23.z.boolean().optional()
+  requireSuccess: import_zod24.z.boolean().optional()
 });
-var BuildPassedConfigSchema = import_zod23.z.strictObject({
+var BuildPassedConfigSchema = import_zod24.z.strictObject({
   /** Allowlisted command only (default at runtime: "yarn build") */
   command: BuildPassedCommandStringSchema.optional(),
   /** Expected exit code (default: 0) */
-  expectedExitCode: import_zod23.z.number().int().optional()
+  expectedExitCode: import_zod24.z.number().int().optional()
 });
-var TimeConfigSchema = import_zod23.z.strictObject({
+var TimeConfigSchema = import_zod24.z.strictObject({
   /** Maximum allowed duration in milliseconds */
-  maxDurationMs: import_zod23.z.number().int().positive()
+  maxDurationMs: import_zod24.z.number().int().positive()
 });
-var LlmJudgeConfigSchema = import_zod23.z.object({
+var LlmJudgeConfigSchema = import_zod24.z.object({
   /**
    * Prompt template with placeholders:
    * - {{output}}: agent's final output
@@ -908,65 +1099,65 @@ var LlmJudgeConfigSchema = import_zod23.z.object({
    * - {{trace}}: step-by-step trace of tool calls
    * - Custom parameters defined in the parameters array
    */
-  prompt: import_zod23.z.string().min(1),
+  prompt: import_zod24.z.string().min(1),
   /** Minimum score to pass (0-10, default 7) */
-  minScore: import_zod23.z.number().int().min(0).max(10).optional(),
+  minScore: import_zod24.z.number().int().min(0).max(10).optional(),
   /** Model for the judge (e.g. claude-3-5-haiku-20241022) */
-  model: import_zod23.z.string().optional(),
+  model: import_zod24.z.string().optional(),
   /** Max output tokens */
-  maxTokens: import_zod23.z.number().int().optional(),
+  maxTokens: import_zod24.z.number().int().optional(),
   /** Temperature (0-1) */
-  temperature: import_zod23.z.number().min(0).max(1).optional(),
+  temperature: import_zod24.z.number().min(0).max(1).optional(),
   /** User-defined parameters for this assertion */
-  parameters: import_zod23.z.array(AssertionParameterSchema).optional()
+  parameters: import_zod24.z.array(AssertionParameterSchema).optional()
 });
-var ApiCallConfigSchema = import_zod23.z.strictObject({
+var ApiCallConfigSchema = import_zod24.z.strictObject({
   /** URL to call */
-  url: import_zod23.z.string().min(1),
+  url: import_zod24.z.string().min(1),
   /** HTTP method (default GET) */
-  method: import_zod23.z.enum(["GET", "POST"]).optional(),
+  method: import_zod24.z.enum(["GET", "POST"]).optional(),
   /** Request body (JSON string, for POST requests) */
-  requestBody: import_zod23.z.string().optional(),
+  requestBody: import_zod24.z.string().optional(),
   /** Expected JSON response to validate against (subset match — extra fields in actual are OK) */
-  expectedResponse: import_zod23.z.string().min(1),
+  expectedResponse: import_zod24.z.string().min(1),
   /** Request headers as JSON string of key-value pairs */
-  requestHeaders: import_zod23.z.string().optional(),
+  requestHeaders: import_zod24.z.string().optional(),
   /** Request timeout in milliseconds (default 30000) */
-  timeoutMs: import_zod23.z.number().int().positive().optional()
+  timeoutMs: import_zod24.z.number().int().positive().optional()
 });
 var AssertionBaseFields = {
   /** When true, the assertion's pass/fail logic is inverted (NOT operator). */
-  negate: import_zod23.z.boolean().optional()
+  negate: import_zod24.z.boolean().optional()
 };
 var SkillWasCalledAssertionSchema = SkillWasCalledConfigSchema.extend({
-  type: import_zod23.z.literal("skill_was_called"),
+  type: import_zod24.z.literal("skill_was_called"),
   ...AssertionBaseFields
 });
 var ToolCalledWithParamAssertionSchema = ToolCalledWithParamConfigSchema.extend({
-  type: import_zod23.z.literal("tool_called_with_param"),
+  type: import_zod24.z.literal("tool_called_with_param"),
   ...AssertionBaseFields
 });
 var BuildPassedAssertionSchema = BuildPassedConfigSchema.extend({
-  type: import_zod23.z.literal("build_passed"),
+  type: import_zod24.z.literal("build_passed"),
   ...AssertionBaseFields
 });
 var CostAssertionSchema = CostConfigSchema.extend({
-  type: import_zod23.z.literal("cost"),
+  type: import_zod24.z.literal("cost"),
   ...AssertionBaseFields
 });
 var LlmJudgeAssertionSchema = LlmJudgeConfigSchema.extend({
-  type: import_zod23.z.literal("llm_judge"),
+  type: import_zod24.z.literal("llm_judge"),
   ...AssertionBaseFields
 });
 var ApiCallAssertionSchema = ApiCallConfigSchema.extend({
-  type: import_zod23.z.literal("api_call"),
+  type: import_zod24.z.literal("api_call"),
   ...AssertionBaseFields
 });
 var TimeAssertionSchema = TimeConfigSchema.extend({
-  type: import_zod23.z.literal("time_limit"),
+  type: import_zod24.z.literal("time_limit"),
   ...AssertionBaseFields
 });
-var AssertionSchema = import_zod23.z.union([
+var AssertionSchema = import_zod24.z.union([
   SkillWasCalledAssertionSchema,
   ToolCalledWithParamAssertionSchema,
   BuildPassedAssertionSchema,
@@ -975,7 +1166,7 @@ var AssertionSchema = import_zod23.z.union([
   LlmJudgeAssertionSchema,
   ApiCallAssertionSchema
 ]);
-var AssertionConfigSchema = import_zod23.z.union([
+var AssertionConfigSchema = import_zod24.z.union([
   LlmJudgeConfigSchema,
   // requires prompt - check first
   SkillWasCalledConfigSchema,
@@ -990,7 +1181,7 @@ var AssertionConfigSchema = import_zod23.z.union([
   // requires maxCostUsd, uses strictObject
   BuildPassedConfigSchema,
   // all optional, uses strictObject to reject unknown keys
-  import_zod23.z.object({})
+  import_zod24.z.object({})
   // fallback empty config
 ]);
 function validateAssertionConfig(type, config) {
