@lightsage/mcp 0.3.0 → 0.4.0

package/src/models/updatev2evalrequest.ts

@@ -4,44 +4,75 @@
  */
 
 import * as z from "zod";
-import { ClosedEnum } from "../types/enums.js";
+import {
+  SearchGroundingMode,
+  SearchGroundingMode$zodSchema,
+} from "./searchgroundingmode.js";
+import {
+  V2InlineMcpDefinition,
+  V2InlineMcpDefinition$zodSchema,
+} from "./v2inlinemcpdefinition.js";
+import {
+  V2SavedMcpReference,
+  V2SavedMcpReference$zodSchema,
+} from "./v2savedmcpreference.js";
 
-export const UpdateV2EvalRequestType = {
-  Workflow: "workflow",
-  Mcp: "mcp",
-  Cli: "cli",
-  Skill: "skill",
-} as const;
-export type UpdateV2EvalRequestType = ClosedEnum<
-  typeof UpdateV2EvalRequestType
->;
+export type UpdateV2EvalRequestMcp =
+  | V2InlineMcpDefinition
+  | V2SavedMcpReference;
 
-export const UpdateV2EvalRequestType$zodSchema = z.enum([
-  "workflow",
-  "mcp",
-  "cli",
-  "skill",
+export const UpdateV2EvalRequestMcp$zodSchema: z.ZodType<
+  UpdateV2EvalRequestMcp
+> = z.union([
+  V2InlineMcpDefinition$zodSchema,
+  V2SavedMcpReference$zodSchema,
 ]);
 
 export type UpdateV2EvalRequest = {
   name?: string | undefined;
-  type?: UpdateV2EvalRequestType | undefined;
-  prompt?: string | undefined;
-  judge?: string | undefined;
-  mcp_server_id?: string | undefined;
-  cli_id?: string | undefined;
-  skill_id?: string | undefined;
+  prompts?: Array<string> | null | undefined;
+  judges?: Array<string> | null | undefined;
+  agent?: Array<string> | null | undefined;
+  persona?: string | undefined;
+  clis?: Array<string> | null | undefined;
+  mcps?: Array<V2InlineMcpDefinition | V2SavedMcpReference> | null | undefined;
+  setup_commands?: Array<string> | null | undefined;
+  cleanup_commands?: Array<string> | null | undefined;
+  search_grounding?: SearchGroundingMode | null | undefined;
+  runs?: number | undefined;
+  tags?: Array<string> | null | undefined;
 };
 
 export const UpdateV2EvalRequest$zodSchema: z.ZodType<UpdateV2EvalRequest> = z
   .object({
-    cli_id: z.string().optional(),
-    judge: z.string().optional().describe(
-      "Plain-text outcome that counts as success.",
+    agent: z.array(z.string()).nullable().optional(),
+    cleanup_commands: z.array(z.string()).nullable().optional().describe(
+      "Replacement cleanup commands. Pass an empty array to clear all saved cleanup commands.",
     ),
-    mcp_server_id: z.string().optional(),
+    clis: z.array(z.string()).nullable().optional().describe(
+      "Replacement saved CLI IDs or shell installation commands. UUID values resolve saved CLIs; every other value is an installation command. Pass an empty array to clear all CLIs.",
+    ),
+    judges: z.array(z.string()).nullable().optional().describe(
+      "Replacement success criteria for the eval.",
+    ),
+    mcps: z.array(z.union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema,
+    ])).nullable().optional(),
     name: z.string().optional(),
-    prompt: z.string().optional(),
-    skill_id: z.string().optional(),
-    type: UpdateV2EvalRequestType$zodSchema.optional(),
+    persona: z.string().optional().describe(
+      "Replacement eval persona. Pass null to clear it; omit it to leave the saved value unchanged.",
+    ),
+    prompts: z.array(z.string()).nullable().optional(),
+    runs: z.int().optional(),
+    search_grounding: SearchGroundingMode$zodSchema.nullable().optional()
+      .describe(
+        "Replacement web-retrieval policy. Omit it to keep the saved value.",
+      ),
+    setup_commands: z.array(z.string()).nullable().optional().describe(
+      "Replacement setup commands. Pass an empty array to clear all saved setup commands.",
+    ),
+    tags: z.array(z.string()).nullable().optional().describe(
+      "Replacement labels for the eval. Pass an empty array to clear all tags.",
+    ),
   });

