@lightsage/mcp 0.3.0 → 0.4.0

package/bin/mcp-server.js

@@ -1453,7 +1453,7 @@ var init_versions = __esm(() => {
   version2 = {
     major: 4,
     minor: 6,
-    patch: 4
+    patch: 5
   };
 });
 
@@ -28171,6 +28171,9 @@ var require_proxy_addr = __commonJS(function(exports, module) {
       if (!isip(addr))
         return false;
       var ip = parseip(addr);
+      if (ip.kind() === "ipv6" && ip.isIPv4MappedAddress()) {
+        ip = ip.toIPv4Address();
+      }
       var ipconv;
       var kind = ip.kind();
       for (var i = 0;i < subnets.length; i++) {
@@ -28183,10 +28186,15 @@ var require_proxy_addr = __commonJS(function(exports, module) {
           if (subnetkind === "ipv4" && !ip.isIPv4MappedAddress()) {
             continue;
           }
+          if (subnetkind !== "ipv4" && !(subnetrange >= 96 && subnetip.isIPv4MappedAddress())) {
+            continue;
+          }
           if (!ipconv) {
             ipconv = subnetkind === "ipv4" ? ip.toIPv4Address() : ip.toIPv4MappedAddress();
           }
           trusted = ipconv;
+        } else if (kind === "ipv6" && subnetip.isIPv4MappedAddress()) {
+          continue;
         }
         if (trusted.match(subnetip, subnetrange)) {
           return true;
@@ -28204,12 +28212,20 @@ var require_proxy_addr = __commonJS(function(exports, module) {
       if (!isip(addr))
         return false;
       var ip = parseip(addr);
+      if (ip.kind() === "ipv6" && ip.isIPv4MappedAddress()) {
+        ip = ip.toIPv4Address();
+      }
       var kind = ip.kind();
       if (kind !== subnetkind) {
         if (subnetisipv4 && !ip.isIPv4MappedAddress()) {
           return false;
         }
+        if (!subnetisipv4 && !(subnetrange >= 96 && subnetip.isIPv4MappedAddress())) {
+          return false;
+        }
         ip = subnetisipv4 ? ip.toIPv4Address() : ip.toIPv4MappedAddress();
+      } else if (kind === "ipv6" && subnetip.isIPv4MappedAddress()) {
+        return false;
       }
       return ip.match(subnetip, subnetrange);
     };
@@ -42217,6 +42233,22 @@ var require_utils4 = __commonJS(function(exports, module) {
     }
     return output;
   }
+  function recomposeZonedIPv6Host(component) {
+    const zone = component.ipv6Zone;
+    if (zone === undefined)
+      return;
+    const host = component.host;
+    const separator = host.indexOf("%");
+    if (separator === -1)
+      return;
+    const address = host.slice(0, separator);
+    const escaped = address + "%25" + zone;
+    const derived = normalizeIPv6("[" + escaped + "]");
+    if (derived.isIPV6 === true && derived.host === host) {
+      return "[" + escaped + "]";
+    }
+    return;
+  }
   function recomposeAuthority(component) {
     const uriTokens = [];
     if (component.userinfo !== undefined) {
@@ -42226,15 +42258,20 @@ var require_utils4 = __commonJS(function(exports, module) {
     if (component.host !== undefined) {
       let host = component.host;
       if (!isIPv4(host)) {
-        let ipV6res = normalizeIPv6(host);
-        if (ipV6res.isIPV6 !== true && ipV6res.isIPVFuture !== true) {
-          host = normalizePercentEncoding(host, true);
-          ipV6res = normalizeIPv6(host);
-        }
-        if (ipV6res.isIPV6 === true || ipV6res.isIPVFuture === true) {
-          host = `[${ipV6res.escapedHost}]`;
+        const zonedHost = recomposeZonedIPv6Host(component);
+        if (zonedHost !== undefined) {
+          host = zonedHost;
         } else {
-          host = reescapeHostDelimiters(host, false);
+          let ipV6res = normalizeIPv6(host);
+          if (ipV6res.isIPV6 !== true && ipV6res.isIPVFuture !== true) {
+            host = normalizePercentEncoding(host, true);
+            ipV6res = normalizeIPv6(host);
+          }
+          if (ipV6res.isIPV6 === true || ipV6res.isIPVFuture === true) {
+            host = `[${ipV6res.escapedHost}]`;
+          } else {
+            host = reescapeHostDelimiters(host, false);
+          }
         }
       }
       uriTokens.push(host);
@@ -42447,6 +42484,7 @@ var require_schemes = __commonJS(function(exports, module) {
   var HEX_PAIR = /^[\da-f]{2}$/iu;
   var MAILTO_DOMAIN_LITERAL = /^\[[\x21-\x5A\x5E-\x7E]*\]$/u;
   var MAILTO_DOMAIN_ERROR = "URI mailto has an invalid recipient domain.";
+  var MAILTO_AUTHORITY_ERROR = "URI mailto must not have an authority component.";
   var HAS_SURROGATE = /[\uD800-\uDFFF]/u;
   function decodeHex(str) {
     if (typeof str !== "string" || str.indexOf("%") === -1) {
@@ -42561,30 +42599,37 @@ var require_schemes = __commonJS(function(exports, module) {
           mailtoComponent.error = mailtoComponent.error || "URI mailto has malformed header fields.";
           continue;
         }
-        const name = eqIdx === -1 ? token : token.slice(0, eqIdx);
+        const name = decodeHex(eqIdx === -1 ? token : token.slice(0, eqIdx));
+        const normalizedName = name.toLowerCase();
         const value = eqIdx === -1 ? "" : token.slice(eqIdx + 1);
-        if (name === "to") {
+        if (normalizedName === "to") {
           const addrs = value.split(",");
           for (let j = 0;j < addrs.length; j++)
             to.push(addrs[j]);
           continue;
         }
-        if (name === "subject") {
+        if (normalizedName === "subject") {
           mailtoComponent.subject = decodeHex(value);
           continue;
         }
-        if (name === "body") {
+        if (normalizedName === "body") {
           mailtoComponent.body = decodeHex(value);
           continue;
         }
         if (headers === null)
           headers = Object.create(null);
-        headers[decodeHex(name)] = decodeHex(value);
+        headers[name] = decodeHex(value);
       }
       if (headers !== null)
         mailtoComponent.headers = headers;
     }
     mailtoComponent.query = undefined;
+    if (to.length > 0 && (mailtoComponent.userinfo !== undefined || mailtoComponent.host !== undefined || mailtoComponent.port !== undefined)) {
+      mailtoComponent.userinfo = undefined;
+      mailtoComponent.host = undefined;
+      mailtoComponent.port = undefined;
+      mailtoComponent.error = mailtoComponent.error || MAILTO_AUTHORITY_ERROR;
+    }
     for (let i = 0;i < to.length; i++) {
       const rawAddr = to[i];
       const atIdx = rawAddr.lastIndexOf("@");
@@ -42601,10 +42646,35 @@ var require_schemes = __commonJS(function(exports, module) {
       mailtoComponent.to = to;
     return mailtoComponent;
   }
+  function mailtoEncodeHeaderName(name) {
+    return encodeWithAllow(name.replace(/%/gu, "%25"), HFNAME);
+  }
   function mailtoSerialize(component, options) {
     const mailtoComponent = component;
     const to = Array.isArray(mailtoComponent.to) ? mailtoComponent.to.slice() : [];
+    const sourceHeaders = mailtoComponent.headers && typeof mailtoComponent.headers === "object" ? Object.assign(Object.create(null), mailtoComponent.headers) : Object.create(null);
+    const headers = Object.create(null);
+    for (const name in sourceHeaders) {
+      const normalizedName = name.toLowerCase();
+      if (normalizedName === "to") {
+        const addrs = String(sourceHeaders[name]).split(",");
+        for (let i = 0;i < addrs.length; i++)
+          to.push(addrs[i]);
+      } else if (normalizedName === "subject" || normalizedName === "body") {
+        headers[normalizedName] = sourceHeaders[name];
+      } else {
+        headers[name] = sourceHeaders[name];
+      }
+    }
+    if (mailtoComponent.subject !== undefined)
+      headers.subject = mailtoComponent.subject;
+    if (mailtoComponent.body !== undefined)
+      headers.body = mailtoComponent.body;
+    mailtoComponent.headers = headers;
     if (to.length) {
+      mailtoComponent.userinfo = undefined;
+      mailtoComponent.host = undefined;
+      mailtoComponent.port = undefined;
       for (let i = 0;i < to.length; i++) {
         const addr = String(to[i]);
         const atIdx = addr.lastIndexOf("@");
@@ -42620,18 +42690,12 @@ var require_schemes = __commonJS(function(exports, module) {
     } else {
       mailtoComponent.path = undefined;
     }
-    const headers = mailtoComponent.headers && typeof mailtoComponent.headers === "object" ? Object.assign(Object.create(null), mailtoComponent.headers) : Object.create(null);
-    if (mailtoComponent.subject)
-      headers.subject = mailtoComponent.subject;
-    if (mailtoComponent.body)
-      headers.body = mailtoComponent.body;
-    mailtoComponent.headers = headers;
     let query = "";
     let count = 0;
     for (const name in headers) {
       if (count++ !== 0)
         query += "&";
-      query += encodeWithAllow(name, HFNAME) + "=" + encodeWithAllow(String(headers[name]), HFNAME);
+      query += mailtoEncodeHeaderName(name) + "=" + encodeWithAllow(String(headers[name]), HFNAME);
     }
     if (count !== 0) {
       mailtoComponent.query = query;
@@ -42724,6 +42788,12 @@ var require_fast_uri = __commonJS(function(exports, module) {
     schemelessOptions.skipEscape = true;
     return serialize(resolved, schemelessOptions);
   }
+  function copyHost(target, source) {
+    target.host = source.host;
+    if (source.ipv6Zone !== undefined) {
+      target.ipv6Zone = source.ipv6Zone;
+    }
+  }
   function resolveComponent(base, relative, options, skipNormalization) {
     const target = {};
     if (!skipNormalization) {
@@ -42734,14 +42804,14 @@ var require_fast_uri = __commonJS(function(exports, module) {
     if (!options.tolerant && relative.scheme) {
       target.scheme = relative.scheme;
       target.userinfo = relative.userinfo;
-      target.host = relative.host;
+      copyHost(target, relative);
       target.port = relative.port;
       target.path = removeDotSegments(relative.path || "");
       target.query = relative.query;
     } else {
       if (relative.userinfo !== undefined || relative.host !== undefined || relative.port !== undefined) {
         target.userinfo = relative.userinfo;
-        target.host = relative.host;
+        copyHost(target, relative);
         target.port = relative.port;
         target.path = removeDotSegments(relative.path || "");
         target.query = relative.query;
@@ -42769,7 +42839,7 @@ var require_fast_uri = __commonJS(function(exports, module) {
           target.query = relative.query;
         }
         target.userinfo = base.userinfo;
-        target.host = base.host;
+        copyHost(target, base);
         target.port = base.port;
       }
       target.scheme = base.scheme;
@@ -42785,6 +42855,7 @@ var require_fast_uri = __commonJS(function(exports, module) {
   function serialize(cmpts, opts) {
     const component = {
       host: cmpts.host,
+      ipv6Zone: cmpts.ipv6Zone,
       scheme: cmpts.scheme,
       userinfo: cmpts.userinfo,
       port: cmpts.port,
@@ -42981,6 +43052,9 @@ var require_fast_uri = __commonJS(function(exports, module) {
           isIP = ipv6result.isIPV6 || ipv6result.isIPVFuture === true;
           malformedIPLiteral = hasIPLiteralBracket && (!bracketedIPLiteral || ipv6result.error === true);
           parsed.host = isIP ? ipv6result.host : ipv6result.host.toLowerCase();
+          if (isIP && ipv6result.isIPV6 === true && parsed.host.indexOf("%") !== -1) {
+            parsed.ipv6Zone = parsed.host.slice(parsed.host.indexOf("%") + 1);
+          }
           if (malformedIPLiteral) {
             parsed.error = parsed.error || "URI host is malformed.";
             malformedAuthorityOrPort = true;
@@ -43005,13 +43079,14 @@ var require_fast_uri = __commonJS(function(exports, module) {
       if (!malformedIPLiteral) {
         malformedHost = canonicalizeHost(parsed, options, schemeHandler, isIP);
       }
-      if (!schemeHandler || schemeHandler && !schemeHandler.skipNormalize) {
-        if (uri.indexOf("%") !== -1) {
-          if (parsed.host !== undefined && !malformedIPLiteral) {
-            const host = isIP ? parsed.host : normalizePercentEncoding(parsed.host, true);
-            parsed.host = reescapeHostDelimiters(host, isIP);
-          }
+      if (uri.indexOf("%") !== -1 && parsed.host !== undefined && !malformedIPLiteral) {
+        let host = isIP ? parsed.host : normalizePercentEncoding(parsed.host, true);
+        if (!isIP) {
+          host = normalizePercentEncoding(host.toLowerCase());
         }
+        parsed.host = reescapeHostDelimiters(host, isIP);
+      }
+      if (!schemeHandler || schemeHandler && !schemeHandler.skipNormalize) {
         if (parsed.path) {
           parsed.path = normalizePathEncoding(parsed.path);
         }
@@ -52769,9 +52844,9 @@ var init_config = __esm(() => {
   SDK_METADATA = {
     language: "typescript",
     openapiDocVersion: "1.0.0",
-    sdkVersion: "0.3.0",
-    genVersion: "2.935.1",
-    userAgent: "speakeasy-sdk/mcp-typescript 0.3.0 2.935.1 1.0.0 @lightsage/mcp"
+    sdkVersion: "0.4.0",
+    genVersion: "2.941.0",
+    userAgent: "speakeasy-sdk/mcp-typescript 0.4.0 2.941.0 1.0.0 @lightsage/mcp"
   };
 });
 
@@ -54953,312 +55028,6 @@ Returns reusable environments without exposing secret values.`,
   };
 });
 
