@wix/evalforge-types 0.20.0 → 0.22.0

package/build/index.mjs

@@ -14,18 +14,21 @@ var TenantEntitySchema = BaseEntitySchema.extend({
 
 // src/common/mcp.ts
 import { z as z2 } from "zod";
-var MCPServerConfigSchema = z2.object({
-  /** Unique name for this MCP server */
-  name: z2.string(),
-  /** Command to start the MCP server */
-  command: z2.string(),
-  /** Command line arguments */
-  args: z2.array(z2.string()).optional(),
-  /** Environment variables for the server process */
-  envVars: z2.record(z2.string(), z2.string()).optional(),
-  /** Tools to disable for this MCP server */
-  disabledTools: z2.array(z2.string()).optional()
+var MCP_SERVERS_JSON_KEY = "mcpServers";
+var MCPEntitySchema = TenantEntitySchema.extend({
+  /** Display name and key in mcp.json mcpServers object */
+  name: z2.string().min(1),
+  /** MCP server config (command/args, url/headers, etc.) - stored as-is for mcp.json */
+  config: z2.record(z2.string(), z2.unknown())
+});
+var CreateMcpInputSchema = MCPEntitySchema.omit({
+  id: true,
+  createdAt: true,
+  updatedAt: true,
+  deleted: true
 });
+var UpdateMcpInputSchema = CreateMcpInputSchema.partial();
+var MCPServerConfigSchema = z2.record(z2.string(), z2.unknown());
 
 // src/common/models.ts
 import { z as z3 } from "zod";
@@ -205,11 +208,26 @@ var CreateSkillsGroupInputSchema = SkillsGroupSchema.omit({
 });
 var UpdateSkillsGroupInputSchema = CreateSkillsGroupInputSchema.partial();
 
+// src/target/sub-agent.ts
+import { z as z7 } from "zod";
+var SubAgentSchema = TargetSchema.extend({
+  /** The full sub-agent markdown content (YAML frontmatter + body) */
+  subAgentMd: z7.string()
+});
+var SubAgentInputBaseSchema = SubAgentSchema.omit({
+  id: true,
+  createdAt: true,
+  updatedAt: true,
+  deleted: true
+});
+var CreateSubAgentInputSchema = SubAgentInputBaseSchema;
+var UpdateSubAgentInputSchema = SubAgentInputBaseSchema.partial();
+
 // src/test/index.ts
-import { z as z17 } from "zod";
+import { z as z18 } from "zod";
 
 // src/test/base.ts
-import { z as z7 } from "zod";
+import { z as z8 } from "zod";
 var TestType = /* @__PURE__ */ ((TestType2) => {
   TestType2["LLM"] = "LLM";
   TestType2["TOOL"] = "TOOL";
@@ -222,7 +240,7 @@ var TestType = /* @__PURE__ */ ((TestType2) => {
   TestType2["PLAYWRIGHT_NL"] = "PLAYWRIGHT_NL";
   return TestType2;
 })(TestType || {});
-var TestTypeSchema = z7.enum(TestType);
+var TestTypeSchema = z8.enum(TestType);
 var TestImportance = /* @__PURE__ */ ((TestImportance2) => {
   TestImportance2["LOW"] = "low";
   TestImportance2["MEDIUM"] = "medium";
@@ -230,153 +248,153 @@ var TestImportance = /* @__PURE__ */ ((TestImportance2) => {
   TestImportance2["CRITICAL"] = "critical";
   return TestImportance2;
 })(TestImportance || {});
-var TestImportanceSchema = z7.enum(TestImportance);
-var BaseTestSchema = z7.object({
-  id: z7.string(),
+var TestImportanceSchema = z8.enum(TestImportance);
+var BaseTestSchema = z8.object({
+  id: z8.string(),
   type: TestTypeSchema,
-  name: z7.string().min(3),
-  description: z7.string().optional(),
+  name: z8.string().min(3),
+  description: z8.string().optional(),
   importance: TestImportanceSchema.optional()
 });
 
 // src/test/llm.ts
-import { z as z8 } from "zod";
+import { z as z9 } from "zod";
 var LLMTestSchema = BaseTestSchema.extend({
-  type: z8.literal("LLM" /* LLM */),
+  type: z9.literal("LLM" /* LLM */),
   /** Maximum steps for the LLM to take */
-  maxSteps: z8.number().min(1).max(100),
+  maxSteps: z9.number().min(1).max(100),
   /** Prompt to send to the evaluator */
-  prompt: z8.string().min(1),
+  prompt: z9.string().min(1),
   /** ID of the evaluator agent to use */
-  evaluatorId: z8.string()
+  evaluatorId: z9.string()
 });
 
 // src/test/tool.ts
-import { z as z9 } from "zod";
+import { z as z10 } from "zod";
 var ToolTestSchema = BaseTestSchema.extend({
-  type: z9.literal("TOOL" /* TOOL */),
+  type: z10.literal("TOOL" /* TOOL */),
   /** Name of the tool that should be called */
-  toolName: z9.string().min(3),
+  toolName: z10.string().min(3),
   /** Expected arguments for the tool call */
-  args: z9.record(z9.string(), z9.any()),
+  args: z10.record(z10.string(), z10.any()),
   /** Expected content in the tool results */
-  resultsContent: z9.string()
+  resultsContent: z10.string()
 });
 
 // src/test/site-config.ts
-import { z as z10 } from "zod";
+import { z as z11 } from "zod";
 var SiteConfigTestSchema = BaseTestSchema.extend({
-  type: z10.literal("SITE_CONFIG" /* SITE_CONFIG */),
+  type: z11.literal("SITE_CONFIG" /* SITE_CONFIG */),
   /** URL to call */
-  url: z10.string().url(),
+  url: z11.string().url(),
   /** HTTP method */
-  method: z10.enum(["GET", "POST"]),
+  method: z11.enum(["GET", "POST"]),
   /** Request body (for POST) */
-  body: z10.string().optional(),
+  body: z11.string().optional(),
   /** Expected HTTP status code */
-  expectedStatusCode: z10.number().int().min(100).max(599),
+  expectedStatusCode: z11.number().int().min(100).max(599),
   /** Expected response content */
-  expectedResponse: z10.string().optional(),
+  expectedResponse: z11.string().optional(),
   /** JMESPath expression to extract from response */
-  expectedResponseJMESPath: z10.string().optional()
+  expectedResponseJMESPath: z11.string().optional()
 });
 