@@ -1236,35 +1427,35 @@ function getSystemAssertion(id) {
 
 // src/scenario/test-scenario.ts
 var MAX_IMAGE_BASE64_LENGTH = 4 * Math.ceil(2 * 1024 * 1024 / 3);
-var TriggerPromptImageSchema = import_zod24.z.object({
+var TriggerPromptImageSchema = import_zod25.z.object({
   /** Base64-encoded image data (no data URL prefix) */
-  base64: import_zod24.z.string().max(MAX_IMAGE_BASE64_LENGTH, "Image exceeds 2 MB size limit"),
+  base64: import_zod25.z.string().max(MAX_IMAGE_BASE64_LENGTH, "Image exceeds 2 MB size limit"),
   /** MIME type of the image */
-  mediaType: import_zod24.z.enum(["image/jpeg", "image/png", "image/gif", "image/webp"]),
+  mediaType: import_zod25.z.enum(["image/jpeg", "image/png", "image/gif", "image/webp"]),
   /** Original filename of the image */
-  name: import_zod24.z.string()
+  name: import_zod25.z.string()
 });
-var ExpectedFileSchema = import_zod24.z.object({
+var ExpectedFileSchema = import_zod25.z.object({
   /** Relative path where the file should be created */
-  path: import_zod24.z.string(),
+  path: import_zod25.z.string(),
   /** Optional expected content */
-  content: import_zod24.z.string().optional()
+  content: import_zod25.z.string().optional()
 });
 var TestScenarioSchema = TenantEntitySchema.extend({
   /** The prompt sent to the agent to trigger the task */
-  triggerPrompt: import_zod24.z.string().min(10),
+  triggerPrompt: import_zod25.z.string().min(10),
   /** ID of the template to use for this scenario (null = no template) */
-  templateId: import_zod24.z.string().nullish(),
+  templateId: import_zod25.z.string().nullish(),
   /** Inline assertions to evaluate for this scenario (legacy) */
-  assertions: import_zod24.z.array(AssertionSchema).optional(),
+  assertions: import_zod25.z.array(AssertionSchema).optional(),
   /** IDs of saved assertions to evaluate (from assertions table) - legacy, use assertionLinks */
-  assertionIds: import_zod24.z.array(import_zod24.z.string()).optional(),
+  assertionIds: import_zod25.z.array(import_zod25.z.string()).optional(),
   /** Linked assertions with per-scenario parameter values */
-  assertionLinks: import_zod24.z.array(ScenarioAssertionLinkSchema).optional(),
+  assertionLinks: import_zod25.z.array(ScenarioAssertionLinkSchema).optional(),
   /** Tags for categorisation and filtering */
-  tags: import_zod24.z.array(import_zod24.z.string()).optional(),
+  tags: import_zod25.z.array(import_zod25.z.string()).optional(),
   /** Base64-encoded images attached to the trigger prompt (max 3) */
-  triggerPromptImages: import_zod24.z.array(TriggerPromptImageSchema).max(3).optional()
+  triggerPromptImages: import_zod25.z.array(TriggerPromptImageSchema).max(3).optional()
 });
 function validateBuildPassedParamsInAssertionLinks(links, ctx) {
   if (!links) return;
@@ -1275,7 +1466,7 @@ function validateBuildPassedParamsInAssertionLinks(links, ctx) {
     if (cmd === void 0 || cmd === null) continue;
     if (typeof cmd !== "string") {
       ctx.addIssue({
-        code: import_zod24.z.ZodIssueCode.custom,
+        code: import_zod25.z.ZodIssueCode.custom,
         message: "build_passed command must be a string",
         path: ["assertionLinks", i, "params", "command"]
       });
@@ -1283,7 +1474,7 @@ function validateBuildPassedParamsInAssertionLinks(links, ctx) {
     }
     if (!isAllowedBuildCommandString(cmd)) {
       ctx.addIssue({
-        code: import_zod24.z.ZodIssueCode.custom,
+        code: import_zod25.z.ZodIssueCode.custom,
         message: "Invalid build_passed command. Allowed: yarn build, npm run build, pnpm run build, pnpm build",
         path: ["assertionLinks", i, "params", "command"]
       });
@@ -1306,19 +1497,19 @@ var UpdateTestScenarioInputSchema = TestScenarioCreateBaseSchema.partial().super
 });
 
 // src/scenario/batch-import.ts
-var import_zod25 = require("zod");
+var import_zod26 = 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_zod25.z.union([
-  import_zod25.z.string().min(1),
+var BatchAssertionLinkSchema = import_zod26.z.union([
+  import_zod26.z.string().min(1),
   ScenarioAssertionLinkSchema
 ]);
-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()
+var BatchScenarioEntrySchema = import_zod26.z.object({
+  name: import_zod26.z.string().min(1, "name: Required"),
+  description: import_zod26.z.string().optional().default(""),
+  triggerPrompt: import_zod26.z.string().min(10, "triggerPrompt: Must be at least 10 characters"),
+  templateId: import_zod26.z.string().nullish(),
+  tags: import_zod26.z.array(import_zod26.z.string()).optional(),
+  assertionLinks: import_zod26.z.array(BatchAssertionLinkSchema).optional()
 }).superRefine((data, ctx) => {
   if (!data.assertionLinks) return;
   const objectLinks = data.assertionLinks.filter(
@@ -1328,8 +1519,8 @@ var BatchScenarioEntrySchema = import_zod25.z.object({
     validateBuildPassedParamsInAssertionLinks(objectLinks, ctx);
   }
 });
-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 BatchImportPayloadSchema = import_zod26.z.object({
+  scenarios: import_zod26.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,
@@ -1351,29 +1542,29 @@ function normalizeBatchAssertionLink(link) {
   }
   return link;
 }
-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({
+var BatchResultItemSchema = import_zod26.z.object({
+  index: import_zod26.z.number(),
+  name: import_zod26.z.string(),
+  status: import_zod26.z.enum(["valid", "invalid"]),
+  id: import_zod26.z.string().nullable().optional(),
+  errors: import_zod26.z.array(import_zod26.z.string()).optional()
+});
+var BatchSummarySchema = import_zod26.z.object({
+  total: import_zod26.z.number(),