package/src/models/v2eval.ts

@@ -4,46 +4,81 @@
  */
 
 import * as z from "zod";
-import { ClosedEnum } from "../types/enums.js";
-import { V2EvalJudge, V2EvalJudge$zodSchema } from "./v2evaljudge.js";
+import {
+  SearchGroundingMode,
+  SearchGroundingMode$zodSchema,
+} from "./searchgroundingmode.js";
+import {
+  V2InlineCliReference,
+  V2InlineCliReference$zodSchema,
+} from "./v2inlineclireference.js";
+import {
+  V2InlineMcpDefinition,
+  V2InlineMcpDefinition$zodSchema,
+} from "./v2inlinemcpdefinition.js";
+import {
+  V2SavedCliReference,
+  V2SavedCliReference$zodSchema,
+} from "./v2savedclireference.js";
+import {
+  V2SavedMcpReference,
+  V2SavedMcpReference$zodSchema,
+} from "./v2savedmcpreference.js";
 
-export const V2EvalType = {
-  Workflow: "workflow",
-  Mcp: "mcp",
-  Cli: "cli",
-  Skill: "skill",
-} as const;
-export type V2EvalType = ClosedEnum<typeof V2EvalType>;
+export type Cli = V2SavedCliReference | V2InlineCliReference;
 
-export const V2EvalType$zodSchema = z.enum([
-  "workflow",
-  "mcp",
-  "cli",
-  "skill",
+export const Cli$zodSchema: z.ZodType<Cli> = z.union([
+  V2SavedCliReference$zodSchema,
+  V2InlineCliReference$zodSchema,
+]);
+
+export type V2EvalMcp = V2InlineMcpDefinition | V2SavedMcpReference;
+
+export const V2EvalMcp$zodSchema: z.ZodType<V2EvalMcp> = z.union([
+  V2InlineMcpDefinition$zodSchema,
+  V2SavedMcpReference$zodSchema,
 ]);
 
 export type V2Eval = {
   id: string;
   name: string;
-  type: V2EvalType;
-  prompt: string;
-  judge: V2EvalJudge;
-  mcp_server_id?: string | null | undefined;
-  cli_id?: string | null | undefined;
-  skill_id?: string | null | undefined;
+  prompts?: Array<string> | undefined;
+  judges?: Array<string> | undefined;
+  agent?: Array<string> | undefined;
+  persona?: string | null | undefined;
+  clis?: Array<V2SavedCliReference | V2InlineCliReference> | undefined;
+  mcps?: Array<V2InlineMcpDefinition | V2SavedMcpReference> | undefined;
+  setup_commands?: Array<string> | undefined;
+  cleanup_commands?: Array<string> | undefined;
+  search_grounding?: SearchGroundingMode | undefined;
+  runs?: number | undefined;
+  tags?: Array<string> | undefined;
   created_at?: string | null | undefined;
   updated_at?: string | null | undefined;
 };
 
 export const V2Eval$zodSchema: z.ZodType<V2Eval> = z.object({
-  cli_id: z.string().nullable().optional(),
+  agent: z.array(z.string()).optional(),
+  cleanup_commands: z.array(z.string()).optional(),
+  clis: z.array(z.union([
+    V2SavedCliReference$zodSchema,
+    V2InlineCliReference$zodSchema,
+  ])).optional(),
   created_at: z.string().nullable().optional(),
   id: z.string(),
-  judge: V2EvalJudge$zodSchema,
-  mcp_server_id: z.string().nullable().optional(),
+  judges: z.array(z.string()).optional(),
+  mcps: z.array(z.union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema,
+  ])).optional(),
   name: z.string(),
-  prompt: z.string(),
-  skill_id: z.string().nullable().optional(),
-  type: V2EvalType$zodSchema,
+  persona: z.string().nullable().optional(),
+  prompts: z.array(z.string()).optional(),
+  runs: z.int().default(1),
+  search_grounding: SearchGroundingMode$zodSchema.optional().describe(
+    "Requested web-retrieval behavior for an eval run.",
+  ),
+  setup_commands: z.array(z.string()).optional(),
+  tags: z.array(z.string()).optional(),
   updated_at: z.string().nullable().optional(),
 });