 // src/test/command-execution.ts
-import { z as z11 } from "zod";
+import { z as z12 } from "zod";
 var AllowedCommands = [
   "yarn install --no-immutable && yarn build",
   "npm run build",
   "yarn typecheck"
 ];
 var CommandExecutionTestSchema = BaseTestSchema.extend({
-  type: z11.literal("COMMAND_EXECUTION" /* COMMAND_EXECUTION */),
+  type: z12.literal("COMMAND_EXECUTION" /* COMMAND_EXECUTION */),
   /** Command to execute (must be in AllowedCommands) */
-  command: z11.string().refine((value) => AllowedCommands.includes(value), {
+  command: z12.string().refine((value) => AllowedCommands.includes(value), {
     message: `Command must be one of: ${AllowedCommands.join(", ")}`
   }),
   /** Expected exit code (default: 0) */
-  expectedExitCode: z11.number().default(0).optional()
+  expectedExitCode: z12.number().default(0).optional()
 });
 
 // src/test/file-presence.ts
-import { z as z12 } from "zod";
+import { z as z13 } from "zod";
 var FilePresenceTestSchema = BaseTestSchema.extend({
-  type: z12.literal("FILE_PRESENCE" /* FILE_PRESENCE */),
+  type: z13.literal("FILE_PRESENCE" /* FILE_PRESENCE */),
   /** Paths to check */
-  paths: z12.array(z12.string()),
+  paths: z13.array(z13.string()),
   /** Whether files should exist (true) or not exist (false) */
-  shouldExist: z12.boolean()
+  shouldExist: z13.boolean()
 });
 
 // src/test/file-content.ts
-import { z as z13 } from "zod";
-var FileContentCheckSchema = z13.object({
+import { z as z14 } from "zod";
+var FileContentCheckSchema = z14.object({
   /** Strings that must be present in the file */
-  contains: z13.array(z13.string()).optional(),
+  contains: z14.array(z14.string()).optional(),
   /** Strings that must NOT be present in the file */
-  notContains: z13.array(z13.string()).optional(),
+  notContains: z14.array(z14.string()).optional(),
   /** Regex pattern the content must match */
-  matches: z13.string().optional(),
+  matches: z14.string().optional(),
   /** JSON path checks for structured content */
-  jsonPath: z13.array(
-    z13.object({
-      path: z13.string(),
-      value: z13.unknown()
+  jsonPath: z14.array(
+    z14.object({
+      path: z14.string(),
+      value: z14.unknown()
     })
   ).optional(),
   /** Lines that should be added (for diff checking) */
-  added: z13.array(z13.string()).optional(),
+  added: z14.array(z14.string()).optional(),
   /** Lines that should be removed (for diff checking) */
-  removed: z13.array(z13.string()).optional()
+  removed: z14.array(z14.string()).optional()
 });
 var FileContentTestSchema = BaseTestSchema.extend({
-  type: z13.literal("FILE_CONTENT" /* FILE_CONTENT */),
+  type: z14.literal("FILE_CONTENT" /* FILE_CONTENT */),
   /** Path to the file to check */
-  path: z13.string(),
+  path: z14.string(),
   /** Content checks to perform */
   checks: FileContentCheckSchema
 });
 
 // src/test/build-check.ts
-import { z as z14 } from "zod";
+import { z as z15 } from "zod";
 var BuildCheckTestSchema = BaseTestSchema.extend({
-  type: z14.literal("BUILD_CHECK" /* BUILD_CHECK */),
+  type: z15.literal("BUILD_CHECK" /* BUILD_CHECK */),
   /** Build command to execute */
-  command: z14.string(),
+  command: z15.string(),
   /** Whether the build should succeed */
-  expectSuccess: z14.boolean(),
+  expectSuccess: z15.boolean(),
   /** Maximum allowed warnings (optional) */
-  allowedWarnings: z14.number().optional(),
+  allowedWarnings: z15.number().optional(),
   /** Timeout in milliseconds */
-  timeout: z14.number().optional()
+  timeout: z15.number().optional()
 });
 
 // src/test/vitest.ts
-import { z as z15 } from "zod";
+import { z as z16 } from "zod";
 var VitestTestSchema = BaseTestSchema.extend({
-  type: z15.literal("VITEST" /* VITEST */),
+  type: z16.literal("VITEST" /* VITEST */),
   /** Test file content */
-  testFile: z15.string(),
+  testFile: z16.string(),
   /** Name of the test file */
-  testFileName: z15.string(),
+  testFileName: z16.string(),
   /** Minimum pass rate required (0-100) */
-  minPassRate: z15.number().min(0).max(100)
+  minPassRate: z16.number().min(0).max(100)
 });
 
 // src/test/playwright-nl.ts
-import { z as z16 } from "zod";
+import { z as z17 } from "zod";
 var PlaywrightNLTestSchema = BaseTestSchema.extend({
-  type: z16.literal("PLAYWRIGHT_NL" /* PLAYWRIGHT_NL */),
+  type: z17.literal("PLAYWRIGHT_NL" /* PLAYWRIGHT_NL */),
   /** Natural language steps to execute */
-  steps: z16.array(z16.string()),
+  steps: z17.array(z17.string()),
   /** Expected outcome description */
-  expectedOutcome: z16.string(),
+  expectedOutcome: z17.string(),
   /** Timeout in milliseconds */
-  timeout: z16.number().optional()
+  timeout: z17.number().optional()
 });
 
 // src/test/index.ts
-var TestSchema = z17.discriminatedUnion("type", [
+var TestSchema = z18.discriminatedUnion("type", [
   LLMTestSchema,
   ToolTestSchema,
   SiteConfigTestSchema,
@@ -389,66 +407,66 @@ var TestSchema = z17.discriminatedUnion("type", [
 ]);
 