-// src/models/v2evalresultsummary.ts
-var V2EvalResultSummary$zodSchema;
-var init_v2evalresultsummary = __esm(() => {
-  init_zod();
-  V2EvalResultSummary$zodSchema = object({
-    agent_id: string2().nullable().optional().describe("Runnable agent ID combining the harness and model."),
-    created_at: string2().nullable().optional(),
-    eval_id: string2().nullable().optional(),
-    failure_reasons: array(string2()).nullable().optional(),
-    id: string2().describe("Result ID, not an eval job ID."),
-    latency_ms: int().nullable().optional(),
-    passed: boolean2().nullable().optional(),
-    score: number2().nullable().optional().describe("Score from 0 to 1.")
-  });
-});
-
-// src/models/v2evalresultsresponse.ts
-var V2EvalResultsResponse$zodSchema;
-var init_v2evalresultsresponse = __esm(() => {
-  init_zod();
-  init_v2evalresultsummary();
-  V2EvalResultsResponse$zodSchema = object({
-    count: int(),
-    data: array(V2EvalResultSummary$zodSchema).optional(),
-    limit: int()
-  });
-});
-
-// src/models/v2evalresultslistop.ts
-var V2EvalResultsListRequest$zodSchema, V2EvalResultsListResponse$zodSchema;
-var init_v2evalresultslistop = __esm(() => {
-  init_zod();
-  init_httpvalidationerror();
-  init_v2evalresultsresponse();
-  V2EvalResultsListRequest$zodSchema = object({
-    limit: int().default(20),
-    since: string2().describe("Relative window such as `7d` or an ISO timestamp.").optional()
-  });
-  V2EvalResultsListResponse$zodSchema = union([
-    V2EvalResultsResponse$zodSchema,
-    HTTPValidationError$zodSchema
-  ]);
-});
-
-// src/funcs/evalResultsV2EvalResultsList.ts
-function evalResultsV2EvalResultsList(client$, request, options) {
-  return new APIPromise($do7(client$, request, options));
-}
-async function $do7(client$, request, options) {
-  const parsed$ = safeParse4(request, (value$) => V2EvalResultsListRequest$zodSchema.optional().parse(value$), "Input validation failed");
-  if (!parsed$.ok) {
-    return [parsed$, { status: "invalid" }];
-  }
-  const payload$ = parsed$.value;
-  const body$ = null;
-  const path$ = pathToFunc("/v2/eval-results")();
-  const query$ = encodeFormQuery({
-    limit: payload$?.limit,
-    since: payload$?.since
-  });
-  const headers$ = new Headers(compactMap({
-    Accept: "application/json"
-  }));
-  const securityInput = await extractSecurity(client$._options.security);
-  const requestSecurity = resolveGlobalSecurity(securityInput);
-  const context = {
-    options: client$._options,
-    baseURL: options?.serverURL ?? client$._baseURL ?? "",
-    operationID: "v2_eval_results_list",
-    oAuth2Scopes: null,
-    resolvedSecurity: requestSecurity,
-    securitySource: client$._options.security,
-    retryConfig: options?.retries || client$._options.retryConfig || { strategy: "none" },
-    retryCodes: options?.retryCodes || [
-      "429",
-      "500",
-      "502",
-      "503",
-      "504"
-    ]
-  };
-  const requestRes = client$._createRequest(context, {
-    security: requestSecurity,
-    method: "GET",
-    baseURL: options?.serverURL,
-    path: path$,
-    headers: headers$,
-    query: query$,
-    body: body$,
-    userAgent: client$._options.userAgent,
-    timeoutMs: options?.timeoutMs || client$._options.timeoutMs || -1
-  }, options);
-  if (!requestRes.ok) {
-    return [requestRes, { status: "invalid" }];
-  }
-  const req$ = requestRes.value;
-  const doResult = await client$._do(req$, {
-    context,
-    errorCodes: [],
-    retryConfig: context.retryConfig,
-    retryCodes: context.retryCodes
-  });
-  if (!doResult.ok) {
-    return [doResult, { status: "request-error", request: req$ }];
-  }
-  return [doResult, {
-    status: "complete",
-    request: req$,
-    response: doResult.value
-  }];
-}
-var init_evalResultsV2EvalResultsList = __esm(() => {
-  init_encodings();
-  init_primitives();
-  init_schemas4();
-  init_security();
-  init_url();
-  init_v2evalresultslistop();
-  init_async();
-});
-
-// src/mcp-server/tools/evalResultsV2EvalResultsList.ts
-var args4, tool$evalResultsV2EvalResultsList;
-var init_evalResultsV2EvalResultsList2 = __esm(() => {
-  init_evalResultsV2EvalResultsList();
-  init_v2evalresultslistop();
-  init_tools();
-  args4 = {
-    request: V2EvalResultsListRequest$zodSchema.optional()
-  };
-  tool$evalResultsV2EvalResultsList = {
-    name: "eval-results-v2-eval-results-list",
-    description: `List Results
-
-Returns completed custom-eval results from all supported run flows, ordered newest first. Result IDs are different from eval run IDs.`,
-    annotations: {
-      title: "",
-      destructiveHint: false,
-      idempotentHint: false,
-      openWorldHint: false,
-      readOnlyHint: true
-    },
-    args: args4,
-    tool: async (client, args, ctx) => {
-      const [result] = await evalResultsV2EvalResultsList(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
-      if (!result.ok) {
-        return {
-          content: [{ type: "text", text: result.error.message }],
-          isError: true
-        };
-      }
-      return formatResult(result.value);
-    }
-  };
-});
-
-// src/models/v2evalresult.ts
-var V2EvalResult$zodSchema;
-var init_v2evalresult = __esm(() => {
-  init_zod();
-  V2EvalResult$zodSchema = object({
-    agent_id: string2().nullable().optional().describe("Runnable agent ID combining the harness and model."),
-    created_at: string2().nullable().optional(),
-    eval_id: string2().nullable().optional(),
-    execution_stderr: string2().nullable().optional(),
-    execution_stdout: string2().nullable().optional(),
-    exit_code: int().nullable().optional(),
-    failure_reasons: array(string2()).nullable().optional(),
-    id: string2().describe("Result ID, not an eval job ID."),
-    latency_ms: int().nullable().optional(),
-    output: string2().nullable().optional(),
-    passed: boolean2().nullable().optional(),
-    prompt: string2().nullable().optional(),
-    score: number2().nullable().optional().describe("Score from 0 to 1."),
-    tool_calls: array(record(string2(), any())).nullable().optional()
-  });
-});
-
-// src/models/v2evalresultsretrieveop.ts
-var V2EvalResultsRetrieveRequest$zodSchema, V2EvalResultsRetrieveResponse$zodSchema;
-var init_v2evalresultsretrieveop = __esm(() => {
-  init_zod();
-  init_httpvalidationerror();
-  init_v2evalresult();
-  V2EvalResultsRetrieveRequest$zodSchema = object({
-    result_id: string2()
-  });
-  V2EvalResultsRetrieveResponse$zodSchema = union([
-    V2EvalResult$zodSchema,
-    HTTPValidationError$zodSchema
-  ]);
-});
-
-// src/funcs/evalResultsV2EvalResultsRetrieve.ts
-function evalResultsV2EvalResultsRetrieve(client$, request, options) {
-  return new APIPromise($do8(client$, request, options));
-}
-async function $do8(client$, request, options) {
-  const parsed$ = safeParse4(request, (value$) => V2EvalResultsRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
-  if (!parsed$.ok) {
-    return [parsed$, { status: "invalid" }];
-  }
-  const payload$ = parsed$.value;
-  const body$ = null;
-  const pathParams$ = {
-    result_id: encodeSimple("result_id", payload$.result_id, {
-      explode: false,
-      charEncoding: "percent"
-    })
-  };
-  const path$ = pathToFunc("/v2/eval-results/{result_id}")(pathParams$);
-  const headers$ = new Headers(compactMap({
-    Accept: "application/json"
-  }));
-  const securityInput = await extractSecurity(client$._options.security);
-  const requestSecurity = resolveGlobalSecurity(securityInput);
-  const context = {
-    options: client$._options,
-    baseURL: options?.serverURL ?? client$._baseURL ?? "",
-    operationID: "v2_eval_results_retrieve",
-    oAuth2Scopes: null,
-    resolvedSecurity: requestSecurity,
-    securitySource: client$._options.security,
-    retryConfig: options?.retries || client$._options.retryConfig || { strategy: "none" },
-    retryCodes: options?.retryCodes || [
-      "429",
-      "500",
-      "502",
-      "503",
-      "504"
-    ]
-  };
-  const requestRes = client$._createRequest(context, {
-    security: requestSecurity,
-    method: "GET",
-    baseURL: options?.serverURL,
-    path: path$,
-    headers: headers$,
-    body: body$,
-    userAgent: client$._options.userAgent,
-    timeoutMs: options?.timeoutMs || client$._options.timeoutMs || -1
-  }, options);
-  if (!requestRes.ok) {
-    return [requestRes, { status: "invalid" }];
-  }
-  const req$ = requestRes.value;
-  const doResult = await client$._do(req$, {
-    context,
-    errorCodes: [],
-    retryConfig: context.retryConfig,
-    retryCodes: context.retryCodes
-  });
-  if (!doResult.ok) {
-    return [doResult, { status: "request-error", request: req$ }];
-  }
-  return [doResult, {
-    status: "complete",
-    request: req$,
-    response: doResult.value
-  }];
-}
-var init_evalResultsV2EvalResultsRetrieve = __esm(() => {
-  init_encodings();
-  init_primitives();
-  init_schemas4();
-  init_security();
-  init_url();
-  init_v2evalresultsretrieveop();
-  init_async();
-});
-
-// src/mcp-server/tools/evalResultsV2EvalResultsRetrieve.ts
-var args5, tool$evalResultsV2EvalResultsRetrieve;
-var init_evalResultsV2EvalResultsRetrieve2 = __esm(() => {
-  init_evalResultsV2EvalResultsRetrieve();
-  init_v2evalresultsretrieveop();
-  init_tools();
-  args5 = {
-    request: V2EvalResultsRetrieveRequest$zodSchema
-  };
-  tool$evalResultsV2EvalResultsRetrieve = {
-    name: "eval-results-v2-eval-results-retrieve",
-    description: `Retrieve Result
-
-Returns one completed custom-eval result with its output and diagnostic details.`,
-    annotations: {
-      title: "",
-      destructiveHint: false,
-      idempotentHint: false,
-      openWorldHint: false,
-      readOnlyHint: true
-    },
-    args: args5,
-    tool: async (client, args, ctx) => {
-      const [result] = await evalResultsV2EvalResultsRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
-      if (!result.ok) {
-        return {
-          content: [{ type: "text", text: result.error.message }],
-          isError: true
-        };
-      }
-      return formatResult(result.value);
-    }
-  };
-});
-
 // src/models/publicevalrunprogress.ts
 var PublicEvalRunProgress$zodSchema;
 var init_publicevalrunprogress = __esm(() => {
@@ -55267,9 +55036,10 @@ var init_publicevalrunprogress = __esm(() => {
     cancelled: int().default(0),
     completed: int().default(0),
     failed: int().default(0),
-    interrupted: int().default(0),
-    percent: number2().default(0).describe("The percentage of child jobs in a terminal state. It can reach `100` while the parent job is still summarizing, so use `status` to detect completion."),
+    interrupted: int().default(0).describe("Deprecated compatibility alias for `retrying`; use `retrying`."),
+    percent: number2().default(0).describe("The percentage of child jobs in a final state (`completed`, `failed`, or `cancelled`). It can reach `100` while the parent job is still summarizing, so use `status` to detect completion."),
     queued: int().default(0),
+    retrying: int().default(0).describe("Recoverable child jobs waiting to be retried. These remain active and are excluded from `percent`."),
     running: int().default(0),
     total: int().default(0)
   });
@@ -55281,8 +55051,8 @@ var init_v2evalrun = __esm(() => {
   init_zod();
   init_publicevalrunprogress();
   V2EvalRun$zodSchema = object({
-    agent_id: string2().nullable().optional().describe("Runnable agent ID selected for this run."),
-    agent_name: string2().nullable().optional().describe("Display name captured for the selected agent."),
+    agent: array(string2()).optional().describe("Runnable agents selected for this simulation."),
+    attempt_count: int().default(1).describe("Total concrete attempts after all fan-out dimensions."),
     completed_at: string2().nullable().optional(),
     configuration_id: string2().nullable().optional(),
     created_at: string2().nullable().optional(),
@@ -55293,10 +55063,13 @@ var init_v2evalrun = __esm(() => {
     eval_name: string2().nullable().optional().describe("Display name captured for the eval when the run was created."),
     id: string2().describe("Run ID to use when polling eval-run status."),
     job_id: string2().describe("Alias of `id` for the durable parent job."),
+    persona: string2().nullable().optional().describe("Perspective applied to every attempt in this simulation."),
     progress: PublicEvalRunProgress$zodSchema.optional(),
+    prompt_count: int().default(1).describe("Number of prompts included in the simulation."),
+    runs: int().default(1).describe("Independent attempts requested for every prompt-agent pair."),
     stage: string2().nullable().optional(),
     started_at: string2().nullable().optional(),
-    status: string2()
+    status: string2().describe("Terminal values: `completed`, `failed`, and `cancelled`. Recoverable interruptions are reported as `running` with stage `retrying`.")
   });
 });
 
@@ -55331,9 +55104,9 @@ var init_v2evalrunslistop = __esm(() => {
 
 // src/funcs/evalRunsV2EvalRunsList.ts
 function evalRunsV2EvalRunsList(client$, request, options) {
-  return new APIPromise($do9(client$, request, options));
+  return new APIPromise($do7(client$, request, options));
 }
-async function $do9(client$, request, options) {
+async function $do7(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalRunsListRequest$zodSchema.optional().parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -55407,12 +55180,12 @@ var init_evalRunsV2EvalRunsList = __esm(() => {
 });
 
 // src/mcp-server/tools/evalRunsV2EvalRunsList.ts
-var args6, tool$evalRunsV2EvalRunsList;
+var args4, tool$evalRunsV2EvalRunsList;
 var init_evalRunsV2EvalRunsList2 = __esm(() => {
   init_evalRunsV2EvalRunsList();
   init_v2evalrunslistop();
   init_tools();
-  args6 = {
+  args4 = {
     request: V2EvalRunsListRequest$zodSchema.optional()
   };
   tool$evalRunsV2EvalRunsList = {
@@ -55427,7 +55200,7 @@ Returns jobs created through \`POST /v2/eval-runs\`, ordered newest first, with
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args6,
+    args: args4,
     tool: async (client, args, ctx) => {
       const [result] = await evalRunsV2EvalRunsList(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -55441,27 +55214,39 @@ Returns jobs created through \`POST /v2/eval-runs\`, ordered newest first, with
   };
 });
 
-// src/models/v2evalrunsretrieveop.ts
-var V2EvalRunsRetrieveRequest$zodSchema, V2EvalRunsRetrieveResponse$zodSchema;
-var init_v2evalrunsretrieveop = __esm(() => {
+// src/models/v2evalrunscancelop.ts
+var V2EvalRunsCancelRequest$zodSchema, V2EvalRunsCancelServiceUnavailableResponseBody$zodSchema, V2EvalRunsCancelConflictResponseBody$zodSchema, V2EvalRunsCancelNotFoundResponseBody$zodSchema, V2EvalRunsCancelResponse$zodSchema;
+var init_v2evalrunscancelop = __esm(() => {
   init_zod();
   init_httpvalidationerror();
   init_v2evalrun();
-  V2EvalRunsRetrieveRequest$zodSchema = object({
+  V2EvalRunsCancelRequest$zodSchema = object({
     run_id: string2()
   });
-  V2EvalRunsRetrieveResponse$zodSchema = union([
+  V2EvalRunsCancelServiceUnavailableResponseBody$zodSchema = object({
+    detail: string2()
+  }).describe("Eval run cancellation is temporarily unavailable.");
+  V2EvalRunsCancelConflictResponseBody$zodSchema = object({
+    detail: string2()
+  }).describe("The eval run has already completed or failed.");
+  V2EvalRunsCancelNotFoundResponseBody$zodSchema = object({
+    detail: string2()
+  }).describe("Eval run not found. Use the run ID returned when you started the eval.");
+  V2EvalRunsCancelResponse$zodSchema = union([
     V2EvalRun$zodSchema,
+    lazy(() => V2EvalRunsCancelNotFoundResponseBody$zodSchema),
+    lazy(() => V2EvalRunsCancelConflictResponseBody$zodSchema),
+    lazy(() => V2EvalRunsCancelServiceUnavailableResponseBody$zodSchema),
     HTTPValidationError$zodSchema
   ]);
 });
 
-// src/funcs/evalRunsV2EvalRunsRetrieve.ts
-function evalRunsV2EvalRunsRetrieve(client$, request, options) {
-  return new APIPromise($do10(client$, request, options));
+// src/funcs/evalsCancel.ts
+function evalsCancel(client$, request, options) {
+  return new APIPromise($do8(client$, request, options));
 }
-async function $do10(client$, request, options) {
-  const parsed$ = safeParse4(request, (value$) => V2EvalRunsRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
+async function $do8(client$, request, options) {
+  const parsed$ = safeParse4(request, (value$) => V2EvalRunsCancelRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
   }
@@ -55473,7 +55258,7 @@ async function $do10(client$, request, options) {
       charEncoding: "percent"
     })
   };
-  const path$ = pathToFunc("/v2/eval-runs/{run_id}")(pathParams$);
+  const path$ = pathToFunc("/v2/eval-runs/{run_id}/cancel")(pathParams$);
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
   }));
@@ -55482,7 +55267,7 @@ async function $do10(client$, request, options) {
   const context = {
     options: client$._options,
     baseURL: options?.serverURL ?? client$._baseURL ?? "",
-    operationID: "v2_eval_runs_retrieve",
+    operationID: "v2_eval_runs_cancel",
     oAuth2Scopes: null,
     resolvedSecurity: requestSecurity,
     securitySource: client$._options.security,
@@ -55497,7 +55282,7 @@ async function $do10(client$, request, options) {
   };
   const requestRes = client$._createRequest(context, {
     security: requestSecurity,
-    method: "GET",
+    method: "POST",
     baseURL: options?.serverURL,
     path: path$,
     headers: headers$,
@@ -55524,40 +55309,40 @@ async function $do10(client$, request, options) {
     response: doResult.value
   }];
 }
-var init_evalRunsV2EvalRunsRetrieve = __esm(() => {
+var init_evalsCancel = __esm(() => {
   init_encodings();
   init_primitives();
   init_schemas4();
   init_security();
   init_url();
-  init_v2evalrunsretrieveop();
+  init_v2evalrunscancelop();
   init_async();
 });
 
-// src/mcp-server/tools/evalRunsV2EvalRunsRetrieve.ts
-var args7, tool$evalRunsV2EvalRunsRetrieve;
-var init_evalRunsV2EvalRunsRetrieve2 = __esm(() => {
-  init_evalRunsV2EvalRunsRetrieve();
-  init_v2evalrunsretrieveop();
+// src/mcp-server/tools/evalsCancel.ts
+var args5, tool$evalsCancel;
+var init_evalsCancel2 = __esm(() => {
+  init_evalsCancel();
+  init_v2evalrunscancelop();
   init_tools();
-  args7 = {
-    request: V2EvalRunsRetrieveRequest$zodSchema
+  args5 = {
+    request: V2EvalRunsCancelRequest$zodSchema
   };
-  tool$evalRunsV2EvalRunsRetrieve = {
-    name: "eval-runs-v2-eval-runs-retrieve",
-    description: `Retrieve Run
+  tool$evalsCancel = {
+    name: "evals-cancel",
+    description: `Cancel an eval run
 