+  valid: import_zod26.z.number(),
+  invalid: import_zod26.z.number(),
+  created: import_zod26.z.number()
+});
+var BatchImportResponseSchema = import_zod26.z.object({
   summary: BatchSummarySchema,
-  results: import_zod25.z.array(BatchResultItemSchema)
+  results: import_zod26.z.array(BatchResultItemSchema)
 });
 
 // src/suite/test-suite.ts
-var import_zod26 = require("zod");
+var import_zod27 = require("zod");
 var TestSuiteSchema = TenantEntitySchema.extend({
   /** IDs of test scenarios in this suite */
-  scenarioIds: import_zod26.z.array(import_zod26.z.string())
+  scenarioIds: import_zod27.z.array(import_zod27.z.string())
 });
 var CreateTestSuiteInputSchema = TestSuiteSchema.omit({
   id: true,
@@ -1384,21 +1575,21 @@ var CreateTestSuiteInputSchema = TestSuiteSchema.omit({
 var UpdateTestSuiteInputSchema = CreateTestSuiteInputSchema.partial();
 
 // src/evaluation/metrics.ts
-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 import_zod28 = require("zod");
+var TokenUsageSchema = import_zod28.z.object({
+  prompt: import_zod28.z.number(),
+  completion: import_zod28.z.number(),
+  total: import_zod28.z.number()
+});
+var EvalMetricsSchema = import_zod28.z.object({
+  totalAssertions: import_zod28.z.number(),
+  passed: import_zod28.z.number(),
+  failed: import_zod28.z.number(),
+  skipped: import_zod28.z.number(),
+  errors: import_zod28.z.number(),
+  passRate: import_zod28.z.number(),
+  avgDuration: import_zod28.z.number(),
+  totalDuration: import_zod28.z.number()
 });
 var EvalStatus = /* @__PURE__ */ ((EvalStatus2) => {
   EvalStatus2["PENDING"] = "pending";
@@ -1408,7 +1599,7 @@ var EvalStatus = /* @__PURE__ */ ((EvalStatus2) => {
   EvalStatus2["CANCELLED"] = "cancelled";
   return EvalStatus2;
 })(EvalStatus || {});
-var EvalStatusSchema = import_zod27.z.enum(EvalStatus);
+var EvalStatusSchema = import_zod28.z.enum(EvalStatus);
 var LLMStepType = /* @__PURE__ */ ((LLMStepType2) => {
   LLMStepType2["COMPLETION"] = "completion";
   LLMStepType2["TOOL_USE"] = "tool_use";
@@ -1416,54 +1607,54 @@ var LLMStepType = /* @__PURE__ */ ((LLMStepType2) => {
   LLMStepType2["THINKING"] = "thinking";
   return LLMStepType2;
 })(LLMStepType || {});
-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(),
+var LLMTraceStepSchema = import_zod28.z.object({
+  id: import_zod28.z.string(),
+  stepNumber: import_zod28.z.number(),
+  type: import_zod28.z.enum(LLMStepType),
+  model: import_zod28.z.string(),
+  provider: import_zod28.z.string(),
+  startedAt: import_zod28.z.string(),
+  durationMs: import_zod28.z.number(),
   tokenUsage: TokenUsageSchema,
-  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(),
+  costUsd: import_zod28.z.number(),
+  toolName: import_zod28.z.string().optional(),
+  toolArguments: import_zod28.z.string().optional(),
+  inputPreview: import_zod28.z.string().optional(),
+  outputPreview: import_zod28.z.string().optional(),
+  success: import_zod28.z.boolean(),
+  error: import_zod28.z.string().optional(),
+  turnIndex: import_zod28.z.number().optional()
+});
+var LLMBreakdownStatsSchema = import_zod28.z.object({
+  count: import_zod28.z.number(),
+  durationMs: import_zod28.z.number(),
+  tokens: import_zod28.z.number(),
+  costUsd: import_zod28.z.number()
+});
+var LLMTraceSummarySchema = import_zod28.z.object({
+  totalSteps: import_zod28.z.number(),
+  totalTurns: import_zod28.z.number().optional(),
+  totalDurationMs: import_zod28.z.number(),
   totalTokens: TokenUsageSchema,
-  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),
+  totalCostUsd: import_zod28.z.number(),
+  stepTypeBreakdown: import_zod28.z.record(import_zod28.z.string(), LLMBreakdownStatsSchema).optional(),
+  modelBreakdown: import_zod28.z.record(import_zod28.z.string(), LLMBreakdownStatsSchema),
+  modelsUsed: import_zod28.z.array(import_zod28.z.string())
+});
+var LLMTraceSchema = import_zod28.z.object({
+  id: import_zod28.z.string(),
+  steps: import_zod28.z.array(LLMTraceStepSchema),
   summary: LLMTraceSummarySchema
 });
 
 // src/evaluation/eval-result.ts
-var import_zod31 = require("zod");
+var import_zod32 = require("zod");
 
 // src/evaluation/eval-run.ts
-var import_zod29 = require("zod");
+var import_zod30 = require("zod");
 
 // src/evaluation/live-trace.ts
-var import_zod28 = require("zod");
+var import_zod29 = require("zod");
 var LiveTraceEventType = /* @__PURE__ */ ((LiveTraceEventType2) => {
   LiveTraceEventType2["THINKING"] = "thinking";
   LiveTraceEventType2["TOOL_USE"] = "tool_use";
@@ -1477,37 +1668,37 @@ var LiveTraceEventType = /* @__PURE__ */ ((LiveTraceEventType2) => {
   LiveTraceEventType2["USER"] = "user";
   return LiveTraceEventType2;
 })(LiveTraceEventType || {});
-var LiveTraceEventSchema = import_zod28.z.object({
+var LiveTraceEventSchema = import_zod29.z.object({
   /** The evaluation run ID */
-  evalRunId: import_zod28.z.string(),
+  evalRunId: import_zod29.z.string(),
   /** The scenario ID being executed */
-  scenarioId: import_zod28.z.string(),
+  scenarioId: import_zod29.z.string(),
   /** The scenario name for display */
-  scenarioName: import_zod28.z.string(),
+  scenarioName: import_zod29.z.string(),
   /** The target ID (skill, agent, etc.) */
-  targetId: import_zod28.z.string(),
+  targetId: import_zod29.z.string(),
   /** The target name for display */
-  targetName: import_zod28.z.string(),
+  targetName: import_zod29.z.string(),
   /** Step number in the current scenario execution */
-  stepNumber: import_zod28.z.number(),
+  stepNumber: import_zod29.z.number(),
   /** Type of trace event */
-  type: import_zod28.z.enum(LiveTraceEventType),