 // src/scenario/assertions.ts
-import { z as z18 } from "zod";
-var SkillWasCalledAssertionSchema = z18.object({
-  type: z18.literal("skill_was_called"),
+import { z as z19 } from "zod";
+var SkillWasCalledAssertionSchema = z19.object({
+  type: z19.literal("skill_was_called"),
   /** Name of the skill that must have been called (matched against trace Skill tool args) */
-  skillName: z18.string()
+  skillName: z19.string()
 });
-var BuildPassedAssertionSchema = z18.object({
-  type: z18.literal("build_passed"),
+var BuildPassedAssertionSchema = z19.object({
+  type: z19.literal("build_passed"),
   /** Command to run (default: "yarn build") */
-  command: z18.string().optional(),
+  command: z19.string().optional(),
   /** Expected exit code (default: 0) */
-  expectedExitCode: z18.number().int().optional()
+  expectedExitCode: z19.number().int().optional()
 });
-var LlmJudgeAssertionSchema = z18.object({
-  type: z18.literal("llm_judge"),
+var LlmJudgeAssertionSchema = z19.object({
+  type: z19.literal("llm_judge"),
   /** Prompt template; placeholders: {{output}}, {{cwd}}, {{changedFiles}}, {{trace}} */
-  prompt: z18.string(),
+  prompt: z19.string(),
   /** Optional system prompt for the judge (default asks for JSON with score) */
-  systemPrompt: z18.string().optional(),
+  systemPrompt: z19.string().optional(),
   /** Minimum score to pass (0-100, default 70) */
-  minScore: z18.number().int().min(0).max(100).optional(),
+  minScore: z19.number().int().min(0).max(100).optional(),
   /** Model for the judge (e.g. claude-3-5-haiku) */
-  model: z18.string().optional(),
-  maxTokens: z18.number().int().optional(),
-  temperature: z18.number().min(0).max(1).optional()
+  model: z19.string().optional(),
+  maxTokens: z19.number().int().optional(),
+  temperature: z19.number().min(0).max(1).optional()
 });
-var AssertionSchema = z18.union([
+var AssertionSchema = z19.union([
   SkillWasCalledAssertionSchema,
   BuildPassedAssertionSchema,
   LlmJudgeAssertionSchema
 ]);
 
 // src/scenario/environment.ts
-import { z as z19 } from "zod";
-var LocalProjectConfigSchema = z19.object({
+import { z as z20 } from "zod";
+var LocalProjectConfigSchema = z20.object({
   /** Template ID to use for the local project */
-  templateId: z19.string().optional(),
+  templateId: z20.string().optional(),
   /** Files to create in the project */
-  files: z19.array(
-    z19.object({
-      path: z19.string().min(1),
-      content: z19.string().min(1)
+  files: z20.array(
+    z20.object({
+      path: z20.string().min(1),
+      content: z20.string().min(1)
     })
   ).optional()
 });
-var MetaSiteConfigSchema = z19.object({
-  configurations: z19.array(
-    z19.object({
-      name: z19.string().min(1),
-      apiCalls: z19.array(
-        z19.object({
-          url: z19.string().url(),
-          method: z19.enum(["POST", "PUT"]),
-          body: z19.string()
+var MetaSiteConfigSchema = z20.object({
+  configurations: z20.array(
+    z20.object({
+      name: z20.string().min(1),
+      apiCalls: z20.array(
+        z20.object({
+          url: z20.string().url(),
+          method: z20.enum(["POST", "PUT"]),
+          body: z20.string()
         })
       )
     })
   ).optional()
 });
-var EnvironmentSchema = z19.object({
+var EnvironmentSchema = z20.object({
   /** Local project configuration */
   localProject: LocalProjectConfigSchema.optional(),
   /** Meta site configuration */
@@ -456,54 +474,54 @@ var EnvironmentSchema = z19.object({
 });
 
 // src/scenario/test-scenario.ts
-import { z as z21 } from "zod";
+import { z as z22 } from "zod";
 
 // src/assertion/assertion.ts
-import { z as z20 } from "zod";
-var AssertionTypeSchema = z20.enum([
+import { z as z21 } from "zod";
+var AssertionTypeSchema = z21.enum([
   "skill_was_called",
   "build_passed",
   "llm_judge"
 ]);
-var AssertionParameterTypeSchema = z20.enum([
+var AssertionParameterTypeSchema = z21.enum([
   "string",
   "number",
   "boolean"
 ]);
-var AssertionParameterSchema = z20.object({
+var AssertionParameterSchema = z21.object({
   /** Parameter name (used as key in params object) */
-  name: z20.string().min(1),
+  name: z21.string().min(1),
   /** Display label for the parameter */
-  label: z20.string().min(1),
+  label: z21.string().min(1),
   /** Parameter type */
   type: AssertionParameterTypeSchema,
   /** Whether this parameter is required */
-  required: z20.boolean(),
+  required: z21.boolean(),
   /** Default value (optional, used when not provided) */
-  defaultValue: z20.union([z20.string(), z20.number(), z20.boolean()]).optional(),
+  defaultValue: z21.union([z21.string(), z21.number(), z21.boolean()]).optional(),
   /** If true, parameter is hidden by default behind "Show advanced options" */
-  advanced: z20.boolean().optional()
+  advanced: z21.boolean().optional()
 });
-var ScenarioAssertionLinkSchema = z20.object({
+var ScenarioAssertionLinkSchema = z21.object({
   /** ID of the assertion (can be system assertion like 'system:skill_was_called' or custom assertion UUID) */
-  assertionId: z20.string(),
+  assertionId: z21.string(),
   /** Parameter values for this assertion in this scenario */
-  params: z20.record(
-    z20.string(),
-    z20.union([z20.string(), z20.number(), z20.boolean(), z20.null()])
+  params: z21.record(
+    z21.string(),
+    z21.union([z21.string(), z21.number(), z21.boolean(), z21.null()])
   ).optional()
 });
-var SkillWasCalledConfigSchema = z20.object({
+var SkillWasCalledConfigSchema = z21.object({
   /** Name of the skill that must have been called */
-  skillName: z20.string().min(1)
+  skillName: z21.string().min(1)
 });
-var BuildPassedConfigSchema = z20.strictObject({
+var BuildPassedConfigSchema = z21.strictObject({
   /** Command to run (default: "yarn build") */
-  command: z20.string().optional(),
+  command: z21.string().optional(),
   /** Expected exit code (default: 0) */
-  expectedExitCode: z20.number().int().optional()
+  expectedExitCode: z21.number().int().optional()
 });
-var LlmJudgeConfigSchema = z20.object({
+var LlmJudgeConfigSchema = z21.object({
   /**
    * Prompt template with placeholders:
    * - {{output}}: agent's final output
@@ -514,28 +532,28 @@ var LlmJudgeConfigSchema = z20.object({
    * - {{trace}}: step-by-step trace of tool calls
    * - Custom parameters defined in the parameters array
    */
-  prompt: z20.string().min(1),
+  prompt: z21.string().min(1),
   /** Optional system prompt for the judge */
-  systemPrompt: z20.string().optional(),
+  systemPrompt: z21.string().optional(),
   /** Minimum score to pass (0-100, default 70) */
-  minScore: z20.number().int().min(0).max(100).optional(),
+  minScore: z21.number().int().min(0).max(100).optional(),
   /** Model for the judge (e.g. claude-3-5-haiku-20241022) */
-  model: z20.string().optional(),
+  model: z21.string().optional(),
   /** Max output tokens */
-  maxTokens: z20.number().int().optional(),
+  maxTokens: z21.number().int().optional(),
   /** Temperature (0-1) */
-  temperature: z20.number().min(0).max(1).optional(),
+  temperature: z21.number().min(0).max(1).optional(),
   /** User-defined parameters for this assertion */
-  parameters: z20.array(AssertionParameterSchema).optional()
+  parameters: z21.array(AssertionParameterSchema).optional()
 });
-var AssertionConfigSchema = z20.union([
+var AssertionConfigSchema = z21.union([
   LlmJudgeConfigSchema,
   // requires prompt - check first
   SkillWasCalledConfigSchema,
   // requires skillName
   BuildPassedConfigSchema,
   // all optional, uses strictObject to reject unknown keys
-  z20.object({})
+  z21.object({})
   // fallback empty config
 ]);
 var CustomAssertionSchema = TenantEntitySchema.extend({
@@ -580,23 +598,23 @@ function getLlmJudgeConfig(assertion) {
 }
 