package/src/models/v2evalconfiguration.ts

@@ -4,6 +4,10 @@
  */
 
 import * as z from "zod";
+import {
+  SearchGroundingMode,
+  SearchGroundingMode$zodSchema,
+} from "./searchgroundingmode.js";
 
 export type V2EvalConfiguration = {
   id: string;
@@ -16,6 +20,7 @@ export type V2EvalConfiguration = {
   skill_ids?: Array<string> | undefined;
   skills?: Array<string> | undefined;
   include_docs?: boolean | undefined;
+  search_grounding?: SearchGroundingMode | undefined;
   created_at?: string | null | undefined;
   updated_at?: string | null | undefined;
 };
@@ -31,6 +36,9 @@ export const V2EvalConfiguration$zodSchema: z.ZodType<V2EvalConfiguration> = z
     mcp_server_ids: z.array(z.string()).optional(),
     name: z.string(),
     repository_id: z.string().nullable().optional(),
+    search_grounding: SearchGroundingMode$zodSchema.optional().describe(
+      "Requested web-retrieval behavior for an eval run.",
+    ),
     skill_ids: z.array(z.string()).optional(),
     skills: z.array(z.string()).optional().describe(
       "Public GitHub `owner/repository` skill packages installed for this configuration.",

package/src/models/v2evalresultsummary.ts

@@ -4,28 +4,69 @@
  */
 
 import * as z from "zod";
+import {
+  PublicEvalRunEnvironment,
+  PublicEvalRunEnvironment$zodSchema,
+} from "./publicevalrunenvironment.js";
+import {
+  PublicGraderResult,
+  PublicGraderResult$zodSchema,
+} from "./publicgraderresult.js";
+import {
+  SearchGroundingMode,
+  SearchGroundingMode$zodSchema,
+} from "./searchgroundingmode.js";
 
 export type V2EvalResultSummary = {
   id: string;
   eval_id?: string | null | undefined;
-  agent_id?: string | null | undefined;
+  agent?: string | null | undefined;
+  prompt?: string | null | undefined;
+  prompt_index?: number | undefined;
+  repeat_index?: number | undefined;
   passed?: boolean | null | undefined;
   score?: number | null | undefined;
   failure_reasons?: Array<string> | null | undefined;
+  grader_results?: Array<PublicGraderResult> | undefined;
   latency_ms?: number | null | undefined;
+  search_grounding?: SearchGroundingMode | null | undefined;
+  web_retrieval_observed?: boolean | null | undefined;
+  search_observed?: boolean | null | undefined;
+  grounding_qualified?: boolean | null | undefined;
+  grounding_qualification_reason?: string | null | undefined;
   created_at?: string | null | undefined;
+  environment?: PublicEvalRunEnvironment | null | undefined;
 };
 
 export const V2EvalResultSummary$zodSchema: z.ZodType<V2EvalResultSummary> = z
   .object({
-    agent_id: z.string().nullable().optional().describe(
-      "Runnable agent ID combining the harness and model.",
+    agent: z.string().nullable().optional().describe(
+      "Runnable agent combining the harness and model.",
     ),
     created_at: z.string().nullable().optional(),
+    environment: PublicEvalRunEnvironment$zodSchema.nullable().optional()
+      .describe(
+        "The environment this attempt used, without secret values. Null when the attempt had no environment.",
+      ),
     eval_id: z.string().nullable().optional(),
-    failure_reasons: z.array(z.string()).nullable().optional(),
+    failure_reasons: z.array(z.string()).nullable().optional().describe(
+      "Checks that failed, listed even when the result passed overall. Deductions from checks that passed are in `grader_results[].deductions`, not here.",
+    ),
+    grader_results: z.array(PublicGraderResult$zodSchema).optional().describe(
+      "Per-check verdicts behind `score`, including deductions recorded by passing checks.",
+    ),
+    grounding_qualification_reason: z.string().nullable().optional(),
+    grounding_qualified: z.boolean().nullable().optional().describe(
+      "Whether the attempt satisfied the requested grounding experiment. Null means the default mode was used or isolation could not be proven.",
+    ),
     id: z.string().describe("Result ID, not an eval job ID."),
     latency_ms: z.int().nullable().optional(),
     passed: z.boolean().nullable().optional(),
+    prompt: z.string().nullable().optional(),
+    prompt_index: z.int().default(0),
+    repeat_index: z.int().default(0),
     score: z.number().nullable().optional().describe("Score from 0 to 1."),
+    search_grounding: SearchGroundingMode$zodSchema.nullable().optional(),
+    search_observed: z.boolean().nullable().optional(),
+    web_retrieval_observed: z.boolean().nullable().optional(),
   });