+  type: import_zod29.z.enum(LiveTraceEventType),
   /** Tool name if this is a tool_use event */
-  toolName: import_zod28.z.string().optional(),
+  toolName: import_zod29.z.string().optional(),
   /** Tool arguments preview (truncated JSON) */
-  toolArgs: import_zod28.z.string().optional(),
+  toolArgs: import_zod29.z.string().optional(),
   /** Output preview (truncated text) */
-  outputPreview: import_zod28.z.string().optional(),
+  outputPreview: import_zod29.z.string().optional(),
   /** File path for file operations */
-  filePath: import_zod28.z.string().optional(),
+  filePath: import_zod29.z.string().optional(),
   /** Elapsed time in milliseconds for progress events */
-  elapsedMs: import_zod28.z.number().optional(),
+  elapsedMs: import_zod29.z.number().optional(),
   /** Thinking/reasoning text from Claude */
-  thinking: import_zod28.z.string().optional(),
+  thinking: import_zod29.z.string().optional(),
   /** Timestamp when this event occurred */
-  timestamp: import_zod28.z.string(),
+  timestamp: import_zod29.z.string(),
   /** Whether this is the final event for this scenario */
-  isComplete: import_zod28.z.boolean()
+  isComplete: import_zod29.z.boolean()
 });
 var TRACE_EVENT_PREFIX = "TRACE_EVENT:";
 function parseTraceEventLine(line) {
@@ -1536,40 +1727,40 @@ var TriggerType = /* @__PURE__ */ ((TriggerType2) => {
   TriggerType2["SCHEDULED"] = "SCHEDULED";
   return TriggerType2;
 })(TriggerType || {});
-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 TriggerMetadataSchema = import_zod30.z.object({
+  version: import_zod30.z.string().optional(),
+  resourceUpdated: import_zod30.z.array(import_zod30.z.string()).optional(),
+  scheduleId: import_zod30.z.string().optional()
 });
-var TriggerSchema = import_zod29.z.object({
-  id: import_zod29.z.string(),
+var TriggerSchema = import_zod30.z.object({
+  id: import_zod30.z.string(),
   metadata: TriggerMetadataSchema.optional(),
-  type: import_zod29.z.nativeEnum(TriggerType)
+  type: import_zod30.z.nativeEnum(TriggerType)
 });
-var DiffLineTypeSchema = import_zod29.z.enum(["added", "removed", "unchanged"]);
-var DiffLineSchema = import_zod29.z.object({
+var DiffLineTypeSchema = import_zod30.z.enum(["added", "removed", "unchanged"]);
+var DiffLineSchema = import_zod30.z.object({
   type: DiffLineTypeSchema,
-  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(),
+  content: import_zod30.z.string(),
+  lineNumber: import_zod30.z.number()
+});
+var DiffContentSchema = import_zod30.z.object({
+  path: import_zod30.z.string(),
+  expected: import_zod30.z.string(),
+  actual: import_zod30.z.string(),
+  diffLines: import_zod30.z.array(DiffLineSchema),
+  renamedFrom: import_zod30.z.string().optional(),
   /** Whether this file is an infrastructure/config file (e.g. .claude/settings.json, .mcp.json) */
-  isInfrastructure: import_zod29.z.boolean().optional()
+  isInfrastructure: import_zod30.z.boolean().optional()
 });
-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 CommandExecutionSchema = import_zod30.z.object({
+  command: import_zod30.z.string(),
+  exitCode: import_zod30.z.number(),
+  output: import_zod30.z.string().optional(),
+  duration: import_zod30.z.number()
 });
-var FileModificationSchema = import_zod29.z.object({
-  path: import_zod29.z.string(),
-  action: import_zod29.z.enum(["created", "modified", "deleted"])
+var FileModificationSchema = import_zod30.z.object({
+  path: import_zod30.z.string(),
+  action: import_zod30.z.enum(["created", "modified", "deleted"])
 });
 var TemplateFileStatus = /* @__PURE__ */ ((TemplateFileStatus2) => {
   TemplateFileStatus2["NEW"] = "new";
@@ -1577,62 +1768,62 @@ var TemplateFileStatus = /* @__PURE__ */ ((TemplateFileStatus2) => {
   TemplateFileStatus2["UNCHANGED"] = "unchanged";
   return TemplateFileStatus2;
 })(TemplateFileStatus || {});
-var TemplateFileSchema = import_zod29.z.object({
+var TemplateFileSchema = import_zod30.z.object({
   /** Relative path within the template */
-  path: import_zod29.z.string(),
+  path: import_zod30.z.string(),
   /** Full file content after execution */
-  content: import_zod29.z.string(),
+  content: import_zod30.z.string(),
   /** File status (new, modified, unchanged) */
-  status: import_zod29.z.enum(["new", "modified", "unchanged"]),
+  status: import_zod30.z.enum(["new", "modified", "unchanged"]),
   /** Whether this file is an infrastructure/config file (e.g. .claude/settings.json, .mcp.json) */
-  isInfrastructure: import_zod29.z.boolean().optional()
+  isInfrastructure: import_zod30.z.boolean().optional()
 });
-var ApiCallSchema = import_zod29.z.object({
-  endpoint: import_zod29.z.string(),
-  tokensUsed: import_zod29.z.number(),
-  duration: import_zod29.z.number()
+var ApiCallSchema = import_zod30.z.object({
+  endpoint: import_zod30.z.string(),
+  tokensUsed: import_zod30.z.number(),
+  duration: import_zod30.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 ExecutionTraceSchema = import_zod30.z.object({
+  commands: import_zod30.z.array(CommandExecutionSchema),
+  filesModified: import_zod30.z.array(FileModificationSchema),
+  apiCalls: import_zod30.z.array(ApiCallSchema),
+  totalDuration: import_zod30.z.number()
 });
-var RunAnalysisFindingSchema = import_zod29.z.object({
-  category: import_zod29.z.enum([
+var RunAnalysisFindingSchema = import_zod30.z.object({
+  category: import_zod30.z.enum([
     "failure_pattern",
     "cost_waste",
     "flakiness",
     "inefficiency",
     "positive"
   ]),
-  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()
+  severity: import_zod30.z.enum(["high", "medium", "low"]),