 // src/scenario/test-scenario.ts
-var ExpectedFileSchema = z21.object({
+var ExpectedFileSchema = z22.object({
   /** Relative path where the file should be created */
-  path: z21.string(),
+  path: z22.string(),
   /** Optional expected content */
-  content: z21.string().optional()
+  content: z22.string().optional()
 });
 var TestScenarioSchema = TenantEntitySchema.extend({
   /** The prompt sent to the agent to trigger the task */
-  triggerPrompt: z21.string().min(10),
+  triggerPrompt: z22.string().min(10),
   /** ID of the template to use for this scenario (null = no template) */
-  templateId: z21.string().nullish(),
+  templateId: z22.string().nullish(),
   /** Inline assertions to evaluate for this scenario (legacy) */
-  assertions: z21.array(AssertionSchema).optional(),
+  assertions: z22.array(AssertionSchema).optional(),
   /** IDs of saved assertions to evaluate (from assertions table) - legacy, use assertionLinks */
-  assertionIds: z21.array(z21.string()).optional(),
+  assertionIds: z22.array(z22.string()).optional(),
   /** Linked assertions with per-scenario parameter values */
-  assertionLinks: z21.array(ScenarioAssertionLinkSchema).optional()
+  assertionLinks: z22.array(ScenarioAssertionLinkSchema).optional()
 });
 var CreateTestScenarioInputSchema = TestScenarioSchema.omit({
   id: true,
@@ -607,10 +625,10 @@ var CreateTestScenarioInputSchema = TestScenarioSchema.omit({
 var UpdateTestScenarioInputSchema = CreateTestScenarioInputSchema.partial();
 
 // src/suite/test-suite.ts
-import { z as z22 } from "zod";
+import { z as z23 } from "zod";
 var TestSuiteSchema = TenantEntitySchema.extend({
   /** IDs of test scenarios in this suite */
-  scenarioIds: z22.array(z22.string())
+  scenarioIds: z23.array(z23.string())
 });
 var CreateTestSuiteInputSchema = TestSuiteSchema.omit({
   id: true,
@@ -621,21 +639,21 @@ var CreateTestSuiteInputSchema = TestSuiteSchema.omit({
 var UpdateTestSuiteInputSchema = CreateTestSuiteInputSchema.partial();
 