-Returns the status and progress of an eval job. Poll until the status reaches a terminal value.`,
+Marks an active eval run as cancelled and prevents queued attempts from starting. Running sandbox attempts are stopped asynchronously. Requests already sent to an external provider cannot be recalled, but any later response does not change the run from cancelled. Repeating this request for an already-cancelled run is safe.`,
     annotations: {
       title: "",
       destructiveHint: false,
       idempotentHint: false,
       openWorldHint: false,
-      readOnlyHint: true
+      readOnlyHint: false
     },
-    args: args7,
+    args: args5,
     tool: async (client, args, ctx) => {
-      const [result] = await evalRunsV2EvalRunsRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
+      const [result] = await evalsCancel(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
         return {
           content: [{ type: "text", text: result.error.message }],
@@ -55569,18 +55354,107 @@ Returns the status and progress of an eval job. Poll until the status reaches a
   };
 });
 
+// src/models/publicevalrunenvironment.ts
+var PublicEvalRunEnvironment$zodSchema;
+var init_publicevalrunenvironment = __esm(() => {
+  init_zod();
+  PublicEvalRunEnvironment$zodSchema = object({
+    changed_since_queued: boolean2().nullable().optional().describe("True when the saved environment has been edited or deleted since the run was queued, so a rerun would use different values. Null when unknown or for inline environments."),
+    id: string2().nullable().optional().describe("Environment ID, as listed by the environments operation."),
+    inline: boolean2().default(false).describe("True when variables were supplied inline with the run request instead of from a saved environment."),
+    name: string2().nullable().optional().describe("Environment name when the run was queued."),
+    queued_at: string2().nullable().optional().describe("When the run was queued with this environment."),
+    updated_at: string2().nullable().optional().describe("Environment version the run used: when its name or variables last changed before the run was queued. Null if the environment had not changed since versioning was introduced."),
+    variable_names: array(string2()).optional().describe("Variable names supplied to the run. Values are never returned.")
+  }).describe("Non-secret summary of the environment an eval run was queued with.");
+});
+
+// src/models/publiccriterionverdict.ts
+var Verdict$zodSchema, PublicCriterionVerdict$zodSchema;
+var init_publiccriterionverdict = __esm(() => {
+  init_zod();
+  Verdict$zodSchema = _enum([
+    "PASS",
+    "FAIL",
+    "NOT_VERIFIED"
+  ]).describe("`PASS` when the evidence shows the requirement was met, `FAIL` when it shows it was not, and `NOT_VERIFIED` when the evidence cannot confirm it. Any `FAIL` or `NOT_VERIFIED` fails the run, whatever the score.");
+  PublicCriterionVerdict$zodSchema = object({
+    criterion: string2().describe("One mandatory requirement from the judge criteria."),
+    evidence: string2().nullable().optional().describe("Short evidence the judge gave for the verdict."),
+    verdict: Verdict$zodSchema.describe("`PASS` when the evidence shows the requirement was met, `FAIL` when it shows it was not, and `NOT_VERIFIED` when the evidence cannot confirm it. Any `FAIL` or `NOT_VERIFIED` fails the run, whatever the score.")
+  });
+});
+
+// src/models/publicgraderresult.ts
+var PublicGraderResult$zodSchema;
+var init_publicgraderresult = __esm(() => {
+  init_zod();
+  init_publiccriterionverdict();
+  PublicGraderResult$zodSchema = object({
+    criteria_verdicts: array(PublicCriterionVerdict$zodSchema).optional().describe("Judges only: a verdict for each mandatory requirement in the judge criteria. Empty for other checks and for runs graded before per-requirement verdicts were recorded."),
+    deductions: array(string2()).optional().describe("Specific shortcomings the check recorded, even when it passed. A judge can pass with a score below 1 and still list deductions here; those checks do not appear in `failure_reasons`."),
+    details: string2().nullable().optional().describe("Reasoning or check output explaining the verdict."),
+    label: string2().nullable().optional().describe("Display name for the check, when the grader sets one."),
+    name: string2().describe("Grader name as configured on the eval."),
+    passed: boolean2().nullable().optional().describe("Whether this check passed."),
+    score: number2().nullable().optional().describe("Score from 0 to 1 for checks that score partially, such as judges; null for pass/fail checks.")
+  });
+});
+
+// src/models/searchgroundingmode.ts
+var SearchGroundingMode$zodSchema;
+var init_searchgroundingmode = __esm(() => {
+  init_zod();
+  SearchGroundingMode$zodSchema = _enum([
+    "disabled",
+    "enabled",
+    "default"
+  ]).describe("Requested web-retrieval behavior for an eval run.");
+});
+
+// src/models/v2evalresultsummary.ts
+var V2EvalResultSummary$zodSchema;
+var init_v2evalresultsummary = __esm(() => {
+  init_zod();
+  init_publicevalrunenvironment();
+  init_publicgraderresult();
+  init_searchgroundingmode();
+  V2EvalResultSummary$zodSchema = object({
+    agent: string2().nullable().optional().describe("Runnable agent combining the harness and model."),
+    created_at: string2().nullable().optional(),
+    environment: PublicEvalRunEnvironment$zodSchema.nullable().optional().describe("The environment this attempt used, without secret values. Null when the attempt had no environment."),
+    eval_id: string2().nullable().optional(),
+    failure_reasons: array(string2()).nullable().optional().describe("Checks that failed, listed even when the result passed overall. Deductions from checks that passed are in `grader_results[].deductions`, not here."),
+    grader_results: array(PublicGraderResult$zodSchema).optional().describe("Per-check verdicts behind `score`, including deductions recorded by passing checks."),
+    grounding_qualification_reason: string2().nullable().optional(),
+    grounding_qualified: boolean2().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: string2().describe("Result ID, not an eval job ID."),
+    latency_ms: int().nullable().optional(),
+    passed: boolean2().nullable().optional(),
+    prompt: string2().nullable().optional(),
+    prompt_index: int().default(0),
+    repeat_index: int().default(0),
+    score: number2().nullable().optional().describe("Score from 0 to 1."),
+    search_grounding: SearchGroundingMode$zodSchema.nullable().optional(),
+    search_observed: boolean2().nullable().optional(),
+    web_retrieval_observed: boolean2().nullable().optional()
+  });
+});
+
 // src/models/v2evalrunanalysis.ts
 var V2EvalRunAnalysis$zodSchema;
 var init_v2evalrunanalysis = __esm(() => {
   init_zod();
   V2EvalRunAnalysis$zodSchema = object({
-    agent_id: string2().nullable().optional().describe("Runnable agent ID used for this concrete attempt."),
+    agent: string2().nullable().optional().describe("Runnable agent used for this concrete attempt."),
     analysis: record(string2(), any()).optional().describe("Persisted eval_run_analysis payload, including verdict, outcome, findings, bottleneck, and analysis sections."),
     analysis_version: string2().nullable().optional().describe("Version of the analysis pipeline that produced this payload."),
     attempt_id: string2().describe("Child eval-attempt ID used to correlate the analysis with trace events."),
     created_at: string2().nullable().optional(),
     eval_id: string2().nullable().optional().describe("Public eval definition ID, not the backing executable eval ID."),
     id: string2().describe("Persisted eval analysis ID."),
+    prompt: string2().nullable().optional().describe("Exact task prompt used for this concrete attempt."),
+    prompt_index: int().default(0).describe("Zero-based prompt index within the submitted simulation."),
     repeat_index: int().default(0).describe("Zero-based repeat index within the submitted run."),
     result_id: string2().describe("Concrete eval result ID analyzed for this attempt."),
     updated_at: string2().nullable().optional(),
@@ -55588,46 +55462,64 @@ var init_v2evalrunanalysis = __esm(() => {
   });
 });
 
-// src/models/v2evalrunanalysisresponse.ts
-var V2EvalRunAnalysisResponse$zodSchema;
-var init_v2evalrunanalysisresponse = __esm(() => {
+// src/models/v2evalrundetail.ts
+var V2EvalRunDetail$zodSchema;
+var init_v2evalrundetail = __esm(() => {
   init_zod();
   init_publicevalrunprogress();
+  init_v2evalresultsummary();
   init_v2evalrunanalysis();
-  V2EvalRunAnalysisResponse$zodSchema = object({
-    data: array(V2EvalRunAnalysis$zodSchema).optional(),
+  V2EvalRunDetail$zodSchema = object({
+    agent: array(string2()).optional().describe("Runnable agents selected for this simulation."),
+    analysis: array(V2EvalRunAnalysis$zodSchema).optional().describe("Per-attempt analysis for this run. Empty until an attempt has been analyzed."),
+    attempt_count: int().default(1).describe("Total concrete attempts after all fan-out dimensions."),
+    completed_at: string2().nullable().optional(),
+    configuration_id: string2().nullable().optional(),
+    created_at: string2().nullable().optional(),
+    dispatch_id: string2().nullable().optional(),
     error: string2().nullable().optional(),
+    estimated_cost: number2().nullable().optional(),
+    eval_id: string2().nullable().optional().describe("Saved eval definition ID, including one created by a direct run."),
+    eval_name: string2().nullable().optional().describe("Display name captured for the eval when the run was created."),
+    id: string2().describe("Run ID to use when polling eval-run status."),
+    job_id: string2().describe("Alias of `id` for the durable parent job."),
+    persona: string2().nullable().optional().describe("Perspective applied to every attempt in this simulation."),
     progress: PublicEvalRunProgress$zodSchema.optional(),
-    run_id: string2(),
-    status: string2()
+    prompt_count: int().default(1).describe("Number of prompts included in the simulation."),
+    results: array(V2EvalResultSummary$zodSchema).optional().describe("Per-attempt outcomes for this run. Empty while the run is active; populated as attempts finish."),
+    results_truncated: boolean2().default(false).describe("True when the run produced more results than the inline cap. The remaining results are not included in this response."),
+    runs: int().default(1).describe("Independent attempts requested for every prompt-agent pair."),
+    stage: string2().nullable().optional(),
+    started_at: string2().nullable().optional(),
+    status: string2().describe("Terminal values: `completed`, `failed`, and `cancelled`. Recoverable interruptions are reported as `running` with stage `retrying`.")
   });
 });
 
-// src/models/v2evalrunsanalysisop.ts
-var V2EvalRunsAnalysisRequest$zodSchema, V2EvalRunsAnalysisResponseBody$zodSchema, V2EvalRunsAnalysisResponse$zodSchema;
-var init_v2evalrunsanalysisop = __esm(() => {
+// src/models/v2evalrunsretrieveop.ts
+var V2EvalRunsRetrieveRequest$zodSchema, V2EvalRunsRetrieveResponseBody$zodSchema, V2EvalRunsRetrieveResponse$zodSchema;
+var init_v2evalrunsretrieveop = __esm(() => {
   init_zod();
   init_httpvalidationerror();
-  init_v2evalrunanalysisresponse();
-  V2EvalRunsAnalysisRequest$zodSchema = object({
+  init_v2evalrundetail();
+  V2EvalRunsRetrieveRequest$zodSchema = object({
     run_id: string2()
   });
-  V2EvalRunsAnalysisResponseBody$zodSchema = object({
+  V2EvalRunsRetrieveResponseBody$zodSchema = object({
     detail: string2()
   }).describe("Eval run not found. Use the run ID returned when you started the eval.");
-  V2EvalRunsAnalysisResponse$zodSchema = union([
-    V2EvalRunAnalysisResponse$zodSchema,
-    lazy(() => V2EvalRunsAnalysisResponseBody$zodSchema),
+  V2EvalRunsRetrieveResponse$zodSchema = union([
+    V2EvalRunDetail$zodSchema,
+    lazy(() => V2EvalRunsRetrieveResponseBody$zodSchema),
     HTTPValidationError$zodSchema
   ]);
 });
 
-// src/funcs/evalsAnalysis.ts
-function evalsAnalysis(client$, request, options) {
-  return new APIPromise($do11(client$, request, options));
+// src/funcs/evalsGet.ts
+function evalsGet(client$, request, options) {
+  return new APIPromise($do9(client$, request, options));
 }
-async function $do11(client$, request, options) {
-  const parsed$ = safeParse4(request, (value$) => V2EvalRunsAnalysisRequest$zodSchema.parse(value$), "Input validation failed");
+async function $do9(client$, request, options) {
+  const parsed$ = safeParse4(request, (value$) => V2EvalRunsRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
   }
@@ -55639,7 +55531,7 @@ async function $do11(client$, request, options) {
       charEncoding: "percent"
     })
   };
-  const path$ = pathToFunc("/v2/eval-runs/{run_id}/analysis")(pathParams$);
+  const path$ = pathToFunc("/v2/eval-runs/{run_id}")(pathParams$);
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
   }));
@@ -55648,7 +55540,7 @@ async function $do11(client$, request, options) {
   const context = {
     options: client$._options,
     baseURL: options?.serverURL ?? client$._baseURL ?? "",
-    operationID: "v2_eval_runs_analysis",
+    operationID: "v2_eval_runs_retrieve",
     oAuth2Scopes: null,
     resolvedSecurity: requestSecurity,
     securitySource: client$._options.security,
@@ -55690,30 +55582,30 @@ async function $do11(client$, request, options) {
     response: doResult.value
   }];
 }
-var init_evalsAnalysis = __esm(() => {
+var init_evalsGet = __esm(() => {
   init_encodings();
   init_primitives();
   init_schemas4();
   init_security();
   init_url();
-  init_v2evalrunsanalysisop();
+  init_v2evalrunsretrieveop();
   init_async();
 });
 
-// src/mcp-server/tools/evalsAnalysis.ts
-var args8, tool$evalsAnalysis;
-var init_evalsAnalysis2 = __esm(() => {
-  init_evalsAnalysis();
-  init_v2evalrunsanalysisop();
+// src/mcp-server/tools/evalsGet.ts
+var args6, tool$evalsGet;
+var init_evalsGet2 = __esm(() => {
+  init_evalsGet();
+  init_v2evalrunsretrieveop();
   init_tools();
-  args8 = {
-    request: V2EvalRunsAnalysisRequest$zodSchema
+  args6 = {
+    request: V2EvalRunsRetrieveRequest$zodSchema
   };
-  tool$evalsAnalysis = {
-    name: "evals-analysis",
-    description: `Retrieve eval run analysis
+  tool$evalsGet = {
+    name: "evals-get",
+    description: `Retrieve an eval run
 
-Returns the persisted analysis produced for each completed attempt in one eval run. The analysis includes the verdict, outcome explanation, findings, bottleneck, and analysis sections. Active runs return the analysis available so far together with current run status and progress.`,
+Returns the run's current status and progress. Once its attempts finish, the response also includes their results and analysis. Poll this until the status reaches a terminal value. Use the trace operation for the raw execution event log, or the monitor operation to follow a run live.`,
     annotations: {
       title: "",
       destructiveHint: false,
@@ -55721,9 +55613,9 @@ Returns the persisted analysis produced for each completed attempt in one eval r
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args8,
+    args: args6,
     tool: async (client, args, ctx) => {
-      const [result] = await evalsAnalysis(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
+      const [result] = await evalsGet(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
         return {
           content: [{ type: "text", text: result.error.message }],
@@ -55735,27 +55627,76 @@ Returns the persisted analysis produced for each completed attempt in one eval r
   };
 });
 