+  description: import_zod30.z.string(),
+  affectedScenarios: import_zod30.z.array(import_zod30.z.string()),
+  recommendation: import_zod30.z.string().optional()
 });
-var RunAnalysisSchema = import_zod29.z.object({
-  generatedAt: import_zod29.z.string(),
-  summary: import_zod29.z.string(),
-  findings: import_zod29.z.array(RunAnalysisFindingSchema)
+var RunAnalysisSchema = import_zod30.z.object({
+  generatedAt: import_zod30.z.string(),
+  summary: import_zod30.z.string(),
+  findings: import_zod30.z.array(RunAnalysisFindingSchema)
 });
 var EvalRunSchema = TenantEntitySchema.extend({
   /** Agent ID for this run */
-  agentId: import_zod29.z.string().optional(),
+  agentId: import_zod30.z.string().optional(),
   /** Preset ID that originated this run (optional) */
-  presetId: import_zod29.z.string().optional(),
+  presetId: import_zod30.z.string().optional(),
   /** Skill IDs for this run */
-  skillIds: import_zod29.z.array(import_zod29.z.string()).optional(),
+  skillIds: import_zod30.z.array(import_zod30.z.string()).optional(),
   /** Map of skillId to skillVersionId for this run */
-  skillVersions: import_zod29.z.record(import_zod29.z.string(), import_zod29.z.string()).optional(),
+  skillVersions: import_zod30.z.record(import_zod30.z.string(), import_zod30.z.string()).optional(),
   /** Scenario IDs to run (always present — resolved server-side from tags when needed) */
-  scenarioIds: import_zod29.z.array(import_zod29.z.string()),
+  scenarioIds: import_zod30.z.array(import_zod30.z.string()),
   /** Current status */
   status: EvalStatusSchema,
   /** Progress percentage (0-100) */
-  progress: import_zod29.z.number(),
+  progress: import_zod30.z.number(),
   /** Results for each scenario/target combination (lazy to break eval-result ↔ eval-run cycle) */
-  results: import_zod29.z.array(import_zod29.z.lazy(() => EvalRunResultSchema)),
+  results: import_zod30.z.array(import_zod30.z.lazy(() => EvalRunResultSchema)),
   /** Aggregated metrics across all results */
   aggregateMetrics: EvalMetricsSchema,
   /** Aggregated LLM trace summary */
@@ -1640,41 +1831,45 @@ var EvalRunSchema = TenantEntitySchema.extend({
   /** What triggered this run */
   trigger: TriggerSchema.optional(),
   /** When the run started (set when evaluation is triggered) */
-  startedAt: import_zod29.z.string().optional(),
+  startedAt: import_zod30.z.string().optional(),
   /** When the run completed */
-  completedAt: import_zod29.z.string().optional(),
+  completedAt: import_zod30.z.string().optional(),
   /** Live trace events captured during execution (for playback on results page) */
-  liveTraceEvents: import_zod29.z.array(LiveTraceEventSchema).optional(),
+  liveTraceEvents: import_zod30.z.array(LiveTraceEventSchema).optional(),
   /** Remote job ID for tracking execution in Dev Machines */
-  jobId: import_zod29.z.string().optional(),
+  jobId: import_zod30.z.string().optional(),
   /** Remote job status from the Dev Machine API (PENDING, RUNNING, COMPLETED, FAILED, CANCELLED) */
-  jobStatus: import_zod29.z.string().optional(),
+  jobStatus: import_zod30.z.string().optional(),
   /** Remote job error message if the job failed */
-  jobError: import_zod29.z.string().optional(),
+  jobError: import_zod30.z.string().optional(),
   /** Timestamp of the last job status check */
-  jobStatusCheckedAt: import_zod29.z.string().optional(),
+  jobStatusCheckedAt: import_zod30.z.string().optional(),
   /** MCP server IDs to enable for this run (optional) */
-  mcpIds: import_zod29.z.array(import_zod29.z.string()).optional(),
+  mcpIds: import_zod30.z.array(import_zod30.z.string()).optional(),
   /** Sub-agent IDs to enable for this run (optional) */
-  subAgentIds: import_zod29.z.array(import_zod29.z.string()).optional(),
+  subAgentIds: import_zod30.z.array(import_zod30.z.string()).optional(),
   /** Rule IDs to enable for this run (optional) */
-  ruleIds: import_zod29.z.array(import_zod29.z.string()).optional(),
+  ruleIds: import_zod30.z.array(import_zod30.z.string()).optional(),
+  /** Unified capability IDs (replaces skill/mcp/subAgent/ruleIds) */
+  capabilityIds: import_zod30.z.array(import_zod30.z.string()).optional(),
+  /** Map of capabilityId to capabilityVersionId for version pinning */
+  capabilityVersions: import_zod30.z.record(import_zod30.z.string(), import_zod30.z.string()).optional(),
   /** Tags used to select scenarios for this run (for traceability) */
-  tags: import_zod29.z.array(import_zod29.z.string()).optional(),
+  tags: import_zod30.z.array(import_zod30.z.string()).optional(),
   /** How many times each scenario is executed within this eval run. Default: 1. Max: 20. */
-  runsPerScenario: import_zod29.z.number().int().min(1).max(20).optional(),
+  runsPerScenario: import_zod30.z.number().int().min(1).max(20).optional(),
   /** Snapshot of agent configuration captured at run creation time */
-  agentSnapshot: import_zod29.z.object({
-    name: import_zod29.z.string().optional(),
+  agentSnapshot: import_zod30.z.object({
+    name: import_zod30.z.string().optional(),
     agentType: AgentTypeSchema.optional(),
     runCommand: AgentRunCommandSchema.optional(),
-    systemPrompt: import_zod29.z.string().nullable().optional(),
+    systemPrompt: import_zod30.z.string().nullable().optional(),
     modelConfig: ModelConfigSchema.optional()
   }).optional(),
   /** UUID linking all runs in a comparison group */
-  comparisonGroupId: import_zod29.z.string().optional(),
+  comparisonGroupId: import_zod30.z.string().optional(),
   /** Human-readable label for this variant (e.g., "MCP: Wix Stores") */
-  comparisonLabel: import_zod29.z.string().optional(),