 // src/evaluation/metrics.ts
-import { z as z23 } from "zod";
-var TokenUsageSchema = z23.object({
-  prompt: z23.number(),
-  completion: z23.number(),
-  total: z23.number()
-});
-var EvalMetricsSchema = z23.object({
-  totalAssertions: z23.number(),
-  passed: z23.number(),
-  failed: z23.number(),
-  skipped: z23.number(),
-  errors: z23.number(),
-  passRate: z23.number(),
-  avgDuration: z23.number(),
-  totalDuration: z23.number()
+import { z as z24 } from "zod";
+var TokenUsageSchema = z24.object({
+  prompt: z24.number(),
+  completion: z24.number(),
+  total: z24.number()
+});
+var EvalMetricsSchema = z24.object({
+  totalAssertions: z24.number(),
+  passed: z24.number(),
+  failed: z24.number(),
+  skipped: z24.number(),
+  errors: z24.number(),
+  passRate: z24.number(),
+  avgDuration: z24.number(),
+  totalDuration: z24.number()
 });
 var EvalStatus = /* @__PURE__ */ ((EvalStatus2) => {
   EvalStatus2["PENDING"] = "pending";
@@ -645,7 +663,7 @@ var EvalStatus = /* @__PURE__ */ ((EvalStatus2) => {
   EvalStatus2["CANCELLED"] = "cancelled";
   return EvalStatus2;
 })(EvalStatus || {});
-var EvalStatusSchema = z23.enum(EvalStatus);
+var EvalStatusSchema = z24.enum(EvalStatus);
 var LLMStepType = /* @__PURE__ */ ((LLMStepType2) => {
   LLMStepType2["COMPLETION"] = "completion";
   LLMStepType2["TOOL_USE"] = "tool_use";
@@ -653,52 +671,52 @@ var LLMStepType = /* @__PURE__ */ ((LLMStepType2) => {
   LLMStepType2["THINKING"] = "thinking";
   return LLMStepType2;
 })(LLMStepType || {});
-var LLMTraceStepSchema = z23.object({
-  id: z23.string(),
-  stepNumber: z23.number(),
-  type: z23.enum(LLMStepType),
-  model: z23.string(),
-  provider: z23.string(),
-  startedAt: z23.string(),
-  durationMs: z23.number(),
+var LLMTraceStepSchema = z24.object({
+  id: z24.string(),
+  stepNumber: z24.number(),
+  type: z24.enum(LLMStepType),
+  model: z24.string(),
+  provider: z24.string(),
+  startedAt: z24.string(),
+  durationMs: z24.number(),
   tokenUsage: TokenUsageSchema,
-  costUsd: z23.number(),
-  toolName: z23.string().optional(),
-  toolArguments: z23.string().optional(),
-  inputPreview: z23.string().optional(),
-  outputPreview: z23.string().optional(),
-  success: z23.boolean(),
-  error: z23.string().optional()
-});
-var LLMBreakdownStatsSchema = z23.object({
-  count: z23.number(),
-  durationMs: z23.number(),
-  tokens: z23.number(),
-  costUsd: z23.number()
-});
-var LLMTraceSummarySchema = z23.object({
-  totalSteps: z23.number(),
-  totalDurationMs: z23.number(),
+  costUsd: z24.number(),
+  toolName: z24.string().optional(),
+  toolArguments: z24.string().optional(),
+  inputPreview: z24.string().optional(),
+  outputPreview: z24.string().optional(),
+  success: z24.boolean(),
+  error: z24.string().optional()
+});
+var LLMBreakdownStatsSchema = z24.object({
+  count: z24.number(),
+  durationMs: z24.number(),
+  tokens: z24.number(),
+  costUsd: z24.number()
+});
+var LLMTraceSummarySchema = z24.object({
+  totalSteps: z24.number(),
+  totalDurationMs: z24.number(),
   totalTokens: TokenUsageSchema,
-  totalCostUsd: z23.number(),
-  stepTypeBreakdown: z23.record(z23.string(), LLMBreakdownStatsSchema).optional(),
-  modelBreakdown: z23.record(z23.string(), LLMBreakdownStatsSchema),
-  modelsUsed: z23.array(z23.string())
-});
-var LLMTraceSchema = z23.object({
-  id: z23.string(),
-  steps: z23.array(LLMTraceStepSchema),
+  totalCostUsd: z24.number(),
+  stepTypeBreakdown: z24.record(z24.string(), LLMBreakdownStatsSchema).optional(),
+  modelBreakdown: z24.record(z24.string(), LLMBreakdownStatsSchema),
+  modelsUsed: z24.array(z24.string())
+});
+var LLMTraceSchema = z24.object({
+  id: z24.string(),
+  steps: z24.array(LLMTraceStepSchema),
   summary: LLMTraceSummarySchema
 });
 
 // src/evaluation/eval-result.ts
-import { z as z26 } from "zod";
+import { z as z27 } from "zod";
 
 // src/evaluation/eval-run.ts
-import { z as z25 } from "zod";
+import { z as z26 } from "zod";
 