+// src/models/v2inlinemcpdefinition.ts
+var Type$zodSchema, V2InlineMcpDefinition$zodSchema;
+var init_v2inlinemcpdefinition = __esm(() => {
+  init_zod();
+  Type$zodSchema = _enum([
+    "http",
+    "sse",
+    "stdio"
+  ]);
+  V2InlineMcpDefinition$zodSchema = object({
+    args: array(string2()).optional(),
+    command: string2().nullable().optional().describe("Executable command for a `stdio` MCP server."),
+    env: record(string2(), string2()).optional().describe("Environment mapping for a `stdio` MCP server. Values may reference run secrets with `${VARIABLE_NAME}`."),
+    headers: record(string2(), string2()).optional().describe("HTTP headers. Reference run secrets with `${VARIABLE_NAME}` instead of embedding credentials."),
+    name: string2().describe("Stable server name used in the selected agent's MCP config."),
+    type: Type$zodSchema,
+    url: string2().nullable().optional().describe("HTTPS endpoint for an `http` or `sse` MCP server.")
+  });
+});
+
+// src/models/v2savedmcpreference.ts
+var V2SavedMcpReference$zodSchema;
+var init_v2savedmcpreference = __esm(() => {
+  init_zod();
+  V2SavedMcpReference$zodSchema = object({
+    id: string2().describe("Saved MCP server ID from `GET /v2/mcp-servers`.")
+  });
+});
+
 // src/models/createv2evalrunrequest.ts
-var CreateV2EvalRunRequest$zodSchema;
+var CreateV2EvalRunRequestMcp$zodSchema, CreateV2EvalRunRequest$zodSchema;
 var init_createv2evalrunrequest = __esm(() => {
   init_zod();
+  init_searchgroundingmode();
+  init_v2inlinemcpdefinition();
+  init_v2savedmcpreference();
+  CreateV2EvalRunRequestMcp$zodSchema = union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema
+  ]);
   CreateV2EvalRunRequest$zodSchema = object({
-    agent_id: string2().describe("Required in both request forms. Runnable agent ID from `GET /v2/agents`."),
-    configuration_id: string2().optional().describe("Saved-run form only. Saved runtime configuration; required when `eval_id` is provided."),
-    env: record(string2(), string2()).optional().describe("Direct create-and-run form only. Write-only environment variables; values are stored only in the queued job's private configuration."),
+    agent: array(string2()).nullable().optional().describe("Runnable agents from `GET /v2/agents`. Required unless the saved eval already has agents. CLI clients expose repeatable `--agent`."),
+    cleanup_commands: array(string2()).nullable().optional().describe("Best-effort shell commands run sequentially after agent execution and result capture, before sandbox deletion. In saved mode, an explicit empty array disables saved cleanup commands for this launch. CLI clients expose repeatable `--cleanup-command`."),
+    clis: array(string2()).nullable().optional().describe("Saved CLI IDs or shell installation commands. UUID values resolve saved CLIs from `GET /v2/clis`; every other value is run as an installation command. In saved mode, an explicit empty array disables saved CLIs for this launch. CLI clients expose repeatable `--cli`."),
+    configuration_id: string2().optional().describe("Saved-run form only. Optional saved repository, environment, skills, and legacy tool configuration."),
+    env: record(string2(), string2()).optional().describe("Write-only environment variables for this launch. Values are stored only in the queued job's private configuration."),
     eval_id: string2().optional().describe("Saved-run form only. Existing eval definition to run. Omit this field to use the direct create-and-run form."),
-    judges: array(string2()).optional().describe("Direct create-and-run form only. Provide at least one success criterion. CLI clients may expose this string array as a repeatable `--judge` flag."),
-    prompt: string2().optional().describe("Direct create-and-run form only. Task prompt; required when `eval_id` is omitted."),
+    judges: array(string2()).nullable().optional().describe("Success criteria. Required in direct mode; in saved mode, replaces saved judges for this launch. CLI clients expose repeatable `--judge`."),
+    mcps: array(union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema
+    ])).nullable().optional().describe("Saved MCP references or inline portable definitions. In saved mode, an explicit empty array disables saved MCPs for this launch."),
+    name: string2().optional().describe("Direct create-and-run form only. Display name for the durable eval. Defaults to the first prompt."),
+    persona: string2().optional().describe("Optional perspective for every attempt in this launch. It does not modify prompts or add a fan-out dimension."),
+    prompts: array(string2()).nullable().optional().describe("Task prompts. Required in direct mode; in saved mode, replaces saved prompts for this launch. CLI clients expose repeatable `--prompt`."),
     repository: string2().optional().describe("Direct create-and-run form only. A saved repository ID, built-in alias such as `nextjs-starter`, or a public GitHub HTTPS URL."),
-    skills: array(string2()).optional().describe("Additional public skill packages to install before the agent starts in either request form, using GitHub `owner/repository` references such as `resend/resend-skills`. CLI clients may expose this string array as a repeatable `--skill` flag.")
-  }).describe("Start an eval run using one of two flat request forms. Saved runs require `eval_id`, `configuration_id`, and `agent_id`. Direct create-and-run requests require `prompt`, at least one item in `judges`, and `agent_id`. Do not combine saved-run and direct-run fields; no separate mode field is required.");
+    runs: int().optional().describe("Independent attempts for every prompt-agent pair. Defaults to the saved eval value or 1 for a direct run."),
+    search_grounding: SearchGroundingMode$zodSchema.nullable().optional().describe("Web-retrieval policy for this launch. Overrides the selected configuration or saved eval when provided; direct runs default to `default`."),
+    setup_commands: array(string2()).nullable().optional().describe("Shell commands run sequentially in the prepared repository before the agent starts. In saved mode, an explicit empty array disables saved setup commands for this launch. CLI clients expose repeatable `--setup-command`."),
+    skill_ids: array(string2()).optional().describe("Direct create-and-run form only. Reusable skill IDs from `GET /v2/skills` to install before the agent starts."),
+    skills: array(string2()).optional().describe("Additional public skill packages to install before the agent starts in either request form, using GitHub `owner/repository` references such as `resend/resend-skills`. CLI clients may expose this string array as a repeatable `--skill` flag."),
+    tags: array(string2()).optional().describe("Direct create-and-run form only. Optional labels saved on the created eval. CLI clients may expose this string array as a repeatable `--tag` flag.")
+  }).describe("Start an eval run using one of two flat request forms. Saved runs require `eval_id` and use saved simulation defaults unless the request replaces them. Direct create-and-run requests require `prompts`, `judges`, and `agent`. No separate mode field is required.");
 });
 
 // src/funcs/evalsRun.ts
 function evalsRun(client$, request, options) {
-  return new APIPromise($do12(client$, request, options));
+  return new APIPromise($do10(client$, request, options));
 }
-async function $do12(client$, request, options) {
+async function $do10(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => CreateV2EvalRunRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -55825,26 +55766,28 @@ var init_evalsRun = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsRun.ts
-var args9, tool$evalsRun;
+var args7, tool$evalsRun;
 var init_evalsRun2 = __esm(() => {
   init_evalsRun();
   init_createv2evalrunrequest();
   init_tools();
-  args9 = {
+  args7 = {
     request: CreateV2EvalRunRequest$zodSchema
   };
   tool$evalsRun = {
     name: "evals-run",
     description: `Start an eval run
 
-Queues a durable eval job and returns immediately with a polling ID. Poll \`GET /v2/eval-runs/{run_id}\` until its status reaches a terminal value.
+Queues a durable eval job and returns immediately with a polling ID. Poll \`GET /v2/eval-runs/{run_id}\` until its status reaches a terminal value; once finished it also returns the results and analysis. Follow a run live with the monitor operation.
 
 Send one of these flat request forms:
 
-- **Saved eval:** provide \`eval_id\`, \`configuration_id\`, and \`agent_id\`.
-- **Direct create-and-run:** provide \`prompt\`, \`judges\`, and \`agent_id\`. You may also provide a repository, write-only environment variables, and public skill packages.
+- **Saved eval:** provide \`eval_id\`; optionally replace its prompts, persona, agents, CLIs, MCPs, setup commands, cleanup commands, or runs for this launch. A saved configuration can still provide repository, environment, skills, and legacy tool selections.
+- **Direct create-and-run:** provide \`prompts\`, \`judges\`, and \`agent\`. You may also provide a persona, tags, repository, write-only environment variables, saved or inline CLIs and MCP servers, skills, setup commands, and cleanup commands.
+
+The run fans out across prompts × agents × runs. Persona is shared metadata and does not multiply attempts. No separate \`mode\` field is required. If the organization lacks the estimated credits, the request fails with \`402\` before dispatching work.
 
-Do not combine saved-run and direct-run fields. No separate \`mode\` field is required.`,
+A saved or inline environment containing an empty or placeholder credential is rejected with \`400\` before any evaluation is dispatched.`,
     annotations: {
       title: "",
       destructiveHint: false,
@@ -55852,7 +55795,7 @@ Do not combine saved-run and direct-run fields. No separate \`mode\` field is re
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args9,
+    args: args7,
     tool: async (client, args, ctx) => {
       const [result] = await evalsRun(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -55871,8 +55814,7 @@ var V2EvalRunTraceEvent$zodSchema;
 var init_v2evalruntraceevent = __esm(() => {
   init_zod();
   V2EvalRunTraceEvent$zodSchema = object({
-    agent_id: string2().nullable().optional(),
-    agent_name: string2().nullable().optional(),
+    agent: string2().nullable().optional(),
     category: string2(),
     child_job_id: string2().describe("Child eval-attempt ID associated with the event."),
     configuration_id: string2().nullable().optional(),
@@ -55881,9 +55823,11 @@ var init_v2evalruntraceevent = __esm(() => {
     eval_name: string2().nullable().optional(),
     event_id: int(),
     message: string2(),
+    prompt_index: int().default(0),
+    repeat_index: int().default(0),
     run_id: string2(),
     timestamp: string2().nullable().optional(),
-    type: string2()
+    type: string2().describe("Event type. Events are not tool invocations: one invocation usually emits a `tool_call` event and a `tool_result` event, and can emit extra `tool_call` updates as its arguments stream. Count invocations with the result's `tool_calls`, not with trace events.")
   });
 });
 
@@ -55927,9 +55871,9 @@ var init_v2evalrunstraceop = __esm(() => {
 
 // src/funcs/evalsTrace.ts
 function evalsTrace(client$, request, options) {
-  return new APIPromise($do13(client$, request, options));
+  return new APIPromise($do11(client$, request, options));
 }
-async function $do13(client$, request, options) {
+async function $do11(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalRunsTraceRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -56009,12 +55953,12 @@ var init_evalsTrace = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsTrace.ts
-var args10, tool$evalsTrace;
+var args8, tool$evalsTrace;
 var init_evalsTrace2 = __esm(() => {
   init_evalsTrace();
   init_v2evalrunstraceop();
   init_tools();
-  args10 = {
+  args8 = {
     request: V2EvalRunsTraceRequest$zodSchema
   };
   tool$evalsTrace = {
@@ -56029,7 +55973,7 @@ Returns a finite page of sanitized execution events for one eval run. Use \`next
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args10,
+    args: args8,
     tool: async (client, args, ctx) => {
       const [result] = await evalsTrace(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -56047,6 +55991,7 @@ Returns a finite page of sanitized execution events for one eval run. Use \`next
 var CreateV2EvalConfigurationRequest$zodSchema;
 var init_createv2evalconfigurationrequest = __esm(() => {
   init_zod();
+  init_searchgroundingmode();
   CreateV2EvalConfigurationRequest$zodSchema = object({
     cli_ids: array(string2()).optional(),
     environment_id: string2().optional(),
@@ -56054,6 +55999,7 @@ var init_createv2evalconfigurationrequest = __esm(() => {
     mcp_server_ids: array(string2()).optional(),
     name: string2(),
     repository_id: string2().optional(),
+    search_grounding: SearchGroundingMode$zodSchema.optional().describe("Requested web-retrieval behavior for an eval run."),
     skill_ids: array(string2()).optional(),
     skills: array(string2()).optional().describe("Public skill packages saved with this configuration, using GitHub `owner/repository` references such as `resend/resend-skills`.")
   });
@@ -56063,6 +56009,7 @@ var init_createv2evalconfigurationrequest = __esm(() => {
 var V2EvalConfiguration$zodSchema;
 var init_v2evalconfiguration = __esm(() => {
   init_zod();
+  init_searchgroundingmode();
   V2EvalConfiguration$zodSchema = object({
     cli_ids: array(string2()).optional(),
     created_at: string2().nullable().optional(),
@@ -56073,6 +56020,7 @@ var init_v2evalconfiguration = __esm(() => {
     mcp_server_ids: array(string2()).optional(),
     name: string2(),
     repository_id: string2().nullable().optional(),
+    search_grounding: SearchGroundingMode$zodSchema.optional().describe("Requested web-retrieval behavior for an eval run."),
     skill_ids: array(string2()).optional(),
     skills: array(string2()).optional().describe("Public GitHub `owner/repository` skill packages installed for this configuration."),
     updated_at: string2().nullable().optional()
@@ -56098,9 +56046,9 @@ var init_v2evalconfigurationscreateop = __esm(() => {
 
 // src/funcs/evalsV2EvalConfigurationsCreate.ts
 function evalsV2EvalConfigurationsCreate(client$, request, options) {
-  return new APIPromise($do14(client$, request, options));
+  return new APIPromise($do12(client$, request, options));
 }
-async function $do14(client$, request, options) {
+async function $do12(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalConfigurationsCreateRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -56176,12 +56124,12 @@ var init_evalsV2EvalConfigurationsCreate = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsV2EvalConfigurationsCreate.ts
-var args11, tool$evalsV2EvalConfigurationsCreate;
+var args9, tool$evalsV2EvalConfigurationsCreate;
 var init_evalsV2EvalConfigurationsCreate2 = __esm(() => {
   init_evalsV2EvalConfigurationsCreate();
   init_v2evalconfigurationscreateop();
   init_tools();
-  args11 = {
+  args9 = {
     request: V2EvalConfigurationsCreateRequest$zodSchema
   };
   tool$evalsV2EvalConfigurationsCreate = {
@@ -56196,7 +56144,7 @@ Creates a saved runtime configuration that pairs an eval with a repository, envi
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args11,
+    args: args9,
     tool: async (client, args, ctx) => {
       const [result] = await evalsV2EvalConfigurationsCreate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -56228,9 +56176,9 @@ var init_v2evalconfigurationsdeleteop = __esm(() => {
 
 // src/funcs/evalsV2EvalConfigurationsDelete.ts
 function evalsV2EvalConfigurationsDelete(client$, request, options) {
-  return new APIPromise($do15(client$, request, options));
+  return new APIPromise($do13(client$, request, options));
 }
-async function $do15(client$, request, options) {
+async function $do13(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalConfigurationsDeleteRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -56306,12 +56254,12 @@ var init_evalsV2EvalConfigurationsDelete = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsV2EvalConfigurationsDelete.ts
-var args12, tool$evalsV2EvalConfigurationsDelete;
+var args10, tool$evalsV2EvalConfigurationsDelete;
 var init_evalsV2EvalConfigurationsDelete2 = __esm(() => {
   init_evalsV2EvalConfigurationsDelete();
   init_v2evalconfigurationsdeleteop();
   init_tools();
-  args12 = {
+  args10 = {
     request: V2EvalConfigurationsDeleteRequest$zodSchema
   };
   tool$evalsV2EvalConfigurationsDelete = {
@@ -56326,7 +56274,7 @@ Permanently deletes an eval configuration and returns its last saved state.`,
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args12,
+    args: args10,
     tool: async (client, args, ctx) => {
       const [result] = await evalsV2EvalConfigurationsDelete(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -56368,9 +56316,9 @@ var init_v2evalconfigurationslistop = __esm(() => {
 
 // src/funcs/evalsV2EvalConfigurationsList.ts
 function evalsV2EvalConfigurationsList(client$, request, options) {
-  return new APIPromise($do16(client$, request, options));
+  return new APIPromise($do14(client$, request, options));
 }
-async function $do16(client$, request, options) {
+async function $do14(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalConfigurationsListRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -56445,12 +56393,12 @@ var init_evalsV2EvalConfigurationsList = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsV2EvalConfigurationsList.ts
-var args13, tool$evalsV2EvalConfigurationsList;
+var args11, tool$evalsV2EvalConfigurationsList;
 var init_evalsV2EvalConfigurationsList2 = __esm(() => {
   init_evalsV2EvalConfigurationsList();
   init_v2evalconfigurationslistop();
   init_tools();
-  args13 = {
+  args11 = {
     request: V2EvalConfigurationsListRequest$zodSchema
   };
   tool$evalsV2EvalConfigurationsList = {
@@ -56465,7 +56413,7 @@ Returns the saved runtime configurations for an eval, including repository, envi
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args13,
+    args: args11,
     tool: async (client, args, ctx) => {
       const [result] = await evalsV2EvalConfigurationsList(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -56497,9 +56445,9 @@ var init_v2evalconfigurationsretrieveop = __esm(() => {
 
 // src/funcs/evalsV2EvalConfigurationsRetrieve.ts
 function evalsV2EvalConfigurationsRetrieve(client$, request, options) {
-  return new APIPromise($do17(client$, request, options));
+  return new APIPromise($do15(client$, request, options));
 }
-async function $do17(client$, request, options) {
+async function $do15(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalConfigurationsRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -56575,12 +56523,12 @@ var init_evalsV2EvalConfigurationsRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsV2EvalConfigurationsRetrieve.ts
-var args14, tool$evalsV2EvalConfigurationsRetrieve;
+var args12, tool$evalsV2EvalConfigurationsRetrieve;
 var init_evalsV2EvalConfigurationsRetrieve2 = __esm(() => {
   init_evalsV2EvalConfigurationsRetrieve();
   init_v2evalconfigurationsretrieveop();
   init_tools();
-  args14 = {
+  args12 = {
     request: V2EvalConfigurationsRetrieveRequest$zodSchema
   };
   tool$evalsV2EvalConfigurationsRetrieve = {
@@ -56595,7 +56543,7 @@ Returns one saved configuration for an eval.`,
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args14,
+    args: args12,
     tool: async (client, args, ctx) => {
       const [result] = await evalsV2EvalConfigurationsRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -56613,6 +56561,7 @@ Returns one saved configuration for an eval.`,
 var UpdateV2EvalConfigurationRequest$zodSchema;
 var init_updatev2evalconfigurationrequest = __esm(() => {
   init_zod();
+  init_searchgroundingmode();
   UpdateV2EvalConfigurationRequest$zodSchema = object({
     cli_ids: array(string2()).nullable().optional(),
     environment_id: string2().optional(),
@@ -56620,6 +56569,7 @@ var init_updatev2evalconfigurationrequest = __esm(() => {
     mcp_server_ids: array(string2()).nullable().optional(),
     name: string2().optional(),
     repository_id: string2().optional(),
+    search_grounding: SearchGroundingMode$zodSchema.nullable().optional().describe("Replacement web-retrieval policy. Omit it to keep the saved value."),
     skill_ids: array(string2()).nullable().optional(),
     skills: array(string2()).nullable().optional().describe("Public skill packages saved with this configuration, using GitHub `owner/repository` references.")
   });
@@ -56645,9 +56595,9 @@ var init_v2evalconfigurationsupdateop = __esm(() => {
 