+  comparisonLabel: import_zod30.z.string().optional(),
   /** LLM-generated analysis of the completed run */
   runAnalysis: RunAnalysisSchema.optional()
 });
@@ -1692,60 +1887,60 @@ var CreateEvalRunInputSchema = EvalRunSchema.omit({
   agentSnapshot: true
 }).extend({
   /** Optional on input — backend resolves from tags when not provided */
-  scenarioIds: import_zod29.z.array(import_zod29.z.string()).optional()
+  scenarioIds: import_zod30.z.array(import_zod30.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_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()
+var EvaluationProgressSchema = import_zod30.z.object({
+  runId: import_zod30.z.string(),
+  targetId: import_zod30.z.string(),
+  totalScenarios: import_zod30.z.number(),
+  completedScenarios: import_zod30.z.number(),
+  scenarioProgress: import_zod30.z.array(
+    import_zod30.z.object({
+      scenarioId: import_zod30.z.string(),
+      currentStep: import_zod30.z.string(),
+      error: import_zod30.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()
+  createdAt: import_zod30.z.number()
+});
+var EvaluationLogSchema = import_zod30.z.object({
+  runId: import_zod30.z.string(),
+  scenarioId: import_zod30.z.string(),
+  log: import_zod30.z.object({
+    level: import_zod30.z.enum(["info", "error", "debug"]),
+    message: import_zod30.z.string().optional(),
+    args: import_zod30.z.array(import_zod30.z.any()).optional(),
+    error: import_zod30.z.string().optional()
   })
 });
 var LLM_TIMEOUT = 12e4;
 
 // src/evaluation/conversation.ts
-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", [
+var import_zod31 = require("zod");
+var TextBlockSchema = import_zod31.z.object({
+  type: import_zod31.z.literal("text"),
+  text: import_zod31.z.string()
+});
+var ThinkingBlockSchema = import_zod31.z.object({
+  type: import_zod31.z.literal("thinking"),
+  thinking: import_zod31.z.string()
+});
+var ToolUseBlockSchema = import_zod31.z.object({
+  type: import_zod31.z.literal("tool_use"),
+  toolName: import_zod31.z.string(),
+  toolId: import_zod31.z.string(),
+  input: import_zod31.z.unknown()
+});
+var ToolResultBlockSchema = import_zod31.z.object({
+  type: import_zod31.z.literal("tool_result"),
+  toolUseId: import_zod31.z.string(),
+  content: import_zod31.z.string(),
+  isError: import_zod31.z.boolean().optional()
+});
+var ConversationBlockSchema = import_zod31.z.discriminatedUnion("type", [
   TextBlockSchema,
   ThinkingBlockSchema,
   ToolUseBlockSchema,
@@ -1756,18 +1951,18 @@ var ConversationMessageRoles = [
   "user",
   "system"
 ];
-var ConversationMessageSchema = import_zod30.z.object({
-  role: import_zod30.z.enum(ConversationMessageRoles),
-  content: import_zod30.z.array(ConversationBlockSchema),
-  timestamp: import_zod30.z.string()
+var ConversationMessageSchema = import_zod31.z.object({
+  role: import_zod31.z.enum(ConversationMessageRoles),
+  content: import_zod31.z.array(ConversationBlockSchema),
+  timestamp: import_zod31.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()
+var ScenarioConversationSchema = import_zod31.z.object({
+  id: import_zod31.z.string(),
+  projectId: import_zod31.z.string(),
+  evalRunId: import_zod31.z.string(),
+  resultId: import_zod31.z.string(),
+  messages: import_zod31.z.array(ConversationMessageSchema),
+  createdAt: import_zod31.z.string()
 });
 
 // src/evaluation/eval-result.ts
@@ -1778,98 +1973,98 @@ var AssertionResultStatus = /* @__PURE__ */ ((AssertionResultStatus2) => {
   AssertionResultStatus2["ERROR"] = "error";
   return AssertionResultStatus2;
 })(AssertionResultStatus || {});
-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(),
+var AssertionResultSchema = import_zod32.z.object({
+  id: import_zod32.z.string(),
+  assertionId: import_zod32.z.string(),
+  assertionType: import_zod32.z.string(),
+  assertionName: import_zod32.z.string(),
+  status: import_zod32.z.enum(AssertionResultStatus),
+  message: import_zod32.z.string().optional(),
+  expected: import_zod32.z.string().optional(),
+  actual: import_zod32.z.string().optional(),
+  duration: import_zod32.z.number().optional(),
+  details: import_zod32.z.record(import_zod32.z.string(), import_zod32.z.unknown()).optional(),
+  llmTraceSteps: import_zod32.z.array(LLMTraceStepSchema).optional()
+});
+var EvalRunResultSchema = import_zod32.z.object({
+  id: import_zod32.z.string(),
+  targetId: import_zod32.z.string(),
+  targetName: import_zod32.z.string().optional(),
   /** SkillVersion ID used for this evaluation (for version tracking) */
-  skillVersionId: import_zod31.z.string().optional(),
+  skillVersionId: import_zod32.z.string().optional(),
   /** SkillVersion semver string (e.g., "1.0.0", "1.2.3") for display */
-  skillVersion: import_zod31.z.string().optional(),
-  scenarioId: import_zod31.z.string(),
-  scenarioName: import_zod31.z.string(),
+  skillVersion: import_zod32.z.string().optional(),
+  scenarioId: import_zod32.z.string(),
+  scenarioName: import_zod32.z.string(),
   /** Snapshot of the trigger prompt used during the run (prevents stale display after edits) */
-  triggerPrompt: import_zod31.z.string().optional(),
+  triggerPrompt: import_zod32.z.string().optional(),
   modelConfig: ModelConfigSchema.optional(),
-  assertionResults: import_zod31.z.array(AssertionResultSchema),
+  assertionResults: import_zod32.z.array(AssertionResultSchema),
   metrics: EvalMetricsSchema.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(),
+  passed: import_zod32.z.number(),
+  failed: import_zod32.z.number(),
+  passRate: import_zod32.z.number(),
+  duration: import_zod32.z.number(),