 // src/evaluation/live-trace.ts
-import { z as z24 } from "zod";
+import { z as z25 } from "zod";
 var LiveTraceEventType = /* @__PURE__ */ ((LiveTraceEventType2) => {
   LiveTraceEventType2["THINKING"] = "thinking";
   LiveTraceEventType2["TOOL_USE"] = "tool_use";
@@ -712,37 +730,37 @@ var LiveTraceEventType = /* @__PURE__ */ ((LiveTraceEventType2) => {
   LiveTraceEventType2["USER"] = "user";
   return LiveTraceEventType2;
 })(LiveTraceEventType || {});
-var LiveTraceEventSchema = z24.object({
+var LiveTraceEventSchema = z25.object({
   /** The evaluation run ID */
-  evalRunId: z24.string(),
+  evalRunId: z25.string(),
   /** The scenario ID being executed */
-  scenarioId: z24.string(),
+  scenarioId: z25.string(),
   /** The scenario name for display */
-  scenarioName: z24.string(),
+  scenarioName: z25.string(),
   /** The target ID (skill, agent, etc.) */
-  targetId: z24.string(),
+  targetId: z25.string(),
   /** The target name for display */
-  targetName: z24.string(),
+  targetName: z25.string(),
   /** Step number in the current scenario execution */
-  stepNumber: z24.number(),
+  stepNumber: z25.number(),
   /** Type of trace event */
-  type: z24.enum(LiveTraceEventType),
+  type: z25.enum(LiveTraceEventType),
   /** Tool name if this is a tool_use event */
-  toolName: z24.string().optional(),
+  toolName: z25.string().optional(),
   /** Tool arguments preview (truncated JSON) */
-  toolArgs: z24.string().optional(),
+  toolArgs: z25.string().optional(),
   /** Output preview (truncated text) */
-  outputPreview: z24.string().optional(),
+  outputPreview: z25.string().optional(),
   /** File path for file operations */
-  filePath: z24.string().optional(),
+  filePath: z25.string().optional(),
   /** Elapsed time in milliseconds for progress events */
-  elapsedMs: z24.number().optional(),
+  elapsedMs: z25.number().optional(),
   /** Thinking/reasoning text from Claude */
-  thinking: z24.string().optional(),
+  thinking: z25.string().optional(),
   /** Timestamp when this event occurred */
-  timestamp: z24.string(),
+  timestamp: z25.string(),
   /** Whether this is the final event for this scenario */
-  isComplete: z24.boolean()
+  isComplete: z25.boolean()
 });
 var TRACE_EVENT_PREFIX = "TRACE_EVENT:";
 function parseTraceEventLine(line) {
@@ -770,14 +788,14 @@ var TriggerType = /* @__PURE__ */ ((TriggerType2) => {
   TriggerType2["MANUAL"] = "MANUAL";
   return TriggerType2;
 })(TriggerType || {});
-var TriggerMetadataSchema = z25.object({
-  version: z25.string().optional(),
-  resourceUpdated: z25.array(z25.string()).optional()
+var TriggerMetadataSchema = z26.object({
+  version: z26.string().optional(),
+  resourceUpdated: z26.array(z26.string()).optional()
 });
-var TriggerSchema = z25.object({
-  id: z25.string(),
+var TriggerSchema = z26.object({
+  id: z26.string(),
   metadata: TriggerMetadataSchema.optional(),
-  type: z25.enum(TriggerType)
+  type: z26.enum(TriggerType)
 });
 var FailureCategory = /* @__PURE__ */ ((FailureCategory2) => {
   FailureCategory2["MISSING_FILE"] = "missing_file";
@@ -795,28 +813,28 @@ var FailureSeverity = /* @__PURE__ */ ((FailureSeverity2) => {
   FailureSeverity2["LOW"] = "low";
   return FailureSeverity2;
 })(FailureSeverity || {});
-var DiffLineTypeSchema = z25.enum(["added", "removed", "unchanged"]);
-var DiffLineSchema = z25.object({
+var DiffLineTypeSchema = z26.enum(["added", "removed", "unchanged"]);
+var DiffLineSchema = z26.object({
   type: DiffLineTypeSchema,
-  content: z25.string(),
-  lineNumber: z25.number()
-});
-var DiffContentSchema = z25.object({
-  path: z25.string(),
-  expected: z25.string(),
-  actual: z25.string(),
-  diffLines: z25.array(DiffLineSchema),
-  renamedFrom: z25.string().optional()
-});
-var CommandExecutionSchema = z25.object({
-  command: z25.string(),
-  exitCode: z25.number(),
-  output: z25.string().optional(),
-  duration: z25.number()
-});
-var FileModificationSchema = z25.object({
-  path: z25.string(),
-  action: z25.enum(["created", "modified", "deleted"])
+  content: z26.string(),
+  lineNumber: z26.number()
+});
+var DiffContentSchema = z26.object({
+  path: z26.string(),
+  expected: z26.string(),
+  actual: z26.string(),
+  diffLines: z26.array(DiffLineSchema),
+  renamedFrom: z26.string().optional()
+});
+var CommandExecutionSchema = z26.object({
+  command: z26.string(),
+  exitCode: z26.number(),
+  output: z26.string().optional(),
+  duration: z26.number()
+});
+var FileModificationSchema = z26.object({
+  path: z26.string(),
+  action: z26.enum(["created", "modified", "deleted"])
 });
 var TemplateFileStatus = /* @__PURE__ */ ((TemplateFileStatus2) => {
   TemplateFileStatus2["NEW"] = "new";
@@ -824,75 +842,79 @@ var TemplateFileStatus = /* @__PURE__ */ ((TemplateFileStatus2) => {
   TemplateFileStatus2["UNCHANGED"] = "unchanged";
   return TemplateFileStatus2;
 })(TemplateFileStatus || {});
-var TemplateFileSchema = z25.object({
+var TemplateFileSchema = z26.object({
   /** Relative path within the template */
-  path: z25.string(),
+  path: z26.string(),
   /** Full file content after execution */
-  content: z25.string(),
+  content: z26.string(),
   /** File status (new, modified, unchanged) */
-  status: z25.enum(["new", "modified", "unchanged"])
-});
-var ApiCallSchema = z25.object({
-  endpoint: z25.string(),
-  tokensUsed: z25.number(),
-  duration: z25.number()
-});
-var ExecutionTraceSchema = z25.object({
-  commands: z25.array(CommandExecutionSchema),
-  filesModified: z25.array(FileModificationSchema),
-  apiCalls: z25.array(ApiCallSchema),
-  totalDuration: z25.number()
-});
-var FailureAnalysisSchema = z25.object({
-  category: z25.enum(FailureCategory),
-  severity: z25.enum(FailureSeverity),
-  summary: z25.string(),
-  details: z25.string(),
-  rootCause: z25.string(),
-  suggestedFix: z25.string(),
-  relatedAssertions: z25.array(z25.string()),
-  codeSnippet: z25.string().optional(),
-  similarIssues: z25.array(z25.string()).optional(),
-  patternId: z25.string().optional(),
+  status: z26.enum(["new", "modified", "unchanged"])
+});
+var ApiCallSchema = z26.object({
+  endpoint: z26.string(),
+  tokensUsed: z26.number(),
+  duration: z26.number()
+});
+var ExecutionTraceSchema = z26.object({
+  commands: z26.array(CommandExecutionSchema),
+  filesModified: z26.array(FileModificationSchema),
+  apiCalls: z26.array(ApiCallSchema),
+  totalDuration: z26.number()
+});
+var FailureAnalysisSchema = z26.object({
+  category: z26.enum(FailureCategory),
+  severity: z26.enum(FailureSeverity),
+  summary: z26.string(),
+  details: z26.string(),
+  rootCause: z26.string(),
+  suggestedFix: z26.string(),
+  relatedAssertions: z26.array(z26.string()),
+  codeSnippet: z26.string().optional(),
+  similarIssues: z26.array(z26.string()).optional(),
+  patternId: z26.string().optional(),
   // Extended fields for detailed debugging
   diff: DiffContentSchema.optional(),
   executionTrace: ExecutionTraceSchema.optional()
 });
 var EvalRunSchema = TenantEntitySchema.extend({
   /** Agent ID for this run */
-  agentId: z25.string().optional(),
+  agentId: z26.string().optional(),
   /** Skills group ID for this run */
-  skillsGroupId: z25.string().optional(),
+  skillsGroupId: z26.string().optional(),
   /** Scenario IDs to run */
-  scenarioIds: z25.array(z25.string()),
+  scenarioIds: z26.array(z26.string()),
   /** Current status */
   status: EvalStatusSchema,
   /** Progress percentage (0-100) */
-  progress: z25.number(),
+  progress: z26.number(),
   /** Results for each scenario/target combination (lazy to break eval-result ↔ eval-run cycle) */
-  results: z25.array(z25.lazy(() => EvalRunResultSchema)),
+  results: z26.array(z26.lazy(() => EvalRunResultSchema)),
   /** Aggregated metrics across all results */
   aggregateMetrics: EvalMetricsSchema,
   /** Failure analyses */
-  failureAnalyses: z25.array(FailureAnalysisSchema).optional(),
+  failureAnalyses: z26.array(FailureAnalysisSchema).optional(),
   /** Aggregated LLM trace summary */
   llmTraceSummary: LLMTraceSummarySchema.optional(),
   /** What triggered this run */
   trigger: TriggerSchema.optional(),
   /** When the run started (set when evaluation is triggered) */
-  startedAt: z25.string().optional(),