 // src/funcs/evalsV2EvalConfigurationsUpdate.ts
 function evalsV2EvalConfigurationsUpdate(client$, request, options) {
-  return new APIPromise($do18(client$, request, options));
+  return new APIPromise($do16(client$, request, options));
 }
-async function $do18(client$, request, options) {
+async function $do16(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalConfigurationsUpdateRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -56724,12 +56674,12 @@ var init_evalsV2EvalConfigurationsUpdate = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsV2EvalConfigurationsUpdate.ts
-var args15, tool$evalsV2EvalConfigurationsUpdate;
+var args13, tool$evalsV2EvalConfigurationsUpdate;
 var init_evalsV2EvalConfigurationsUpdate2 = __esm(() => {
   init_evalsV2EvalConfigurationsUpdate();
   init_v2evalconfigurationsupdateop();
   init_tools();
-  args15 = {
+  args13 = {
     request: V2EvalConfigurationsUpdateRequest$zodSchema
   };
   tool$evalsV2EvalConfigurationsUpdate = {
@@ -56744,7 +56694,7 @@ Updates an eval configuration; omitted fields remain unchanged.`,
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args15,
+    args: args13,
     tool: async (client, args, ctx) => {
       const [result] = await evalsV2EvalConfigurationsUpdate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -56759,31 +56709,40 @@ Updates an eval configuration; omitted fields remain unchanged.`,
 });
 
 // src/models/createv2evalrequest.ts
-var CreateV2EvalRequestType$zodSchema, CreateV2EvalRequest$zodSchema;
+var CreateV2EvalRequestMcp$zodSchema, CreateV2EvalRequest$zodSchema;
 var init_createv2evalrequest = __esm(() => {
   init_zod();
-  CreateV2EvalRequestType$zodSchema = _enum([
-    "workflow",
-    "mcp",
-    "cli",
-    "skill"
+  init_searchgroundingmode();
+  init_v2inlinemcpdefinition();
+  init_v2savedmcpreference();
+  CreateV2EvalRequestMcp$zodSchema = union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema
   ]);
   CreateV2EvalRequest$zodSchema = object({
-    cli_id: string2().optional(),
-    judge: string2().describe("Plain-text outcome that counts as success."),
-    mcp_server_id: string2().optional(),
+    agent: array(string2()).optional().describe("Runnable agents from `GET /v2/agents`. An empty array saves the eval without default agents. CLI clients expose repeatable `--agent`."),
+    cleanup_commands: array(string2()).optional().describe("Best-effort shell commands run sequentially after agent execution and result capture, before sandbox deletion. CLI clients expose repeatable `--cleanup-command`."),
+    clis: array(string2()).optional().describe("Saved CLI IDs or shell installation commands. UUID values resolve saved CLIs from `GET /v2/clis`; every other value is run as an installation command. CLI clients expose repeatable `--cli`."),
+    judges: array(string2()).describe("Plain-text success criteria. CLI clients expose this as repeatable `--judge`."),
+    mcps: array(union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema
+    ])).optional().describe("Saved MCP references or inline portable MCP definitions."),
     name: string2(),
-    prompt: string2(),
-    skill_id: string2().optional(),
-    type: CreateV2EvalRequestType$zodSchema
+    persona: string2().optional().describe("Optional perspective applied to the whole eval. It is stored as metadata and does not change prompts or multiply attempts."),
+    prompts: array(string2()).describe("Task prompts. CLI clients expose this as repeatable `--prompt`."),
+    runs: int().default(1).describe("Independent attempts for every prompt-agent pair."),
+    search_grounding: SearchGroundingMode$zodSchema.optional().describe("Requested web-retrieval behavior for an eval run."),
+    setup_commands: array(string2()).optional().describe("Shell commands run sequentially in the prepared repository before the agent starts. CLI clients expose repeatable `--setup-command`."),
+    tags: array(string2()).optional().describe("Optional labels for organizing the eval. CLI clients may expose this string array as a repeatable `--tag` flag.")
   });
 });
 
 // src/funcs/evalsV2EvalsCreate.ts
 function evalsV2EvalsCreate(client$, request, options) {
-  return new APIPromise($do19(client$, request, options));
+  return new APIPromise($do17(client$, request, options));
 }
-async function $do19(client$, request, options) {
+async function $do17(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => CreateV2EvalRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -56853,19 +56812,19 @@ var init_evalsV2EvalsCreate = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsV2EvalsCreate.ts
-var args16, tool$evalsV2EvalsCreate;
+var args14, tool$evalsV2EvalsCreate;
 var init_evalsV2EvalsCreate2 = __esm(() => {
   init_evalsV2EvalsCreate();
   init_createv2evalrequest();
   init_tools();
-  args16 = {
+  args14 = {
     request: CreateV2EvalRequest$zodSchema
   };
   tool$evalsV2EvalsCreate = {
     name: "evals-v2-evals-create",
     description: `Create Eval
 
-Creates an eval with a task prompt, subject, and plain-text success criterion in \`judge\`. The eval can then be run through eval runs.`,
+Creates a reusable simulation definition with one or more task prompts and judges. Optional agents, persona, CLI installations, MCP servers, setup commands, cleanup commands, grounding policy, repeat count, and tags are saved with the eval. \`search_grounding\` supports \`disabled\`, \`enabled\`, and \`default\`; \`default\` preserves normal platform behavior. Creating the eval does not start a run and does not accept secret environment values.`,
     annotations: {
       title: "",
       destructiveHint: false,
@@ -56873,7 +56832,7 @@ Creates an eval with a task prompt, subject, and plain-text success criterion in
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args16,
+    args: args14,
     tool: async (client, args, ctx) => {
       const [result] = await evalsV2EvalsCreate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -56887,36 +56846,62 @@ Creates an eval with a task prompt, subject, and plain-text success criterion in
   };
 });
 
-// src/models/v2evaljudge.ts
-var V2EvalJudge$zodSchema;
-var init_v2evaljudge = __esm(() => {
+// src/models/v2inlineclireference.ts
+var V2InlineCliReference$zodSchema;
+var init_v2inlineclireference = __esm(() => {
+  init_zod();
+  V2InlineCliReference$zodSchema = object({
+    install_command: string2().describe("Shell command to install this CLI before the agent starts.")
+  });
+});
+
+// src/models/v2savedclireference.ts
+var V2SavedCliReference$zodSchema;
+var init_v2savedclireference = __esm(() => {
   init_zod();
-  V2EvalJudge$zodSchema = object({
-    criteria: string2().describe("The outcome that counts as success.")
+  V2SavedCliReference$zodSchema = object({
+    id: string2().describe("Saved CLI ID from `GET /v2/clis`.")
   });
 });
 
 // src/models/v2eval.ts
-var V2EvalType$zodSchema, V2Eval$zodSchema;
+var Cli$zodSchema, V2EvalMcp$zodSchema, V2Eval$zodSchema;
 var init_v2eval = __esm(() => {
   init_zod();
-  init_v2evaljudge();
-  V2EvalType$zodSchema = _enum([
-    "workflow",
-    "mcp",
-    "cli",
-    "skill"
+  init_searchgroundingmode();
+  init_v2inlineclireference();
+  init_v2inlinemcpdefinition();
+  init_v2savedclireference();
+  init_v2savedmcpreference();
+  Cli$zodSchema = union([
+    V2SavedCliReference$zodSchema,
+    V2InlineCliReference$zodSchema
+  ]);
+  V2EvalMcp$zodSchema = union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema
   ]);
   V2Eval$zodSchema = object({
-    cli_id: string2().nullable().optional(),
+    agent: array(string2()).optional(),
+    cleanup_commands: array(string2()).optional(),
+    clis: array(union([
+      V2SavedCliReference$zodSchema,
+      V2InlineCliReference$zodSchema
+    ])).optional(),
     created_at: string2().nullable().optional(),
     id: string2(),
-    judge: V2EvalJudge$zodSchema,
-    mcp_server_id: string2().nullable().optional(),
+    judges: array(string2()).optional(),
+    mcps: array(union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema
+    ])).optional(),
     name: string2(),
-    prompt: string2(),
-    skill_id: string2().nullable().optional(),
-    type: V2EvalType$zodSchema,
+    persona: string2().nullable().optional(),
+    prompts: array(string2()).optional(),
+    runs: int().default(1),
+    search_grounding: SearchGroundingMode$zodSchema.optional().describe("Requested web-retrieval behavior for an eval run."),
+    setup_commands: array(string2()).optional(),
+    tags: array(string2()).optional(),
     updated_at: string2().nullable().optional()
   });
 });
@@ -56938,9 +56923,9 @@ var init_v2evalsdeleteop = __esm(() => {
 
 // src/funcs/evalsV2EvalsDelete.ts
 function evalsV2EvalsDelete(client$, request, options) {
-  return new APIPromise($do20(client$, request, options));
+  return new APIPromise($do18(client$, request, options));
 }
-async function $do20(client$, request, options) {
+async function $do18(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalsDeleteRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -57015,12 +57000,12 @@ var init_evalsV2EvalsDelete = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsV2EvalsDelete.ts
-var args17, tool$evalsV2EvalsDelete;
+var args15, tool$evalsV2EvalsDelete;
 var init_evalsV2EvalsDelete2 = __esm(() => {
   init_evalsV2EvalsDelete();
   init_v2evalsdeleteop();
   init_tools();
-  args17 = {
+  args15 = {
     request: V2EvalsDeleteRequest$zodSchema
   };
   tool$evalsV2EvalsDelete = {
@@ -57035,7 +57020,7 @@ Permanently deletes an eval and all of its saved configurations.`,
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args17,
+    args: args15,
     tool: async (client, args, ctx) => {
       const [result] = await evalsV2EvalsDelete(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -57051,9 +57036,9 @@ Permanently deletes an eval and all of its saved configurations.`,
 
 // src/funcs/evalsV2EvalsList.ts
 function evalsV2EvalsList(client$, options) {
-  return new APIPromise($do21(client$, options));
+  return new APIPromise($do19(client$, options));
 }
-async function $do21(client$, options) {
+async function $do19(client$, options) {
   const path$ = pathToFunc("/v2/evals")();
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
@@ -57120,7 +57105,7 @@ var init_evalsV2EvalsList2 = __esm(() => {
     name: "evals-v2-evals-list",
     description: `List Evals
 
-Returns all active evals including workflow, MCP, CLI, and skill evals owned by the organization.`,
+Returns all active eval definitions owned by the organization.`,
     annotations: {
       title: "",
       destructiveHint: false,
@@ -57158,9 +57143,9 @@ var init_v2evalsretrieveop = __esm(() => {
 
 // src/funcs/evalsV2EvalsRetrieve.ts
 function evalsV2EvalsRetrieve(client$, request, options) {
-  return new APIPromise($do22(client$, request, options));
+  return new APIPromise($do20(client$, request, options));
 }
-async function $do22(client$, request, options) {
+async function $do20(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalsRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -57235,19 +57220,19 @@ var init_evalsV2EvalsRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsV2EvalsRetrieve.ts
-var args18, tool$evalsV2EvalsRetrieve;
+var args16, tool$evalsV2EvalsRetrieve;
 var init_evalsV2EvalsRetrieve2 = __esm(() => {
   init_evalsV2EvalsRetrieve();
   init_v2evalsretrieveop();
   init_tools();
-  args18 = {
+  args16 = {
     request: V2EvalsRetrieveRequest$zodSchema
   };
   tool$evalsV2EvalsRetrieve = {
     name: "evals-v2-evals-retrieve",
     description: `Retrieve Eval
 
-Returns one eval and its judge configuration.`,
+Returns one reusable eval simulation definition.`,
     annotations: {
       title: "",
       destructiveHint: false,
@@ -57255,7 +57240,7 @@ Returns one eval and its judge configuration.`,
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args18,
+    args: args16,
     tool: async (client, args, ctx) => {
       const [result] = await evalsV2EvalsRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -57270,23 +57255,32 @@ Returns one eval and its judge configuration.`,
 });
 
 // src/models/updatev2evalrequest.ts
-var UpdateV2EvalRequestType$zodSchema, UpdateV2EvalRequest$zodSchema;
+var UpdateV2EvalRequestMcp$zodSchema, UpdateV2EvalRequest$zodSchema;
 var init_updatev2evalrequest = __esm(() => {
   init_zod();
-  UpdateV2EvalRequestType$zodSchema = _enum([
-    "workflow",
-    "mcp",
-    "cli",
-    "skill"
+  init_searchgroundingmode();
+  init_v2inlinemcpdefinition();
+  init_v2savedmcpreference();
+  UpdateV2EvalRequestMcp$zodSchema = union([
+    V2InlineMcpDefinition$zodSchema,
+    V2SavedMcpReference$zodSchema
   ]);
   UpdateV2EvalRequest$zodSchema = object({
-    cli_id: string2().optional(),
-    judge: string2().optional().describe("Plain-text outcome that counts as success."),
-    mcp_server_id: string2().optional(),
+    agent: array(string2()).nullable().optional(),
+    cleanup_commands: array(string2()).nullable().optional().describe("Replacement cleanup commands. Pass an empty array to clear all saved cleanup commands."),
+    clis: array(string2()).nullable().optional().describe("Replacement saved CLI IDs or shell installation commands. UUID values resolve saved CLIs; every other value is an installation command. Pass an empty array to clear all CLIs."),
+    judges: array(string2()).nullable().optional().describe("Replacement success criteria for the eval."),
+    mcps: array(union([
+      V2InlineMcpDefinition$zodSchema,
+      V2SavedMcpReference$zodSchema
+    ])).nullable().optional(),
     name: string2().optional(),
-    prompt: string2().optional(),
-    skill_id: string2().optional(),
-    type: UpdateV2EvalRequestType$zodSchema.optional()
+    persona: string2().optional().describe("Replacement eval persona. Pass null to clear it; omit it to leave the saved value unchanged."),
+    prompts: array(string2()).nullable().optional(),
+    runs: int().optional(),
+    search_grounding: SearchGroundingMode$zodSchema.nullable().optional().describe("Replacement web-retrieval policy. Omit it to keep the saved value."),
+    setup_commands: array(string2()).nullable().optional().describe("Replacement setup commands. Pass an empty array to clear all saved setup commands."),
+    tags: array(string2()).nullable().optional().describe("Replacement labels for the eval. Pass an empty array to clear all tags.")
   });
 });
 
@@ -57309,9 +57303,9 @@ var init_v2evalsupdateop = __esm(() => {
 
 // src/funcs/evalsV2EvalsUpdate.ts
 function evalsV2EvalsUpdate(client$, request, options) {
-  return new APIPromise($do23(client$, request, options));
+  return new APIPromise($do21(client$, request, options));
 }
-async function $do23(client$, request, options) {
+async function $do21(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2EvalsUpdateRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -57387,19 +57381,19 @@ var init_evalsV2EvalsUpdate = __esm(() => {
 });
 
 // src/mcp-server/tools/evalsV2EvalsUpdate.ts
-var args19, tool$evalsV2EvalsUpdate;
+var args17, tool$evalsV2EvalsUpdate;
 var init_evalsV2EvalsUpdate2 = __esm(() => {
   init_evalsV2EvalsUpdate();
   init_v2evalsupdateop();
   init_tools();
-  args19 = {
+  args17 = {
     request: V2EvalsUpdateRequest$zodSchema
   };
   tool$evalsV2EvalsUpdate = {
     name: "evals-v2-evals-update",
     description: `Update Eval
 
-Updates an eval's task prompt, subject, plain-text \`judge\`, or name. Omitted fields remain unchanged.`,
+Updates a reusable eval simulation definition. Omitted fields remain unchanged. Empty \`agent\`, \`clis\`, \`mcps\`, \`setup_commands\`, \`cleanup_commands\`, or \`tags\` arrays clear those collections, and \`persona: null\` clears the saved persona.`,
     annotations: {
       title: "",
       destructiveHint: false,
@@ -57407,7 +57401,7 @@ Updates an eval's task prompt, subject, plain-text \`judge\`, or name. Omitted f
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args19,
+    args: args17,
     tool: async (client, args, ctx) => {
       const [result] = await evalsV2EvalsUpdate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -57423,9 +57417,9 @@ Updates an eval's task prompt, subject, plain-text \`judge\`, or name. Omitted f
 
 // src/funcs/mcpServersV2MCPServersList.ts
 function mcpServersV2MCPServersList(client$, options) {
-  return new APIPromise($do24(client$, options));
+  return new APIPromise($do22(client$, options));
 }
-async function $do24(client$, options) {
+async function $do22(client$, options) {
   const path$ = pathToFunc("/v2/mcp-servers")();
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
@@ -57607,9 +57601,9 @@ var init_v2opportunitieslistop = __esm(() => {
 