+  outputText: import_zod32.z.string().optional(),
+  files: import_zod32.z.array(ExpectedFileSchema).optional(),
+  fileDiffs: import_zod32.z.array(DiffContentSchema).optional(),
   /** Full template files after execution with status indicators */
-  templateFiles: import_zod31.z.array(TemplateFileSchema).optional(),
-  startedAt: import_zod31.z.string().optional(),
-  completedAt: import_zod31.z.string().optional(),
+  templateFiles: import_zod32.z.array(TemplateFileSchema).optional(),
+  startedAt: import_zod32.z.string().optional(),
+  completedAt: import_zod32.z.string().optional(),
   llmTrace: LLMTraceSchema.optional(),
   /** Full conversation messages (only present in transit; stripped before DB storage) */
-  conversation: import_zod31.z.array(ConversationMessageSchema).optional(),
+  conversation: import_zod32.z.array(ConversationMessageSchema).optional(),
   /** 0-based iteration index when a scenario is run multiple times within a single eval run */
-  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()
+  iterationIndex: import_zod32.z.number().int().min(0).optional()
+});
+var PromptResultSchema = import_zod32.z.object({
+  text: import_zod32.z.string(),
+  files: import_zod32.z.array(import_zod32.z.unknown()).optional(),
+  finishReason: import_zod32.z.string().optional(),
+  reasoning: import_zod32.z.string().optional(),
+  reasoningDetails: import_zod32.z.unknown().optional(),
+  toolCalls: import_zod32.z.array(import_zod32.z.unknown()).optional(),
+  toolResults: import_zod32.z.array(import_zod32.z.unknown()).optional(),
+  warnings: import_zod32.z.array(import_zod32.z.unknown()).optional(),
+  sources: import_zod32.z.array(import_zod32.z.unknown()).optional(),
+  steps: import_zod32.z.array(import_zod32.z.unknown()),
+  generationTimeMs: import_zod32.z.number(),
+  prompt: import_zod32.z.string(),
+  systemPrompt: import_zod32.z.string(),
+  usage: import_zod32.z.object({
+    totalTokens: import_zod32.z.number().optional(),
+    totalMicrocentsSpent: import_zod32.z.number().optional()
   })
 });
-var EvaluationResultSchema = import_zod31.z.object({
-  id: import_zod31.z.string(),
-  runId: import_zod31.z.string(),
-  timestamp: import_zod31.z.number(),
+var EvaluationResultSchema = import_zod32.z.object({
+  id: import_zod32.z.string(),
+  runId: import_zod32.z.string(),
+  timestamp: import_zod32.z.number(),
   promptResult: PromptResultSchema,
-  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()
+  testResults: import_zod32.z.array(import_zod32.z.unknown()),
+  tags: import_zod32.z.array(import_zod32.z.string()).optional(),
+  feedback: import_zod32.z.string().optional(),
+  score: import_zod32.z.number(),
+  suiteId: import_zod32.z.string().optional()
+});
+var LeanEvaluationResultSchema = import_zod32.z.object({
+  id: import_zod32.z.string(),
+  runId: import_zod32.z.string(),
+  timestamp: import_zod32.z.number(),
+  tags: import_zod32.z.array(import_zod32.z.string()).optional(),
+  scenarioId: import_zod32.z.string(),
+  scenarioVersion: import_zod32.z.number().optional(),
+  targetId: import_zod32.z.string(),
+  targetVersion: import_zod32.z.number().optional(),
+  suiteId: import_zod32.z.string().optional(),
+  score: import_zod32.z.number(),
+  time: import_zod32.z.number().optional(),
+  microcentsSpent: import_zod32.z.number().optional()
 });
 
 // src/evaluation/eval-run-folder.ts
-var import_zod32 = require("zod");
+var import_zod33 = require("zod");
 var EvalRunFolderSchema = TenantEntitySchema.extend({});
 var CreateEvalRunFolderInputSchema = EvalRunFolderSchema.omit({
   id: true,
@@ -1883,26 +2078,26 @@ var UpdateEvalRunFolderInputSchema = EvalRunFolderSchema.omit({
   updatedAt: true,
   deleted: true
 }).partial();
-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()
+var EvalRunFolderMembershipSchema = import_zod33.z.object({
+  folderId: import_zod33.z.string(),
+  evalRunId: import_zod33.z.string(),
+  projectId: import_zod33.z.string(),
+  createdAt: import_zod33.z.string()
 });
 
 // src/project/project.ts
-var import_zod33 = require("zod");
+var import_zod34 = require("zod");
 var ProjectSchema = BaseEntitySchema.extend({
-  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"),
+  appId: import_zod34.z.string().optional().describe("The ID of the app in Dev Center"),
+  scenarioTags: import_zod34.z.array(import_zod34.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_zod33.z.string().nullable().optional().describe("Wix auth token for CLI/MCP authentication (encrypted at rest)"),
+  wixAuthToken: import_zod34.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_zod33.z.string().nullable().optional().describe("Base64-encoded Base44 auth file content (encrypted at rest)"),
+  base44AuthFile: import_zod34.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_zod33.z.string().optional().describe("Email associated with the Wix auth token (resolved at runtime)"),
+  wixAuthEmail: import_zod34.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_zod33.z.string().optional().describe("Email from the Base44 auth file (resolved at runtime)")
+  base44AuthEmail: import_zod34.z.string().optional().describe("Email from the Base44 auth file (resolved at runtime)")
 });
 var CreateProjectInputSchema = ProjectSchema.omit({
   id: true,
@@ -1911,6 +2106,10 @@ var CreateProjectInputSchema = ProjectSchema.omit({
   deleted: true,
   wixAuthEmail: true,
   base44AuthEmail: true
+}).extend({
+  appId: import_zod34.z.string().describe(
+    "Required: The ID of the app in Dev Center for credential scoping"
+  )
 });
 var UpdateProjectInputSchema = CreateProjectInputSchema.partial();
 