+  startedAt: z26.string().optional(),
   /** When the run completed */
-  completedAt: z25.string().optional(),
+  completedAt: z26.string().optional(),
   /** Live trace events captured during execution (for playback on results page) */
-  liveTraceEvents: z25.array(LiveTraceEventSchema).optional(),
+  liveTraceEvents: z26.array(LiveTraceEventSchema).optional(),
   /** Remote job ID for tracking execution in Dev Machines */
-  jobId: z25.string().optional(),
+  jobId: z26.string().optional(),
   /** Remote job status from the Dev Machine API (PENDING, RUNNING, COMPLETED, FAILED, CANCELLED) */
-  jobStatus: z25.string().optional(),
+  jobStatus: z26.string().optional(),
   /** Remote job error message if the job failed */
-  jobError: z25.string().optional(),
+  jobError: z26.string().optional(),
   /** Timestamp of the last job status check */
-  jobStatusCheckedAt: z25.string().optional()
+  jobStatusCheckedAt: z26.string().optional(),
+  /** MCP server IDs to enable for this run (optional) */
+  mcpIds: z26.array(z26.string()).optional(),
+  /** Sub-agent IDs to enable for this run (optional) */
+  subAgentIds: z26.array(z26.string()).optional()
 });
 var CreateEvalRunInputSchema = EvalRunSchema.omit({
   id: true,
@@ -905,28 +927,28 @@ var CreateEvalRunInputSchema = EvalRunSchema.omit({
   startedAt: true,
   completedAt: true
 });
-var EvaluationProgressSchema = z25.object({
-  runId: z25.string(),
-  targetId: z25.string(),
-  totalScenarios: z25.number(),
-  completedScenarios: z25.number(),
-  scenarioProgress: z25.array(
-    z25.object({
-      scenarioId: z25.string(),
-      currentStep: z25.string(),
-      error: z25.string().optional()
+var EvaluationProgressSchema = z26.object({
+  runId: z26.string(),
+  targetId: z26.string(),
+  totalScenarios: z26.number(),
+  completedScenarios: z26.number(),
+  scenarioProgress: z26.array(
+    z26.object({
+      scenarioId: z26.string(),
+      currentStep: z26.string(),
+      error: z26.string().optional()
     })
   ),
-  createdAt: z25.number()
+  createdAt: z26.number()
 });
-var EvaluationLogSchema = z25.object({
-  runId: z25.string(),
-  scenarioId: z25.string(),
-  log: z25.object({
-    level: z25.enum(["info", "error", "debug"]),
-    message: z25.string().optional(),
-    args: z25.array(z25.any()).optional(),
-    error: z25.string().optional()
+var EvaluationLogSchema = z26.object({
+  runId: z26.string(),
+  scenarioId: z26.string(),
+  log: z26.object({
+    level: z26.enum(["info", "error", "debug"]),
+    message: z26.string().optional(),
+    args: z26.array(z26.any()).optional(),
+    error: z26.string().optional()
   })
 });
 var LLM_TIMEOUT = 12e4;
@@ -939,91 +961,91 @@ var AssertionResultStatus = /* @__PURE__ */ ((AssertionResultStatus2) => {
   AssertionResultStatus2["ERROR"] = "error";
   return AssertionResultStatus2;
 })(AssertionResultStatus || {});
-var AssertionResultSchema = z26.object({
-  id: z26.string(),
-  assertionId: z26.string(),
-  assertionType: z26.string(),
-  assertionName: z26.string(),
-  status: z26.enum(AssertionResultStatus),
-  message: z26.string().optional(),
-  expected: z26.string().optional(),
-  actual: z26.string().optional(),
-  duration: z26.number().optional(),
-  details: z26.record(z26.string(), z26.unknown()).optional(),
-  llmTraceSteps: z26.array(LLMTraceStepSchema).optional()
-});
-var EvalRunResultSchema = z26.object({
-  id: z26.string(),
-  targetId: z26.string(),
-  targetName: z26.string().optional(),
-  scenarioId: z26.string(),
-  scenarioName: z26.string(),
+var AssertionResultSchema = z27.object({
+  id: z27.string(),
+  assertionId: z27.string(),
+  assertionType: z27.string(),
+  assertionName: z27.string(),
+  status: z27.enum(AssertionResultStatus),
+  message: z27.string().optional(),
+  expected: z27.string().optional(),
+  actual: z27.string().optional(),
+  duration: z27.number().optional(),
+  details: z27.record(z27.string(), z27.unknown()).optional(),
+  llmTraceSteps: z27.array(LLMTraceStepSchema).optional()
+});
+var EvalRunResultSchema = z27.object({
+  id: z27.string(),
+  targetId: z27.string(),
+  targetName: z27.string().optional(),
+  scenarioId: z27.string(),
+  scenarioName: z27.string(),
   modelConfig: ModelConfigSchema.optional(),
-  assertionResults: z26.array(AssertionResultSchema),
+  assertionResults: z27.array(AssertionResultSchema),
   metrics: EvalMetricsSchema.optional(),
-  passed: z26.number(),
-  failed: z26.number(),
-  passRate: z26.number(),
-  duration: z26.number(),
-  outputText: z26.string().optional(),
-  files: z26.array(ExpectedFileSchema).optional(),
-  fileDiffs: z26.array(DiffContentSchema).optional(),
+  passed: z27.number(),
+  failed: z27.number(),
+  passRate: z27.number(),
+  duration: z27.number(),
+  outputText: z27.string().optional(),
+  files: z27.array(ExpectedFileSchema).optional(),
+  fileDiffs: z27.array(DiffContentSchema).optional(),
   /** Full template files after execution with status indicators */
-  templateFiles: z26.array(TemplateFileSchema).optional(),
-  startedAt: z26.string().optional(),
-  completedAt: z26.string().optional(),
+  templateFiles: z27.array(TemplateFileSchema).optional(),
+  startedAt: z27.string().optional(),
+  completedAt: z27.string().optional(),
   llmTrace: LLMTraceSchema.optional()
 });
-var PromptResultSchema = z26.object({
-  text: z26.string(),
-  files: z26.array(z26.unknown()).optional(),
-  finishReason: z26.string().optional(),
-  reasoning: z26.string().optional(),
-  reasoningDetails: z26.unknown().optional(),
-  toolCalls: z26.array(z26.unknown()).optional(),
-  toolResults: z26.array(z26.unknown()).optional(),
-  warnings: z26.array(z26.unknown()).optional(),
-  sources: z26.array(z26.unknown()).optional(),
-  steps: z26.array(z26.unknown()),
-  generationTimeMs: z26.number(),
-  prompt: z26.string(),
-  systemPrompt: z26.string(),
-  usage: z26.object({
-    totalTokens: z26.number().optional(),
-    totalMicrocentsSpent: z26.number().optional()
+var PromptResultSchema = z27.object({
+  text: z27.string(),
+  files: z27.array(z27.unknown()).optional(),
+  finishReason: z27.string().optional(),
+  reasoning: z27.string().optional(),
+  reasoningDetails: z27.unknown().optional(),
+  toolCalls: z27.array(z27.unknown()).optional(),
+  toolResults: z27.array(z27.unknown()).optional(),
+  warnings: z27.array(z27.unknown()).optional(),
+  sources: z27.array(z27.unknown()).optional(),
+  steps: z27.array(z27.unknown()),
+  generationTimeMs: z27.number(),
+  prompt: z27.string(),
+  systemPrompt: z27.string(),
+  usage: z27.object({
+    totalTokens: z27.number().optional(),
+    totalMicrocentsSpent: z27.number().optional()
   })
 });
-var EvaluationResultSchema = z26.object({
-  id: z26.string(),
-  runId: z26.string(),
-  timestamp: z26.number(),
+var EvaluationResultSchema = z27.object({
+  id: z27.string(),
+  runId: z27.string(),
+  timestamp: z27.number(),
   promptResult: PromptResultSchema,
-  testResults: z26.array(z26.unknown()),
-  tags: z26.array(z26.string()).optional(),
-  feedback: z26.string().optional(),
-  score: z26.number(),
-  suiteId: z26.string().optional()
-});
-var LeanEvaluationResultSchema = z26.object({
-  id: z26.string(),
-  runId: z26.string(),
-  timestamp: z26.number(),
-  tags: z26.array(z26.string()).optional(),
-  scenarioId: z26.string(),
-  scenarioVersion: z26.number().optional(),
-  targetId: z26.string(),
-  targetVersion: z26.number().optional(),
-  suiteId: z26.string().optional(),
-  score: z26.number(),
-  time: z26.number().optional(),
-  microcentsSpent: z26.number().optional()
+  testResults: z27.array(z27.unknown()),
+  tags: z27.array(z27.string()).optional(),
+  feedback: z27.string().optional(),
+  score: z27.number(),
+  suiteId: z27.string().optional()
+});
+var LeanEvaluationResultSchema = z27.object({
+  id: z27.string(),
+  runId: z27.string(),
+  timestamp: z27.number(),
+  tags: z27.array(z27.string()).optional(),
+  scenarioId: z27.string(),
+  scenarioVersion: z27.number().optional(),
+  targetId: z27.string(),
+  targetVersion: z27.number().optional(),
+  suiteId: z27.string().optional(),
+  score: z27.number(),
+  time: z27.number().optional(),
+  microcentsSpent: z27.number().optional()
 });
 