 // src/funcs/opportunitiesV2OpportunitiesList.ts
 function opportunitiesV2OpportunitiesList(client$, request, options) {
-  return new APIPromise($do25(client$, request, options));
+  return new APIPromise($do23(client$, request, options));
 }
-async function $do25(client$, request, options) {
+async function $do23(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2OpportunitiesListRequest$zodSchema.optional().parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -57687,12 +57681,12 @@ var init_opportunitiesV2OpportunitiesList = __esm(() => {
 });
 
 // src/mcp-server/tools/opportunitiesV2OpportunitiesList.ts
-var args20, tool$opportunitiesV2OpportunitiesList;
+var args18, tool$opportunitiesV2OpportunitiesList;
 var init_opportunitiesV2OpportunitiesList2 = __esm(() => {
   init_opportunitiesV2OpportunitiesList();
   init_v2opportunitieslistop();
   init_tools();
-  args20 = {
+  args18 = {
     request: V2OpportunitiesListRequest$zodSchema.optional()
   };
   tool$opportunitiesV2OpportunitiesList = {
@@ -57707,7 +57701,7 @@ Returns ranked opportunities computed from stored prompt visibility results. Eac
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args20,
+    args: args18,
     tool: async (client, args, ctx) => {
       const [result] = await opportunitiesV2OpportunitiesList(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -57779,9 +57773,9 @@ var init_v2opportunitiesretrieveop = __esm(() => {
 
 // src/funcs/opportunitiesV2OpportunitiesRetrieve.ts
 function opportunitiesV2OpportunitiesRetrieve(client$, request, options) {
-  return new APIPromise($do26(client$, request, options));
+  return new APIPromise($do24(client$, request, options));
 }
-async function $do26(client$, request, options) {
+async function $do24(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2OpportunitiesRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -57862,12 +57856,12 @@ var init_opportunitiesV2OpportunitiesRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/opportunitiesV2OpportunitiesRetrieve.ts
-var args21, tool$opportunitiesV2OpportunitiesRetrieve;
+var args19, tool$opportunitiesV2OpportunitiesRetrieve;
 var init_opportunitiesV2OpportunitiesRetrieve2 = __esm(() => {
   init_opportunitiesV2OpportunitiesRetrieve();
   init_v2opportunitiesretrieveop();
   init_tools();
-  args21 = {
+  args19 = {
     request: V2OpportunitiesRetrieveRequest$zodSchema
   };
   tool$opportunitiesV2OpportunitiesRetrieve = {
@@ -57882,7 +57876,7 @@ Returns one computed opportunity and its execution context for the selected date
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args21,
+    args: args19,
     tool: async (client, args, ctx) => {
       const [result] = await opportunitiesV2OpportunitiesRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -57951,7 +57945,7 @@ var init_v2promptrunsummary = __esm(() => {
     response_preview: string2().nullable().optional(),
     run_date: string2(),
     target_kind: V2PromptRunSummaryTargetKind$zodSchema.nullable().optional(),
-    tool_call_count: int().nullable().optional()
+    tool_call_count: int().nullable().optional().describe("Coding-agent tool invocations, counting a call and its result once. Matches the length of `tool_calls` on the detail response; null for answer-engine runs.")
   });
 });
 
@@ -57990,9 +57984,9 @@ var init_v2promptrunslistop = __esm(() => {
 
 // src/funcs/promptsV2PromptRunsList.ts
 function promptsV2PromptRunsList(client$, request, options) {
-  return new APIPromise($do27(client$, request, options));
+  return new APIPromise($do25(client$, request, options));
 }
-async function $do27(client$, request, options) {
+async function $do25(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2PromptRunsListRequest$zodSchema.optional().parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -58070,19 +58064,19 @@ var init_promptsV2PromptRunsList = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptRunsList.ts
-var args22, tool$promptsV2PromptRunsList;
+var args20, tool$promptsV2PromptRunsList;
 var init_promptsV2PromptRunsList2 = __esm(() => {
   init_promptsV2PromptRunsList();
   init_v2promptrunslistop();
   init_tools();
-  args22 = {
+  args20 = {
     request: V2PromptRunsListRequest$zodSchema.optional()
   };
   tool$promptsV2PromptRunsList = {
     name: "prompts-v2-prompt-runs-list",
     description: `List Prompt Runs
 
-Returns stored prompt runs without exposing raw runtime or provider configuration. Filterable by prompt, platform, and date range.`,
+Returns stored prompt runs with limit-and-offset pagination, without exposing raw runtime or provider configuration. Filter by prompt, platform, or date range.`,
     annotations: {
       title: "",
       destructiveHint: false,
@@ -58090,7 +58084,7 @@ Returns stored prompt runs without exposing raw runtime or provider configuratio
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args22,
+    args: args20,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptRunsList(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -58161,6 +58155,7 @@ var init_v2promptrundetail = __esm(() => {
   V2PromptRunDetail$zodSchema = object({
     citation_count: int().default(0),
     citations: array(V2PromptRunCitation$zodSchema).optional(),
+    conversation_turns: array(record(string2(), any())).optional().describe("Sanitized interleaved text and tool-call trace captured for this run. It interleaves the agent's `text` turns with `tool_call` turns, so its length is not a tool-call count; use `tool_calls` or `tool_call_count`."),
     created_at: string2().nullable().optional(),
     harness: string2().nullable().optional().describe("Coding-agent harness for agent runs; otherwise the answer platform identifier."),
     id: string2(),
@@ -58174,7 +58169,8 @@ var init_v2promptrundetail = __esm(() => {
     response_preview: string2().nullable().optional(),
     run_date: string2(),
     target_kind: V2PromptRunDetailTargetKind$zodSchema.nullable().optional(),
-    tool_call_count: int().nullable().optional()
+    tool_call_count: int().nullable().optional().describe("Coding-agent tool invocations, counting a call and its result once. Matches the length of `tool_calls` on the detail response; null for answer-engine runs."),
+    tool_calls: array(record(string2(), any())).optional().describe("Sanitized coding-agent tool invocations captured for this run, one entry per invocation with its result merged in.")
   });
 });
 
@@ -58195,9 +58191,9 @@ var init_v2promptrunsretrieveop = __esm(() => {
 
 // src/funcs/promptsV2PromptRunsRetrieve.ts
 function promptsV2PromptRunsRetrieve(client$, request, options) {
-  return new APIPromise($do28(client$, request, options));
+  return new APIPromise($do26(client$, request, options));
 }
-async function $do28(client$, request, options) {
+async function $do26(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2PromptRunsRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -58272,19 +58268,19 @@ var init_promptsV2PromptRunsRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptRunsRetrieve.ts
-var args23, tool$promptsV2PromptRunsRetrieve;
+var args21, tool$promptsV2PromptRunsRetrieve;
 var init_promptsV2PromptRunsRetrieve2 = __esm(() => {
   init_promptsV2PromptRunsRetrieve();
   init_v2promptrunsretrieveop();
   init_tools();
-  args23 = {
+  args21 = {
     request: V2PromptRunsRetrieveRequest$zodSchema
   };
   tool$promptsV2PromptRunsRetrieve = {
     name: "prompts-v2-prompt-runs-retrieve",
     description: `Retrieve Prompt Run
 
-Returns one stored prompt run, including its mentions and citations.`,
+Returns one stored prompt run, including mentions, citations, and any sanitized coding-agent trace captured for the run.`,
     annotations: {
       title: "",
       destructiveHint: false,
@@ -58292,7 +58288,7 @@ Returns one stored prompt run, including its mentions and citations.`,
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args23,
+    args: args21,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptRunsRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -58329,9 +58325,9 @@ var init_createv2promptsrequest = __esm(() => {
 
 // src/funcs/promptsV2PromptsCreate.ts
 function promptsV2PromptsCreate(client$, request, options) {
-  return new APIPromise($do29(client$, request, options));
+  return new APIPromise($do27(client$, request, options));
 }
-async function $do29(client$, request, options) {
+async function $do27(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => CreateV2PromptsRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -58401,12 +58397,12 @@ var init_promptsV2PromptsCreate = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptsCreate.ts
-var args24, tool$promptsV2PromptsCreate;
+var args22, tool$promptsV2PromptsCreate;
 var init_promptsV2PromptsCreate2 = __esm(() => {
   init_promptsV2PromptsCreate();
   init_createv2promptsrequest();
   init_tools();
-  args24 = {
+  args22 = {
     request: CreateV2PromptsRequest$zodSchema
   };
   tool$promptsV2PromptsCreate = {
@@ -58421,7 +58417,7 @@ Creates prompts in a batch. Each item is validated independently, so the respons
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args24,
+    args: args22,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptsCreate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -58461,9 +58457,9 @@ var init_v2promptsdeleteop = __esm(() => {
 
 // src/funcs/promptsV2PromptsDelete.ts
 function promptsV2PromptsDelete(client$, request, options) {
-  return new APIPromise($do30(client$, request, options));
+  return new APIPromise($do28(client$, request, options));
 }
-async function $do30(client$, request, options) {
+async function $do28(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2PromptsDeleteRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -58538,12 +58534,12 @@ var init_promptsV2PromptsDelete = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptsDelete.ts
-var args25, tool$promptsV2PromptsDelete;
+var args23, tool$promptsV2PromptsDelete;
 var init_promptsV2PromptsDelete2 = __esm(() => {
   init_promptsV2PromptsDelete();
   init_v2promptsdeleteop();
   init_tools();
-  args25 = {
+  args23 = {
     request: V2PromptsDeleteRequest$zodSchema
   };
   tool$promptsV2PromptsDelete = {
@@ -58558,7 +58554,7 @@ Archives the prompt and excludes it from normal listings while preserving its hi
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args25,
+    args: args23,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptsDelete(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -58668,9 +58664,9 @@ var init_v2promptslistop = __esm(() => {
 
 // src/funcs/promptsV2PromptsList.ts
 function promptsV2PromptsList(client$, request, options) {
-  return new APIPromise($do31(client$, request, options));
+  return new APIPromise($do29(client$, request, options));
 }
-async function $do31(client$, request, options) {
+async function $do29(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2PromptsListRequest$zodSchema.optional().parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -58745,12 +58741,12 @@ var init_promptsV2PromptsList = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptsList.ts
-var args26, tool$promptsV2PromptsList;
+var args24, tool$promptsV2PromptsList;
 var init_promptsV2PromptsList2 = __esm(() => {
   init_promptsV2PromptsList();
   init_v2promptslistop();
   init_tools();
-  args26 = {
+  args24 = {
     request: V2PromptsListRequest$zodSchema.optional()
   };
   tool$promptsV2PromptsList = {
@@ -58765,7 +58761,7 @@ Returns active and inactive prompts by default. Set \`include_archived\` to incl
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args26,
+    args: args24,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptsList(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -58796,9 +58792,9 @@ var init_v2promptsretrieveop = __esm(() => {
 
 // src/funcs/promptsV2PromptsRetrieve.ts
 function promptsV2PromptsRetrieve(client$, request, options) {
-  return new APIPromise($do32(client$, request, options));
+  return new APIPromise($do30(client$, request, options));
 }
-async function $do32(client$, request, options) {
+async function $do30(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2PromptsRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -58873,12 +58869,12 @@ var init_promptsV2PromptsRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptsRetrieve.ts
-var args27, tool$promptsV2PromptsRetrieve;
+var args25, tool$promptsV2PromptsRetrieve;
 var init_promptsV2PromptsRetrieve2 = __esm(() => {
   init_promptsV2PromptsRetrieve();
   init_v2promptsretrieveop();
   init_tools();
-  args27 = {
+  args25 = {
     request: V2PromptsRetrieveRequest$zodSchema
   };
   tool$promptsV2PromptsRetrieve = {
@@ -58893,7 +58889,7 @@ Returns a prompt with its topic, platforms, tags, and lifecycle status.`,
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args27,
+    args: args25,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptsRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -58942,9 +58938,9 @@ var init_v2promptsupdateop = __esm(() => {
 
 // src/funcs/promptsV2PromptsUpdate.ts
 function promptsV2PromptsUpdate(client$, request, options) {
-  return new APIPromise($do33(client$, request, options));
+  return new APIPromise($do31(client$, request, options));
 }
-async function $do33(client$, request, options) {
+async function $do31(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2PromptsUpdateRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -59020,12 +59016,12 @@ var init_promptsV2PromptsUpdate = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptsUpdate.ts
-var args28, tool$promptsV2PromptsUpdate;
+var args26, tool$promptsV2PromptsUpdate;
 var init_promptsV2PromptsUpdate2 = __esm(() => {
   init_promptsV2PromptsUpdate();
   init_v2promptsupdateop();
   init_tools();
-  args28 = {
+  args26 = {
     request: V2PromptsUpdateRequest$zodSchema
   };
   tool$promptsV2PromptsUpdate = {
@@ -59040,7 +59036,7 @@ Updates a prompt's text, topic, platforms, or lifecycle status. Omitted fields r
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args28,
+    args: args26,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptsUpdate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -59065,9 +59061,9 @@ var init_createprompttopicrequest = __esm(() => {
 
 // src/funcs/promptsV2PromptTopicsCreate.ts
 function promptsV2PromptTopicsCreate(client$, request, options) {
-  return new APIPromise($do34(client$, request, options));
+  return new APIPromise($do32(client$, request, options));
 }
-async function $do34(client$, request, options) {
+async function $do32(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => CreatePromptTopicRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -59137,12 +59133,12 @@ var init_promptsV2PromptTopicsCreate = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptTopicsCreate.ts
-var args29, tool$promptsV2PromptTopicsCreate;
+var args27, tool$promptsV2PromptTopicsCreate;
 var init_promptsV2PromptTopicsCreate2 = __esm(() => {
   init_promptsV2PromptTopicsCreate();
   init_createprompttopicrequest();
   init_tools();
-  args29 = {
+  args27 = {
     request: CreatePromptTopicRequest$zodSchema
   };
   tool$promptsV2PromptTopicsCreate = {
@@ -59157,7 +59153,7 @@ Creates a prompt topic and returns the saved topic.`,
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args29,
+    args: args27,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptTopicsCreate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -59189,9 +59185,9 @@ var init_v2prompttopicsdeleteop = __esm(() => {
 
 // src/funcs/promptsV2PromptTopicsDelete.ts
 function promptsV2PromptTopicsDelete(client$, request, options) {
-  return new APIPromise($do35(client$, request, options));
+  return new APIPromise($do33(client$, request, options));
 }
-async function $do35(client$, request, options) {
+async function $do33(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2PromptTopicsDeleteRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -59270,12 +59266,12 @@ var init_promptsV2PromptTopicsDelete = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptTopicsDelete.ts
-var args30, tool$promptsV2PromptTopicsDelete;
+var args28, tool$promptsV2PromptTopicsDelete;
 var init_promptsV2PromptTopicsDelete2 = __esm(() => {
   init_promptsV2PromptTopicsDelete();
   init_v2prompttopicsdeleteop();
   init_tools();
-  args30 = {
+  args28 = {
     request: V2PromptTopicsDeleteRequest$zodSchema
   };
   tool$promptsV2PromptTopicsDelete = {
@@ -59290,7 +59286,7 @@ Archives a prompt topic. Set \`archive_prompts\` to archive remaining prompts; o
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args30,
+    args: args28,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptTopicsDelete(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -59306,9 +59302,9 @@ Archives a prompt topic. Set \`archive_prompts\` to archive remaining prompts; o
 
 // src/funcs/promptsV2PromptTopicsList.ts
 function promptsV2PromptTopicsList(client$, options) {
-  return new APIPromise($do36(client$, options));
+  return new APIPromise($do34(client$, options));
 }
-async function $do36(client$, options) {
+async function $do34(client$, options) {
   const path$ = pathToFunc("/v2/prompts/topics")();
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
@@ -59424,9 +59420,9 @@ var init_v2prompttopicsupdateop = __esm(() => {
 
 // src/funcs/promptsV2PromptTopicsUpdate.ts
 function promptsV2PromptTopicsUpdate(client$, request, options) {
-  return new APIPromise($do37(client$, request, options));
+  return new APIPromise($do35(client$, request, options));
 }
-async function $do37(client$, request, options) {
+async function $do35(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2PromptTopicsUpdateRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -59502,12 +59498,12 @@ var init_promptsV2PromptTopicsUpdate = __esm(() => {
 });
 
 // src/mcp-server/tools/promptsV2PromptTopicsUpdate.ts
-var args31, tool$promptsV2PromptTopicsUpdate;
+var args29, tool$promptsV2PromptTopicsUpdate;
 var init_promptsV2PromptTopicsUpdate2 = __esm(() => {
   init_promptsV2PromptTopicsUpdate();
   init_v2prompttopicsupdateop();
   init_tools();
-  args31 = {
+  args29 = {
     request: V2PromptTopicsUpdateRequest$zodSchema
   };
   tool$promptsV2PromptTopicsUpdate = {
@@ -59522,7 +59518,7 @@ Renames a prompt topic and returns the updated topic.`,
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args31,
+    args: args29,
     tool: async (client, args, ctx) => {
       const [result] = await promptsV2PromptTopicsUpdate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -59565,9 +59561,9 @@ var init_createv2repositoryrequest = __esm(() => {
 