package/src/models/v2evalrun.ts

@@ -15,8 +15,11 @@ export type V2EvalRun = {
   eval_id?: string | null | undefined;
   eval_name?: string | null | undefined;
   configuration_id?: string | null | undefined;
-  agent_id?: string | null | undefined;
-  agent_name?: string | null | undefined;
+  persona?: string | null | undefined;
+  agent?: Array<string> | undefined;
+  prompt_count?: number | undefined;
+  runs?: number | undefined;
+  attempt_count?: number | undefined;
   status: string;
   stage?: string | null | undefined;
   dispatch_id?: string | null | undefined;
@@ -29,11 +32,11 @@ export type V2EvalRun = {
 };
 
 export const V2EvalRun$zodSchema: z.ZodType<V2EvalRun> = z.object({
-  agent_id: z.string().nullable().optional().describe(
-    "Runnable agent ID selected for this run.",
+  agent: z.array(z.string()).optional().describe(
+    "Runnable agents selected for this simulation.",
   ),
-  agent_name: z.string().nullable().optional().describe(
-    "Display name captured for the selected agent.",
+  attempt_count: z.int().default(1).describe(
+    "Total concrete attempts after all fan-out dimensions.",
   ),
   completed_at: z.string().nullable().optional(),
   configuration_id: z.string().nullable().optional(),
@@ -49,8 +52,19 @@ export const V2EvalRun$zodSchema: z.ZodType<V2EvalRun> = z.object({
   ),
   id: z.string().describe("Run ID to use when polling eval-run status."),
   job_id: z.string().describe("Alias of `id` for the durable parent job."),
+  persona: z.string().nullable().optional().describe(
+    "Perspective applied to every attempt in this simulation.",
+  ),
   progress: PublicEvalRunProgress$zodSchema.optional(),
+  prompt_count: z.int().default(1).describe(
+    "Number of prompts included in the simulation.",
+  ),
+  runs: z.int().default(1).describe(
+    "Independent attempts requested for every prompt-agent pair.",
+  ),
   stage: z.string().nullable().optional(),
   started_at: z.string().nullable().optional(),
-  status: z.string(),
+  status: z.string().describe(
+    "Terminal values: `completed`, `failed`, and `cancelled`. Recoverable interruptions are reported as `running` with stage `retrying`.",
+  ),
 });

package/src/models/v2evalrunanalysis.ts

@@ -10,7 +10,9 @@ export type V2EvalRunAnalysis = {
   result_id: string;
   attempt_id: string;
   eval_id?: string | null | undefined;
-  agent_id?: string | null | undefined;
+  agent?: string | null | undefined;
+  prompt?: string | null | undefined;
+  prompt_index?: number | undefined;
   variant_key?: string | undefined;
   repeat_index?: number | undefined;
   analysis_version?: string | null | undefined;
@@ -21,8 +23,8 @@ export type V2EvalRunAnalysis = {
 