 // src/project/project.ts
-import { z as z27 } from "zod";
+import { z as z28 } from "zod";
 var ProjectSchema = BaseEntitySchema.extend({
-  appId: z27.string().optional().describe("The ID of the app in Dev Center"),
-  appSecret: z27.string().optional().describe("The secret of the app in Dev Center")
+  appId: z28.string().optional().describe("The ID of the app in Dev Center"),
+  appSecret: z28.string().optional().describe("The secret of the app in Dev Center")
 });
 var CreateProjectInputSchema = ProjectSchema.omit({
   id: true,
@@ -1034,10 +1056,10 @@ var CreateProjectInputSchema = ProjectSchema.omit({
 var UpdateProjectInputSchema = CreateProjectInputSchema.partial();
 
 // src/template/template.ts
-import { z as z28 } from "zod";
+import { z as z29 } from "zod";
 var TemplateSchema = TenantEntitySchema.extend({
   /** URL to download the template from */
-  downloadUrl: z28.url()
+  downloadUrl: z29.url()
 });
 var CreateTemplateInputSchema = TemplateSchema.omit({
   id: true,
@@ -1172,9 +1194,11 @@ export {
   CreateAgentInputSchema,
   CreateCustomAssertionInputSchema,
   CreateEvalRunInputSchema,
+  CreateMcpInputSchema,
   CreateProjectInputSchema,
   CreateSkillInputSchema,
   CreateSkillsGroupInputSchema,
+  CreateSubAgentInputSchema,
   CreateTemplateInputSchema,
   CreateTestScenarioInputSchema,
   CreateTestSuiteInputSchema,
@@ -1213,7 +1237,9 @@ export {
   LlmJudgeAssertionSchema,
   LlmJudgeConfigSchema,
   LocalProjectConfigSchema,
+  MCPEntitySchema,
   MCPServerConfigSchema,
+  MCP_SERVERS_JSON_KEY,
   MetaSiteConfigSchema,
   ModelConfigSchema,
   ModelIds,
@@ -1234,6 +1260,7 @@ export {
   SkillWasCalledAssertionSchema,
   SkillWasCalledConfigSchema,
   SkillsGroupSchema,
+  SubAgentSchema,
   TRACE_EVENT_PREFIX,
   TargetSchema,
   TemplateFileSchema,
@@ -1254,9 +1281,11 @@ export {
   TriggerType,
   UpdateAgentInputSchema,
   UpdateCustomAssertionInputSchema,
+  UpdateMcpInputSchema,
   UpdateProjectInputSchema,
   UpdateSkillInputSchema,
   UpdateSkillsGroupInputSchema,
+  UpdateSubAgentInputSchema,
   UpdateTemplateInputSchema,
   UpdateTestScenarioInputSchema,
   UpdateTestSuiteInputSchema,