 // src/funcs/repositoriesV2RepositoriesCreate.ts
 function repositoriesV2RepositoriesCreate(client$, request, options) {
-  return new APIPromise($do38(client$, request, options));
+  return new APIPromise($do36(client$, request, options));
 }
-async function $do38(client$, request, options) {
+async function $do36(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => CreateV2RepositoryRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -59637,12 +59633,12 @@ var init_repositoriesV2RepositoriesCreate = __esm(() => {
 });
 
 // src/mcp-server/tools/repositoriesV2RepositoriesCreate.ts
-var args32, tool$repositoriesV2RepositoriesCreate;
+var args30, tool$repositoriesV2RepositoriesCreate;
 var init_repositoriesV2RepositoriesCreate2 = __esm(() => {
   init_repositoriesV2RepositoriesCreate();
   init_createv2repositoryrequest();
   init_tools();
-  args32 = {
+  args30 = {
     request: CreateV2RepositoryRequest$zodSchema
   };
   tool$repositoriesV2RepositoriesCreate = {
@@ -59657,7 +59653,7 @@ Adds a reusable Git repository workspace for eval configurations.`,
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args32,
+    args: args30,
     tool: async (client, args, ctx) => {
       const [result] = await repositoriesV2RepositoriesCreate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -59723,9 +59719,9 @@ var init_v2repositoriesdeleteop = __esm(() => {
 
 // src/funcs/repositoriesV2RepositoriesDelete.ts
 function repositoriesV2RepositoriesDelete(client$, request, options) {
-  return new APIPromise($do39(client$, request, options));
+  return new APIPromise($do37(client$, request, options));
 }
-async function $do39(client$, request, options) {
+async function $do37(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2RepositoriesDeleteRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -59800,12 +59796,12 @@ var init_repositoriesV2RepositoriesDelete = __esm(() => {
 });
 
 // src/mcp-server/tools/repositoriesV2RepositoriesDelete.ts
-var args33, tool$repositoriesV2RepositoriesDelete;
+var args31, tool$repositoriesV2RepositoriesDelete;
 var init_repositoriesV2RepositoriesDelete2 = __esm(() => {
   init_repositoriesV2RepositoriesDelete();
   init_v2repositoriesdeleteop();
   init_tools();
-  args33 = {
+  args31 = {
     request: V2RepositoriesDeleteRequest$zodSchema
   };
   tool$repositoriesV2RepositoriesDelete = {
@@ -59820,7 +59816,7 @@ Permanently deletes a repository workspace and returns its last saved state.`,
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args33,
+    args: args31,
     tool: async (client, args, ctx) => {
       const [result] = await repositoriesV2RepositoriesDelete(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -59836,9 +59832,9 @@ Permanently deletes a repository workspace and returns its last saved state.`,
 
 // src/funcs/repositoriesV2RepositoriesList.ts
 function repositoriesV2RepositoriesList(client$, options) {
-  return new APIPromise($do40(client$, options));
+  return new APIPromise($do38(client$, options));
 }
-async function $do40(client$, options) {
+async function $do38(client$, options) {
   const path$ = pathToFunc("/v2/repositories")();
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
@@ -59943,9 +59939,9 @@ var init_v2repositoriesretrieveop = __esm(() => {
 
 // src/funcs/repositoriesV2RepositoriesRetrieve.ts
 function repositoriesV2RepositoriesRetrieve(client$, request, options) {
-  return new APIPromise($do41(client$, request, options));
+  return new APIPromise($do39(client$, request, options));
 }
-async function $do41(client$, request, options) {
+async function $do39(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2RepositoriesRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -60020,12 +60016,12 @@ var init_repositoriesV2RepositoriesRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/repositoriesV2RepositoriesRetrieve.ts
-var args34, tool$repositoriesV2RepositoriesRetrieve;
+var args32, tool$repositoriesV2RepositoriesRetrieve;
 var init_repositoriesV2RepositoriesRetrieve2 = __esm(() => {
   init_repositoriesV2RepositoriesRetrieve();
   init_v2repositoriesretrieveop();
   init_tools();
-  args34 = {
+  args32 = {
     request: V2RepositoriesRetrieveRequest$zodSchema
   };
   tool$repositoriesV2RepositoriesRetrieve = {
@@ -60040,7 +60036,7 @@ Returns one reusable repository workspace.`,
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args34,
+    args: args32,
     tool: async (client, args, ctx) => {
       const [result] = await repositoriesV2RepositoriesRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -60100,9 +60096,9 @@ var init_v2repositoriesupdateop = __esm(() => {
 
 // src/funcs/repositoriesV2RepositoriesUpdate.ts
 function repositoriesV2RepositoriesUpdate(client$, request, options) {
-  return new APIPromise($do42(client$, request, options));
+  return new APIPromise($do40(client$, request, options));
 }
-async function $do42(client$, request, options) {
+async function $do40(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2RepositoriesUpdateRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -60178,12 +60174,12 @@ var init_repositoriesV2RepositoriesUpdate = __esm(() => {
 });
 
 // src/mcp-server/tools/repositoriesV2RepositoriesUpdate.ts
-var args35, tool$repositoriesV2RepositoriesUpdate;
+var args33, tool$repositoriesV2RepositoriesUpdate;
 var init_repositoriesV2RepositoriesUpdate2 = __esm(() => {
   init_repositoriesV2RepositoriesUpdate();
   init_v2repositoriesupdateop();
   init_tools();
-  args35 = {
+  args33 = {
     request: V2RepositoriesUpdateRequest$zodSchema
   };
   tool$repositoriesV2RepositoriesUpdate = {
@@ -60198,7 +60194,7 @@ Updates a repository workspace; omitted fields remain unchanged.`,
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args35,
+    args: args33,
     tool: async (client, args, ctx) => {
       const [result] = await repositoriesV2RepositoriesUpdate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -60214,9 +60210,9 @@ Updates a repository workspace; omitted fields remain unchanged.`,
 
 // src/funcs/siteAuditsV2SiteAuditsList.ts
 function siteAuditsV2SiteAuditsList(client$, options) {
-  return new APIPromise($do43(client$, options));
+  return new APIPromise($do41(client$, options));
 }
-async function $do43(client$, options) {
+async function $do41(client$, options) {
   const path$ = pathToFunc("/v2/site-audits")();
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
@@ -60357,9 +60353,9 @@ var init_v2siteauditsretrieveop = __esm(() => {
 
 // src/funcs/siteAuditsV2SiteAuditsRetrieve.ts
 function siteAuditsV2SiteAuditsRetrieve(client$, request, options) {
-  return new APIPromise($do44(client$, request, options));
+  return new APIPromise($do42(client$, request, options));
 }
-async function $do44(client$, request, options) {
+async function $do42(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2SiteAuditsRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -60434,12 +60430,12 @@ var init_siteAuditsV2SiteAuditsRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/siteAuditsV2SiteAuditsRetrieve.ts
-var args36, tool$siteAuditsV2SiteAuditsRetrieve;
+var args34, tool$siteAuditsV2SiteAuditsRetrieve;
 var init_siteAuditsV2SiteAuditsRetrieve2 = __esm(() => {
   init_siteAuditsV2SiteAuditsRetrieve();
   init_v2siteauditsretrieveop();
   init_tools();
-  args36 = {
+  args34 = {
     request: V2SiteAuditsRetrieveRequest$zodSchema
   };
   tool$siteAuditsV2SiteAuditsRetrieve = {
@@ -60454,7 +60450,7 @@ Returns one audited page with its score, priority, supporting evidence, and reco
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args36,
+    args: args34,
     tool: async (client, args, ctx) => {
       const [result] = await siteAuditsV2SiteAuditsRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -60470,9 +60466,9 @@ Returns one audited page with its score, priority, supporting evidence, and reco
 
 // src/funcs/skillsV2SkillsList.ts
 function skillsV2SkillsList(client$, options) {
-  return new APIPromise($do45(client$, options));
+  return new APIPromise($do43(client$, options));
 }
-async function $do45(client$, options) {
+async function $do43(client$, options) {
   const path$ = pathToFunc("/v2/skills")();
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
@@ -60562,9 +60558,9 @@ Returns reusable skills available to workflows and eval configurations.`,
 
 // src/funcs/statusV2AuthStatusGet.ts
 function statusV2AuthStatusGet(client$, options) {
-  return new APIPromise($do46(client$, options));
+  return new APIPromise($do44(client$, options));
 }
-async function $do46(client$, options) {
+async function $do44(client$, options) {
   const path$ = pathToFunc("/v2/auth/status")();
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
@@ -60654,9 +60650,9 @@ Validates the API key and returns its organization ID, API key ID, scopes, and a
 
 // src/funcs/statusV2StatusGet.ts
 function statusV2StatusGet(client$, options) {
-  return new APIPromise($do47(client$, options));
+  return new APIPromise($do45(client$, options));
 }
-async function $do47(client$, options) {
+async function $do45(client$, options) {
   const path$ = pathToFunc("/v2/status")();
   const headers$ = new Headers(compactMap({
     Accept: "application/json"
@@ -60766,11 +60762,11 @@ var init_v2publictask = __esm(() => {
     priority: string2().nullable().optional(),
     rank: int().describe("Sort order within the batch, ascending; values need not start at 1."),
     reasoning: string2().nullable().optional(),
-    score: number2(),
+    score: number2().nullable().optional().describe("Planner impact score; manually created tasks may be unscored."),
     source: string2().nullable().optional(),
     source_key: string2().describe("A finding key namespaced by source, such as `api_perf:https://example.com/docs/cli`."),
     source_refs: record(string2(), any()).optional().describe("Source references needed to understand or execute the task."),
-    source_type: string2().describe("A free-form source type such as `api_performance`; this field is not an enum."),
+    source_type: string2().nullable().optional().describe("A free-form source type such as `api_performance`; manually created tasks may not have a source."),
     status: string2().describe("The workflow status. New tasks start as `new`; `added_to_sheet` and `snoozed` tasks are listed only when `status=all`."),
     task_type: string2().describe("A free-form label such as `Redirect` or `Docs`; this field is not an enum."),
     title: string2(),
@@ -60795,11 +60791,11 @@ var init_v2publictask = __esm(() => {
     priority: nullable(string2()).optional(),
     rank: int(),
     reasoning: nullable(string2()).optional(),
-    score: number2(),
+    score: nullable(number2()).optional(),
     source: nullable(string2()).optional(),
     source_key: string2(),
     source_refs: record(string2(), any()).optional(),
-    source_type: string2(),
+    source_type: nullable(string2()).optional(),
     status: string2(),
     task_type: string2(),
     title: string2(),
@@ -60876,9 +60872,9 @@ var init_v2taskslistop = __esm(() => {
 
 // src/funcs/tasksV2TasksList.ts
 function tasksV2TasksList(client$, request, options) {
-  return new APIPromise($do48(client$, request, options));
+  return new APIPromise($do46(client$, request, options));
 }
-async function $do48(client$, request, options) {
+async function $do46(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2TasksListRequest$zodSchema.optional().parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -60952,12 +60948,12 @@ var init_tasksV2TasksList = __esm(() => {
 });
 
 // src/mcp-server/tools/tasksV2TasksList.ts
-var args37, tool$tasksV2TasksList;
+var args35, tool$tasksV2TasksList;
 var init_tasksV2TasksList2 = __esm(() => {
   init_tasksV2TasksList();
   init_v2taskslistop();
   init_tools();
-  args37 = {
+  args35 = {
     request: V2TasksListRequest$zodSchema.optional()
   };
   tool$tasksV2TasksList = {
@@ -60972,7 +60968,7 @@ Returns persisted daily tasks with execution instructions. Includes per-status c
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args37,
+    args: args35,
     tool: async (client, args, ctx) => {
       const [result] = await tasksV2TasksList(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -61003,9 +60999,9 @@ var init_v2tasksretrieveop = __esm(() => {
 
 // src/funcs/tasksV2TasksRetrieve.ts
 function tasksV2TasksRetrieve(client$, request, options) {
-  return new APIPromise($do49(client$, request, options));
+  return new APIPromise($do47(client$, request, options));
 }
-async function $do49(client$, request, options) {
+async function $do47(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2TasksRetrieveRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -61080,12 +61076,12 @@ var init_tasksV2TasksRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/tasksV2TasksRetrieve.ts
-var args38, tool$tasksV2TasksRetrieve;
+var args36, tool$tasksV2TasksRetrieve;
 var init_tasksV2TasksRetrieve2 = __esm(() => {
   init_tasksV2TasksRetrieve();
   init_v2tasksretrieveop();
   init_tools();
-  args38 = {
+  args36 = {
     request: V2TasksRetrieveRequest$zodSchema
   };
   tool$tasksV2TasksRetrieve = {
@@ -61100,7 +61096,7 @@ Returns one persisted task and its execution prompt.`,
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args38,
+    args: args36,
     tool: async (client, args, ctx) => {
       const [result] = await tasksV2TasksRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -61163,9 +61159,9 @@ var init_v2tasksupdateop = __esm(() => {
 
 // src/funcs/tasksV2TasksUpdate.ts
 function tasksV2TasksUpdate(client$, request, options) {
-  return new APIPromise($do50(client$, request, options));
+  return new APIPromise($do48(client$, request, options));
 }
-async function $do50(client$, request, options) {
+async function $do48(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2TasksUpdateRequest$zodSchema.parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -61241,12 +61237,12 @@ var init_tasksV2TasksUpdate = __esm(() => {
 });
 
 // src/mcp-server/tools/tasksV2TasksUpdate.ts
-var args39, tool$tasksV2TasksUpdate;
+var args37, tool$tasksV2TasksUpdate;
 var init_tasksV2TasksUpdate2 = __esm(() => {
   init_tasksV2TasksUpdate();
   init_v2tasksupdateop();
   init_tools();
-  args39 = {
+  args37 = {
     request: V2TasksUpdateRequest$zodSchema
   };
   tool$tasksV2TasksUpdate = {
@@ -61261,7 +61257,7 @@ Updates a task's status, priority, assignee, or action type. Only sent fields ar
       openWorldHint: false,
       readOnlyHint: false
     },
-    args: args39,
+    args: args37,
     tool: async (client, args, ctx) => {
       const [result] = await tasksV2TasksUpdate(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -61349,9 +61345,9 @@ var init_v2visibilitycitationslistop = __esm(() => {
 
 // src/funcs/visibilityV2VisibilityCitationsList.ts
 function visibilityV2VisibilityCitationsList(client$, request, options) {
-  return new APIPromise($do51(client$, request, options));
+  return new APIPromise($do49(client$, request, options));
 }
-async function $do51(client$, request, options) {
+async function $do49(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2VisibilityCitationsListRequest$zodSchema.optional().parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -61428,12 +61424,12 @@ var init_visibilityV2VisibilityCitationsList = __esm(() => {
 });
 
 // src/mcp-server/tools/visibilityV2VisibilityCitationsList.ts
-var args40, tool$visibilityV2VisibilityCitationsList;
+var args38, tool$visibilityV2VisibilityCitationsList;
 var init_visibilityV2VisibilityCitationsList2 = __esm(() => {
   init_visibilityV2VisibilityCitationsList();
   init_v2visibilitycitationslistop();
   init_tools();
-  args40 = {
+  args38 = {
     request: V2VisibilityCitationsListRequest$zodSchema.optional()
   };
   tool$visibilityV2VisibilityCitationsList = {
@@ -61448,7 +61444,7 @@ Returns URL-level citation counts ranked from most to least cited. Results can b
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args40,
+    args: args38,
     tool: async (client, args, ctx) => {
       const [result] = await visibilityV2VisibilityCitationsList(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -61487,10 +61483,10 @@ var init_publiccurrentvisibilityresponse = __esm(() => {
     date: string2().nullable().optional().describe("Latest prompt-run date in the range."),
     end_date: string2(),
     has_data: boolean2(),
-    share_of_voice: number2().nullable().optional().describe("Share of all brand mentions across tracked companies, 0 to 1."),
+    share_of_voice: number2().nullable().optional().describe("Fraction of successful prompt runs that mention the company, from 0 to 1; for example, 0.25 means one in four runs. Failed provider runs are excluded."),
     share_of_voice_rank: int().nullable().optional(),
     start_date: string2(),
-    visibility_score: number2().nullable().optional().describe("The visibility score, from 0 to 1."),
+    visibility_score: number2().nullable().optional().describe("Mean recommendation strength across successful prompt runs, from 0 to 1. Each run contributes the company's recommendation strength (0 to 1) when it is mentioned and 0 when it is not. Failed provider runs are excluded."),
     visibility_score_rank: int().nullable().optional()
   });
 });
@@ -61513,9 +61509,9 @@ var init_v2visibilitycurrentretrieveop = __esm(() => {
 