 export const V2EvalRunAnalysis$zodSchema: z.ZodType<V2EvalRunAnalysis> = z
   .object({
-    agent_id: z.string().nullable().optional().describe(
-      "Runnable agent ID used for this concrete attempt.",
+    agent: z.string().nullable().optional().describe(
+      "Runnable agent used for this concrete attempt.",
     ),
     analysis: z.record(z.string(), z.any()).optional().describe(
       "Persisted eval_run_analysis payload, including verdict, outcome, findings, bottleneck, and analysis sections.",
@@ -38,6 +40,12 @@ export const V2EvalRunAnalysis$zodSchema: z.ZodType<V2EvalRunAnalysis> = z
       "Public eval definition ID, not the backing executable eval ID.",
     ),
     id: z.string().describe("Persisted eval analysis ID."),
+    prompt: z.string().nullable().optional().describe(
+      "Exact task prompt used for this concrete attempt.",
+    ),
+    prompt_index: z.int().default(0).describe(
+      "Zero-based prompt index within the submitted simulation.",
+    ),
     repeat_index: z.int().default(0).describe(
       "Zero-based repeat index within the submitted run.",
     ),

package/src/models/v2evalrundetail.ts

@@ -0,0 +1,90 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: a4eb73f4b2a9
+ */
+
+import * as z from "zod";
+import {
+  PublicEvalRunProgress,
+  PublicEvalRunProgress$zodSchema,
+} from "./publicevalrunprogress.js";
+import {
+  V2EvalResultSummary,
+  V2EvalResultSummary$zodSchema,
+} from "./v2evalresultsummary.js";
+import {
+  V2EvalRunAnalysis,
+  V2EvalRunAnalysis$zodSchema,
+} from "./v2evalrunanalysis.js";
+
+export type V2EvalRunDetail = {
+  id: string;
+  job_id: string;
+  eval_id?: string | null | undefined;
+  eval_name?: string | null | undefined;
+  configuration_id?: string | null | undefined;
+  persona?: string | null | undefined;
+  agent?: Array<string> | undefined;
+  prompt_count?: number | undefined;
+  runs?: number | undefined;
+  attempt_count?: number | undefined;
+  status: string;
+  stage?: string | null | undefined;
+  dispatch_id?: string | null | undefined;
+  progress?: PublicEvalRunProgress | undefined;
+  estimated_cost?: number | null | undefined;
+  error?: string | null | undefined;
+  created_at?: string | null | undefined;
+  started_at?: string | null | undefined;
+  completed_at?: string | null | undefined;
+  results?: Array<V2EvalResultSummary> | undefined;
+  analysis?: Array<V2EvalRunAnalysis> | undefined;
+  results_truncated?: boolean | undefined;
+};
+
+export const V2EvalRunDetail$zodSchema: z.ZodType<V2EvalRunDetail> = z.object({
+  agent: z.array(z.string()).optional().describe(
+    "Runnable agents selected for this simulation.",
+  ),
+  analysis: z.array(V2EvalRunAnalysis$zodSchema).optional().describe(
+    "Per-attempt analysis for this run. Empty until an attempt has been analyzed.",
+  ),
+  attempt_count: z.int().default(1).describe(
+    "Total concrete attempts after all fan-out dimensions.",
+  ),
+  completed_at: z.string().nullable().optional(),
+  configuration_id: z.string().nullable().optional(),
+  created_at: z.string().nullable().optional(),
+  dispatch_id: z.string().nullable().optional(),
+  error: z.string().nullable().optional(),
+  estimated_cost: z.number().nullable().optional(),
+  eval_id: z.string().nullable().optional().describe(
+    "Saved eval definition ID, including one created by a direct run.",
+  ),
+  eval_name: z.string().nullable().optional().describe(
+    "Display name captured for the eval when the run was created.",
+  ),
+  id: z.string().describe("Run ID to use when polling eval-run status."),
+  job_id: z.string().describe("Alias of `id` for the durable parent job."),
+  persona: z.string().nullable().optional().describe(
+    "Perspective applied to every attempt in this simulation.",
+  ),
+  progress: PublicEvalRunProgress$zodSchema.optional(),
+  prompt_count: z.int().default(1).describe(
+    "Number of prompts included in the simulation.",
+  ),
+  results: z.array(V2EvalResultSummary$zodSchema).optional().describe(
+    "Per-attempt outcomes for this run. Empty while the run is active; populated as attempts finish.",
+  ),
+  results_truncated: z.boolean().default(false).describe(
+    "True when the run produced more results than the inline cap. The remaining results are not included in this response.",
+  ),
+  runs: z.int().default(1).describe(
+    "Independent attempts requested for every prompt-agent pair.",
+  ),
+  stage: z.string().nullable().optional(),
+  started_at: z.string().nullable().optional(),
+  status: z.string().describe(
+    "Terminal values: `completed`, `failed`, and `cancelled`. Recoverable interruptions are reported as `running` with stage `retrying`.",
+  ),
+});