@@ -1928,7 +2127,7 @@ var CreateTemplateInputSchema = TemplateSchema.omit({
 var UpdateTemplateInputSchema = CreateTemplateInputSchema.partial();
 
 // src/schedule/eval-schedule.ts
-var import_zod34 = require("zod");
+var import_zod35 = require("zod");
 var FrequencyType = /* @__PURE__ */ ((FrequencyType2) => {
   FrequencyType2["DAILY"] = "daily";
   FrequencyType2["WEEKDAY"] = "weekday";
@@ -1938,29 +2137,29 @@ var FrequencyType = /* @__PURE__ */ ((FrequencyType2) => {
 })(FrequencyType || {});
 var EvalScheduleSchema = TenantEntitySchema.extend({
   /** Whether the schedule is active */
-  enabled: import_zod34.z.boolean(),
+  enabled: import_zod35.z.boolean(),
   /** Test suite to run */
-  suiteId: import_zod34.z.string(),
+  suiteId: import_zod35.z.string(),
   /** Preset that provides agent + entities for this schedule */
-  presetId: import_zod34.z.string(),
+  presetId: import_zod35.z.string(),
   /** How often to run */
-  frequencyType: import_zod34.z.nativeEnum(FrequencyType),
+  frequencyType: import_zod35.z.nativeEnum(FrequencyType),
   /** Time of day in 24h format (HH:MM), hours 00-23, minutes 00-59 */
-  timeOfDay: import_zod34.z.string().regex(/^([01]\d|2[0-3]):[0-5]\d$/),
+  timeOfDay: import_zod35.z.string().regex(/^([01]\d|2[0-3]):[0-5]\d$/),
   /** Day of week (0=Sun, 6=Sat) for weekly schedules */
-  dayOfWeek: import_zod34.z.number().min(0).max(6).optional(),
+  dayOfWeek: import_zod35.z.number().min(0).max(6).optional(),
   /** Day of month (1-31) for monthly schedules */
-  dayOfMonth: import_zod34.z.number().min(1).max(31).optional(),
+  dayOfMonth: import_zod35.z.number().min(1).max(31).optional(),
   /** IANA timezone (e.g., 'America/New_York') */
-  timezone: import_zod34.z.string(),
+  timezone: import_zod35.z.string(),
   /** ID of the last eval run created by this schedule */
-  lastRunId: import_zod34.z.string().optional(),
+  lastRunId: import_zod35.z.string().optional(),
   /** Denormalized status of the last run */
-  lastRunStatus: import_zod34.z.string().optional(),
+  lastRunStatus: import_zod35.z.string().optional(),
   /** ISO timestamp of the last run */
-  lastRunAt: import_zod34.z.string().optional(),
+  lastRunAt: import_zod35.z.string().optional(),
   /** Next scheduled run time in UTC (pre-computed for efficient querying, set by backend) */
-  nextRunAt: import_zod34.z.string().optional()
+  nextRunAt: import_zod35.z.string().optional()
 });
 function isValidTimezone(tz) {
   try {
@@ -1973,14 +2172,14 @@ function isValidTimezone(tz) {
 function validateScheduleFields(data, ctx, options) {
   if (data.frequencyType === "weekly" /* WEEKLY */ && data.dayOfWeek == null) {
     ctx.addIssue({
-      code: import_zod34.z.ZodIssueCode.custom,
+      code: import_zod35.z.ZodIssueCode.custom,
       message: "dayOfWeek is required for weekly schedules",
       path: ["dayOfWeek"]
     });
   }
   if (data.frequencyType === "monthly" /* MONTHLY */ && data.dayOfMonth == null) {
     ctx.addIssue({
-      code: import_zod34.z.ZodIssueCode.custom,
+      code: import_zod35.z.ZodIssueCode.custom,
       message: "dayOfMonth is required for monthly schedules",
       path: ["dayOfMonth"]
     });
@@ -1988,7 +2187,7 @@ function validateScheduleFields(data, ctx, options) {
   const shouldValidateTz = options.partial ? data.timezone !== void 0 : true;
   if (shouldValidateTz && !isValidTimezone(data.timezone)) {
     ctx.addIssue({
-      code: import_zod34.z.ZodIssueCode.custom,
+      code: import_zod35.z.ZodIssueCode.custom,
       message: "Invalid IANA timezone",
       path: ["timezone"]
     });
@@ -2053,6 +2252,13 @@ var UpdateEvalScheduleInputSchema = BaseCreateScheduleSchema.partial().superRefi
   BulkImportResultItemSchema,
   BulkImportResultSchema,
   BulkImportSkillsInputSchema,
+  CAPABILITY_NAME_REGEX,
+  CapabilityContentSchema,
+  CapabilitySchema,
+  CapabilityTypeSchema,
+  CapabilityVersionOriginSchema,
+  CapabilityVersionSchema,
+  CapabilityWithLatestVersionSchema,
   ClaudeModel,
   ClaudeModelSchema,
   CommandExecutionSchema,
@@ -2063,6 +2269,8 @@ var UpdateEvalScheduleInputSchema = BaseCreateScheduleSchema.partial().superRefi
   CostAssertionSchema,
   CostConfigSchema,
   CreateAgentInputSchema,
+  CreateCapabilityInputSchema,
+  CreateCapabilityVersionInputSchema,
   CreateEvalRunFolderInputSchema,
   CreateEvalRunInputSchema,
   CreateEvalScheduleInputSchema,
@@ -2102,6 +2310,7 @@ var UpdateEvalScheduleInputSchema = BaseCreateScheduleSchema.partial().superRefi
   FilePresenceTestSchema,
   FrequencyType,
   GitHubSourceSchema,
+  InitialCapabilityVersionInputSchema,
   InitialVersionInputSchema,
   LEGACY_MODEL_ID_MAP,
   LLMBreakdownStatsSchema,
@@ -2178,6 +2387,7 @@ var UpdateEvalScheduleInputSchema = BaseCreateScheduleSchema.partial().superRefi
   TriggerSchema,
   TriggerType,
   UpdateAgentInputSchema,
+  UpdateCapabilityInputSchema,
   UpdateEvalRunFolderInputSchema,
   UpdateEvalScheduleInputSchema,
   UpdateMcpInputSchema,
@@ -2190,12 +2400,20 @@ var UpdateEvalScheduleInputSchema = BaseCreateScheduleSchema.partial().superRefi
   UpdateTestScenarioInputSchema,
   UpdateTestSuiteInputSchema,
   VitestTestSchema,
+  capabilityToMcp,
+  capabilityToRule,
+  capabilityToSkill,
+  capabilityToSkillWithLatestVersion,
+  capabilityToSubAgent,
+  capabilityVersionToSkillVersion,
   classifyAssertionRef,
   formatTraceEventLine,
   getSystemAssertion,
   getSystemAssertions,
+  groupCapabilitiesByType,
   isAllowedBuildCommandString,
   isSystemAssertionId,
+  isValidCapabilityName,
   isValidSkillFolderName,
   normalizeBatchAssertionLink,
   normalizeModelId,