 // src/funcs/visibilityV2VisibilityCurrentRetrieve.ts
 function visibilityV2VisibilityCurrentRetrieve(client$, request, options) {
-  return new APIPromise($do52(client$, request, options));
+  return new APIPromise($do50(client$, request, options));
 }
-async function $do52(client$, request, options) {
+async function $do50(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2VisibilityCurrentRetrieveRequest$zodSchema.optional().parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -61589,12 +61585,12 @@ var init_visibilityV2VisibilityCurrentRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/visibilityV2VisibilityCurrentRetrieve.ts
-var args41, tool$visibilityV2VisibilityCurrentRetrieve;
+var args39, tool$visibilityV2VisibilityCurrentRetrieve;
 var init_visibilityV2VisibilityCurrentRetrieve2 = __esm(() => {
   init_visibilityV2VisibilityCurrentRetrieve();
   init_v2visibilitycurrentretrieveop();
   init_tools();
-  args41 = {
+  args39 = {
     request: V2VisibilityCurrentRetrieveRequest$zodSchema.optional()
   };
   tool$visibilityV2VisibilityCurrentRetrieve = {
@@ -61609,7 +61605,7 @@ Returns visibility score, share of voice, ranks, and date-range metadata for the
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args41,
+    args: args39,
     tool: async (client, args, ctx) => {
       const [result] = await visibilityV2VisibilityCurrentRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -61632,9 +61628,9 @@ var init_publiccompetitorrow = __esm(() => {
     average_position: number2().nullable().optional().describe("Mean mention position in AI responses; 1 is first and lower is better. 0 when never mentioned."),
     average_position_rank: int().nullable().optional().describe("Companies with an `average_position` of 0 rank last."),
     company: PublicCompany$zodSchema,
-    share_of_voice: number2().nullable().optional().describe("Share of all brand mentions across tracked companies, 0 to 1; rows sum to about 1."),
+    share_of_voice: number2().nullable().optional().describe("Fraction of successful prompt runs that mention the company, from 0 to 1; for example, 0.25 means one in four runs. Failed provider runs are excluded. A run can mention several companies, so rows do not sum to 1."),
     share_of_voice_rank: int().nullable().optional(),
-    visibility_score: number2().nullable().optional().describe("The company's visibility score, from 0 to 1. Rows do not sum to 1."),
+    visibility_score: number2().nullable().optional().describe("Mean recommendation strength across successful prompt runs, from 0 to 1. Each run contributes the company's recommendation strength (0 to 1) when it is mentioned and 0 when it is not. Failed provider runs are excluded. Rows do not sum to 1."),
     visibility_score_rank: int().nullable().optional()
   });
 });
@@ -61695,10 +61691,10 @@ var init_publicvisibilitysnapshot = __esm(() => {
   PublicVisibilitySnapshot$zodSchema = object({
     average_position: number2().nullable().optional().describe("Mean mention position in AI responses; 1 is first and lower is better."),
     average_position_rank: int().nullable().optional(),
-    share_of_voice: number2().nullable().optional().describe("Share of all brand mentions across tracked companies, 0 to 1."),
+    share_of_voice: number2().nullable().optional().describe("Fraction of successful prompt runs that mention the company, from 0 to 1; for example, 0.25 means one in four runs. Failed provider runs are excluded."),
     share_of_voice_rank: int().nullable().optional(),
     total_citations: int().default(0).describe("Owned-domain citations across all platforms; 0, not `null`, with no prompt runs."),
-    visibility_score: number2().nullable().optional().describe("The visibility score, from 0 to 1."),
+    visibility_score: number2().nullable().optional().describe("Mean recommendation strength across successful prompt runs, from 0 to 1. Each run contributes the company's recommendation strength (0 to 1) when it is mentioned and 0 when it is not. Failed provider runs are excluded."),
     visibility_score_rank: int().nullable().optional()
   });
 });
@@ -61752,9 +61748,9 @@ var init_v2visibilityoverviewretrieveop = __esm(() => {
 
 // src/funcs/visibilityV2VisibilityOverviewRetrieve.ts
 function visibilityV2VisibilityOverviewRetrieve(client$, request, options) {
-  return new APIPromise($do53(client$, request, options));
+  return new APIPromise($do51(client$, request, options));
 }
-async function $do53(client$, request, options) {
+async function $do51(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2VisibilityOverviewRetrieveRequest$zodSchema.optional().parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -61828,12 +61824,12 @@ var init_visibilityV2VisibilityOverviewRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/visibilityV2VisibilityOverviewRetrieve.ts
-var args42, tool$visibilityV2VisibilityOverviewRetrieve;
+var args40, tool$visibilityV2VisibilityOverviewRetrieve;
 var init_visibilityV2VisibilityOverviewRetrieve2 = __esm(() => {
   init_visibilityV2VisibilityOverviewRetrieve();
   init_v2visibilityoverviewretrieveop();
   init_tools();
-  args42 = {
+  args40 = {
     request: V2VisibilityOverviewRetrieveRequest$zodSchema.optional()
   };
   tool$visibilityV2VisibilityOverviewRetrieve = {
@@ -61848,7 +61844,7 @@ Returns aggregate snapshot, daily trends, per-platform breakdowns, and competito
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args42,
+    args: args40,
     tool: async (client, args, ctx) => {
       const [result] = await visibilityV2VisibilityOverviewRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -61882,8 +61878,8 @@ var init_publicpromptvisibilitymetrics = __esm(() => {
     average_position: number2().nullable().optional().describe("Mean mention position in AI responses; 1 is first and lower is better. 0 when runs produced no mentions."),
     citation_count: int().default(0).describe("Owned-domain citations only."),
     run_count: int().default(0).describe("One per prompt, platform, and run date."),
-    share_of_voice: number2().nullable().optional().describe("Share of all brand mentions across tracked companies, 0 to 1."),
-    visibility_score: number2().nullable().optional().describe("The visibility score, from 0 to 1."),
+    share_of_voice: number2().nullable().optional().describe("Fraction of successful prompt runs that mention the company, from 0 to 1; for example, 0.25 means one in four runs. Failed provider runs are excluded."),
+    visibility_score: number2().nullable().optional().describe("Mean recommendation strength across successful prompt runs, from 0 to 1. Each run contributes the company's recommendation strength (0 to 1) when it is mentioned and 0 when it is not. Failed provider runs are excluded."),
     visibility_score_rank: int().nullable().optional()
   });
 });
@@ -61962,9 +61958,9 @@ var init_v2visibilitypromptsretrieveop = __esm(() => {
 
 // src/funcs/visibilityV2VisibilityPromptsRetrieve.ts
 function visibilityV2VisibilityPromptsRetrieve(client$, request, options) {
-  return new APIPromise($do54(client$, request, options));
+  return new APIPromise($do52(client$, request, options));
 }
-async function $do54(client$, request, options) {
+async function $do52(client$, request, options) {
   const parsed$ = safeParse4(request, (value$) => V2VisibilityPromptsRetrieveRequest$zodSchema.optional().parse(value$), "Input validation failed");
   if (!parsed$.ok) {
     return [parsed$, { status: "invalid" }];
@@ -62038,12 +62034,12 @@ var init_visibilityV2VisibilityPromptsRetrieve = __esm(() => {
 });
 
 // src/mcp-server/tools/visibilityV2VisibilityPromptsRetrieve.ts
-var args43, tool$visibilityV2VisibilityPromptsRetrieve;
+var args41, tool$visibilityV2VisibilityPromptsRetrieve;
 var init_visibilityV2VisibilityPromptsRetrieve2 = __esm(() => {
   init_visibilityV2VisibilityPromptsRetrieve();
   init_v2visibilitypromptsretrieveop();
   init_tools();
-  args43 = {
+  args41 = {
     request: V2VisibilityPromptsRetrieveRequest$zodSchema.optional()
   };
   tool$visibilityV2VisibilityPromptsRetrieve = {
@@ -62058,7 +62054,7 @@ Returns per-topic and per-prompt visibility metrics including score, rank, share
       openWorldHint: false,
       readOnlyHint: true
     },
-    args: args43,
+    args: args41,
     tool: async (client, args, ctx) => {
       const [result] = await visibilityV2VisibilityPromptsRetrieve(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
       if (!result.ok) {
@@ -62082,7 +62078,7 @@ function createMCPServer(deps) {
       sizes: ["any"]
     }],
     name: "LightsageMcp",
-    version: "0.3.0"
+    version: "0.4.0"
   });
   const getClient = deps.getSDK || (() => new LightsageMcpCore({
     security: deps.security,
@@ -62139,12 +62135,10 @@ function createMCPServer(deps) {
   tool(tool$evalsV2EvalConfigurationsUpdate);
   tool(tool$evalsV2EvalConfigurationsDelete);
   tool(tool$evalsRun);
-  tool(tool$evalsAnalysis);
+  tool(tool$evalsCancel);
   tool(tool$evalsTrace);
+  tool(tool$evalsGet);
   tool(tool$evalRunsV2EvalRunsList);
-  tool(tool$evalRunsV2EvalRunsRetrieve);
-  tool(tool$evalResultsV2EvalResultsList);
-  tool(tool$evalResultsV2EvalResultsRetrieve);
   tool(tool$agentsV2AgentsList);
   tool(tool$repositoriesV2RepositoriesList);
   tool(tool$repositoriesV2RepositoriesCreate);
@@ -62171,11 +62165,9 @@ var init_server2 = __esm(() => {
   init_contentV2ContentFinalize2();
   init_contentV2ContentRetrieve2();
   init_environmentsV2EnvironmentsList2();
-  init_evalResultsV2EvalResultsList2();
-  init_evalResultsV2EvalResultsRetrieve2();
   init_evalRunsV2EvalRunsList2();
-  init_evalRunsV2EvalRunsRetrieve2();
-  init_evalsAnalysis2();
+  init_evalsCancel2();
+  init_evalsGet2();
   init_evalsRun2();
   init_evalsTrace2();
   init_evalsV2EvalConfigurationsCreate2();
@@ -62293,13 +62285,13 @@ Renames a prompt topic and returns the updated topic.`
       name: "prompts-v2-prompt-runs-list",
       description: `List Prompt Runs
 
-Returns stored prompt runs without exposing raw runtime or provider configuration. Filterable by prompt, platform, and date range.`
+Returns stored prompt runs with limit-and-offset pagination, without exposing raw runtime or provider configuration. Filter by prompt, platform, or date range.`
     },
     {
       name: "prompts-v2-prompt-runs-retrieve",
       description: `Retrieve Prompt Run
 
-Returns one stored prompt run, including its mentions and citations.`
+Returns one stored prompt run, including mentions, citations, and any sanitized coding-agent trace captured for the run.`
     },
     {
       name: "prompts-v2-prompts-retrieve",
@@ -62379,21 +62371,21 @@ Returns one audited page with its score, priority, supporting evidence, and reco
       name: "evals-v2-evals-list",
       description: `List Evals
 
-Returns all active evals including workflow, MCP, CLI, and skill evals owned by the organization.`
+Returns all active eval definitions owned by the organization.`
     },
     {
       name: "evals-v2-evals-create",
-      description: "Create Eval\n\nCreates an eval with a task prompt, subject, and plain-text success criterion in `judge`. The eval can then be run through eval runs."
+      description: "Create Eval\n\nCreates a reusable simulation definition with one or more task prompts and judges. Optional agents, persona, CLI installations, MCP servers, setup commands, cleanup commands, grounding policy, repeat count, and tags are saved with the eval. `search_grounding` supports `disabled`, `enabled`, and `default`; `default` preserves normal platform behavior. Creating the eval does not start a run and does not accept secret environment values."
     },
     {
       name: "evals-v2-evals-retrieve",
       description: `Retrieve Eval
 
-Returns one eval and its judge configuration.`
+Returns one reusable eval simulation definition.`
     },
     {
       name: "evals-v2-evals-update",
-      description: "Update Eval\n\nUpdates an eval's task prompt, subject, plain-text `judge`, or name. Omitted fields remain unchanged."
+      description: "Update Eval\n\nUpdates a reusable eval simulation definition. Omitted fields remain unchanged. Empty `agent`, `clis`, `mcps`, `setup_commands`, `cleanup_commands`, or `tags` arrays clear those collections, and `persona: null` clears the saved persona."
     },
     {
       name: "evals-v2-evals-delete",
@@ -62433,39 +62425,27 @@ Permanently deletes an eval configuration and returns its last saved state.`
     },
     {
       name: "evals-run",
-      description: "Start an eval run\n\nQueues a durable eval job and returns immediately with a polling ID. Poll `GET /v2/eval-runs/{run_id}` until its status reaches a terminal value.\n\nSend one of these flat request forms:\n\n- **Saved eval:** provide `eval_id`, `configuration_id`, and `agent_id`.\n- **Direct create-and-run:** provide `prompt`, `judges`, and `agent_id`. You may also provide a repository, write-only environment variables, and public skill packages.\n\nDo not combine saved-run and direct-run fields. No separate `mode` field is required."
+      description: "Start an eval run\n\nQueues a durable eval job and returns immediately with a polling ID. Poll `GET /v2/eval-runs/{run_id}` until its status reaches a terminal value; once finished it also returns the results and analysis. Follow a run live with the monitor operation.\n\nSend one of these flat request forms:\n\n- **Saved eval:** provide `eval_id`; optionally replace its prompts, persona, agents, CLIs, MCPs, setup commands, cleanup commands, or runs for this launch. A saved configuration can still provide repository, environment, skills, and legacy tool selections.\n- **Direct create-and-run:** provide `prompts`, `judges`, and `agent`. You may also provide a persona, tags, repository, write-only environment variables, saved or inline CLIs and MCP servers, skills, setup commands, and cleanup commands.\n\nThe run fans out across prompts × agents × runs. Persona is shared metadata and does not multiply attempts. No separate `mode` field is required. If the organization lacks the estimated credits, the request fails with `402` before dispatching work.\n\nA saved or inline environment containing an empty or placeholder credential is rejected with `400` before any evaluation is dispatched."
     },
     {
-      name: "evals-analysis",
-      description: `Retrieve eval run analysis
+      name: "evals-cancel",
+      description: `Cancel an eval run
 
-Returns the persisted analysis produced for each completed attempt in one eval run. The analysis includes the verdict, outcome explanation, findings, bottleneck, and analysis sections. Active runs return the analysis available so far together with current run status and progress.`
+Marks an active eval run as cancelled and prevents queued attempts from starting. Running sandbox attempts are stopped asynchronously. Requests already sent to an external provider cannot be recalled, but any later response does not change the run from cancelled. Repeating this request for an already-cancelled run is safe.`
     },
     {
       name: "evals-trace",
       description: "Retrieve an eval run trace\n\nReturns a finite page of sanitized execution events for one eval run. Use `next_after_id` as `after_id` to retrieve the next page. Use the monitor operation instead when following a run live."
     },
     {
-      name: "eval-runs-v2-eval-runs-list",
-      description: "List Runs\n\nReturns jobs created through `POST /v2/eval-runs`, ordered newest first, with lifecycle state and progress."
-    },
-    {
-      name: "eval-runs-v2-eval-runs-retrieve",
-      description: `Retrieve Run
-
-Returns the status and progress of an eval job. Poll until the status reaches a terminal value.`
-    },
-    {
-      name: "eval-results-v2-eval-results-list",
-      description: `List Results
+      name: "evals-get",
+      description: `Retrieve an eval run
 
-Returns completed custom-eval results from all supported run flows, ordered newest first. Result IDs are different from eval run IDs.`
+Returns the run's current status and progress. Once its attempts finish, the response also includes their results and analysis. Poll this until the status reaches a terminal value. Use the trace operation for the raw execution event log, or the monitor operation to follow a run live.`
     },
     {
-      name: "eval-results-v2-eval-results-retrieve",
-      description: `Retrieve Result
-
-Returns one completed custom-eval result with its output and diagnostic details.`
+      name: "eval-runs-v2-eval-runs-list",
+      description: "List Runs\n\nReturns jobs created through `POST /v2/eval-runs`, ordered newest first, with lifecycle state and progress."
     },
     {
       name: "agents-v2-agents-list",
@@ -63431,7 +63411,7 @@ http_headers = { "lightsage-api-key-auth" = "YOUR_LIGHTSAGE_API_KEY_AUTH" }`;
         <h1>Instructions</h1>
         <p>One-click installation for Claude Desktop users</p>
         <div class="instruction-item">
-          <a href="https://github.com/lightsagehq/mcp/releases/download/v0.3.0/mcp-server.mcpb" download="mcp-server.mcpb" class="action-button header-action" style="display: inline-flex; margin-bottom: 16px;">
+          <a href="https://github.com/lightsagehq/mcp/releases/download/v0.4.0/mcp-server.mcpb" download="mcp-server.mcpb" class="action-button header-action" style="display: inline-flex; margin-bottom: 16px;">
             \uD83D\uDCE5 Download MCP Bundle
           </a>
         </div>
@@ -66309,7 +66289,7 @@ var routes = buildRouteMap({
 var app = buildApplication(routes, {
   name: "mcp",
   versionInfo: {
-    currentVersion: "0.3.0"
+    currentVersion: "0.4.0"
   }
 });
 run(app, process3.argv.slice(2), buildContext(process3));
@@ -66317,5 +66297,5 @@ export {
   app
 };
 
-//# debugId=E8194E142237D9EF64756E2164756E21
+//# debugId=0CC0D115366CE4FD64756E2164756E21
 //# sourceMappingURL=mcp-server.js.map