package/src/models/v2evalrunscancelop.ts

@@ -0,0 +1,70 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: 15553511dcfe
+ */
+
+import * as z from "zod";
+import {
+  HTTPValidationError,
+  HTTPValidationError$zodSchema,
+} from "./httpvalidationerror.js";
+import { V2EvalRun, V2EvalRun$zodSchema } from "./v2evalrun.js";
+
+export type V2EvalRunsCancelRequest = { run_id: string };
+
+export const V2EvalRunsCancelRequest$zodSchema: z.ZodType<
+  V2EvalRunsCancelRequest
+> = z.object({
+  run_id: z.string(),
+});
+
+/**
+ * Eval run cancellation is temporarily unavailable.
+ */
+export type V2EvalRunsCancelServiceUnavailableResponseBody = { detail: string };
+
+export const V2EvalRunsCancelServiceUnavailableResponseBody$zodSchema:
+  z.ZodType<V2EvalRunsCancelServiceUnavailableResponseBody> = z.object({
+    detail: z.string(),
+  }).describe("Eval run cancellation is temporarily unavailable.");
+
+/**
+ * The eval run has already completed or failed.
+ */
+export type V2EvalRunsCancelConflictResponseBody = { detail: string };
+
+export const V2EvalRunsCancelConflictResponseBody$zodSchema: z.ZodType<
+  V2EvalRunsCancelConflictResponseBody
+> = z.object({
+  detail: z.string(),
+}).describe("The eval run has already completed or failed.");
+
+/**
+ * Eval run not found. Use the run ID returned when you started the eval.
+ */
+export type V2EvalRunsCancelNotFoundResponseBody = { detail: string };
+
+export const V2EvalRunsCancelNotFoundResponseBody$zodSchema: z.ZodType<
+  V2EvalRunsCancelNotFoundResponseBody
+> = z.object({
+  detail: z.string(),
+}).describe(
+  "Eval run not found. Use the run ID returned when you started the eval.",
+);
+
+export type V2EvalRunsCancelResponse =
+  | V2EvalRun
+  | V2EvalRunsCancelNotFoundResponseBody
+  | V2EvalRunsCancelConflictResponseBody
+  | V2EvalRunsCancelServiceUnavailableResponseBody
+  | HTTPValidationError;
+
+export const V2EvalRunsCancelResponse$zodSchema: z.ZodType<
+  V2EvalRunsCancelResponse
+> = z.union([
+  V2EvalRun$zodSchema,
+  z.lazy(() => V2EvalRunsCancelNotFoundResponseBody$zodSchema),
+  z.lazy(() => V2EvalRunsCancelConflictResponseBody$zodSchema),
+  z.lazy(() => V2EvalRunsCancelServiceUnavailableResponseBody$zodSchema),
+  HTTPValidationError$zodSchema,
+]);

package/src/models/v2evalrunscreateop.ts

@@ -13,25 +13,63 @@ import { V2EvalRun, V2EvalRun$zodSchema } from "./v2evalrun.js";
 /**
  * The requested coding agent is unavailable on the organization's current plan.
  */
-export type V2EvalRunsCreateResponseBody = { detail: string };
+export type V2EvalRunsCreateForbiddenResponseBody = { detail: string };
 
-export const V2EvalRunsCreateResponseBody$zodSchema: z.ZodType<
-  V2EvalRunsCreateResponseBody
+export const V2EvalRunsCreateForbiddenResponseBody$zodSchema: z.ZodType<
+  V2EvalRunsCreateForbiddenResponseBody
 > = z.object({
   detail: z.string(),
 }).describe(
   "The requested coding agent is unavailable on the organization's current plan.",
 );
 
+export type Detail = { error: string; required: number; balance: number };
+
+export const Detail$zodSchema: z.ZodType<Detail> = z.object({
+  balance: z.number().describe("The organization's current credit balance."),
+  error: z.string(),
+  required: z.number().describe("Usage-based credits the requested run needs."),
+});
+
+/**
+ * The organization does not currently have enough credits for the requested run.
+ */
+export type V2EvalRunsCreatePaymentRequiredResponseBody = { detail: Detail };
+
+export const V2EvalRunsCreatePaymentRequiredResponseBody$zodSchema: z.ZodType<
+  V2EvalRunsCreatePaymentRequiredResponseBody
+> = z.object({
+  detail: z.lazy(() => Detail$zodSchema),
+}).describe(
+  "The organization does not currently have enough credits for the requested run.",
+);
+
+/**
+ * The saved/direct mode fields, referenced resource, or environment credentials are invalid.
+ */
+export type V2EvalRunsCreateBadRequestResponseBody = { detail: string };
+
+export const V2EvalRunsCreateBadRequestResponseBody$zodSchema: z.ZodType<
+  V2EvalRunsCreateBadRequestResponseBody
+> = z.object({
+  detail: z.string(),
+}).describe(
+  "The saved/direct mode fields, referenced resource, or environment credentials are invalid.",
+);
+
 export type V2EvalRunsCreateResponse =
   | V2EvalRun
-  | V2EvalRunsCreateResponseBody
+  | V2EvalRunsCreateBadRequestResponseBody
+  | V2EvalRunsCreatePaymentRequiredResponseBody
+  | V2EvalRunsCreateForbiddenResponseBody
   | HTTPValidationError;
 
 export const V2EvalRunsCreateResponse$zodSchema: z.ZodType<
   V2EvalRunsCreateResponse
 > = z.union([
   V2EvalRun$zodSchema,
-  z.lazy(() => V2EvalRunsCreateResponseBody$zodSchema),
+  z.lazy(() => V2EvalRunsCreateBadRequestResponseBody$zodSchema),
+  z.lazy(() => V2EvalRunsCreatePaymentRequiredResponseBody$zodSchema),
+  z.lazy(() => V2EvalRunsCreateForbiddenResponseBody$zodSchema),
   HTTPValidationError$zodSchema,
 ]);

package/src/models/v2evalrunsretrieveop.ts

@@ -8,7 +8,10 @@ import {
   HTTPValidationError,
   HTTPValidationError$zodSchema,
 } from "./httpvalidationerror.js";
-import { V2EvalRun, V2EvalRun$zodSchema } from "./v2evalrun.js";
+import {
+  V2EvalRunDetail,
+  V2EvalRunDetail$zodSchema,
+} from "./v2evalrundetail.js";
 
 export type V2EvalRunsRetrieveRequest = { run_id: string };
 
@@ -18,11 +21,28 @@ export const V2EvalRunsRetrieveRequest$zodSchema: z.ZodType<
   run_id: z.string(),
 });
 
-export type V2EvalRunsRetrieveResponse = V2EvalRun | HTTPValidationError;
+/**
+ * Eval run not found. Use the run ID returned when you started the eval.
+ */
+export type V2EvalRunsRetrieveResponseBody = { detail: string };
+
+export const V2EvalRunsRetrieveResponseBody$zodSchema: z.ZodType<
+  V2EvalRunsRetrieveResponseBody
+> = z.object({
+  detail: z.string(),
+}).describe(
+  "Eval run not found. Use the run ID returned when you started the eval.",
+);
+
+export type V2EvalRunsRetrieveResponse =
+  | V2EvalRunDetail
+  | V2EvalRunsRetrieveResponseBody
+  | HTTPValidationError;
 
 export const V2EvalRunsRetrieveResponse$zodSchema: z.ZodType<
   V2EvalRunsRetrieveResponse
 > = z.union([
-  V2EvalRun$zodSchema,
+  V2EvalRunDetail$zodSchema,
+  z.lazy(() => V2EvalRunsRetrieveResponseBody$zodSchema),
   HTTPValidationError$zodSchema,
 ]);