@agentv/core 4.25.2-next.1 → 4.25.4-next.1

package/dist/index.cjs

@@ -5351,7 +5351,6 @@ var init_code_grader = __esm({
           const rawParsed = parseJsonSafe(stdout);
           const parsed = rawParsed != null && typeof rawParsed === "object" && !Array.isArray(rawParsed) ? rawParsed : void 0;
           const passed = exitCode === 0;
-          const score = parsed != null ? clampScore(typeof parsed.score === "number" ? parsed.score : 0) : passed ? 1 : 0;
           const assertions = parsed != null && Array.isArray(parsed?.assertions) ? parsed.assertions.filter(
             (a) => typeof a === "object" && a !== null && typeof a.text === "string"
           ).map((a) => ({
@@ -5359,6 +5358,9 @@ var init_code_grader = __esm({
             passed: Boolean(a.passed),
             ...typeof a.evidence === "string" ? { evidence: a.evidence } : {}
           })) : parsed == null ? [{ text: stdout.trim() || (passed ? "exit 0" : `exit ${exitCode}`), passed }] : [];
+          const score = parsed != null ? clampScore(
+            typeof parsed.score === "number" ? parsed.score : assertions.length > 0 ? assertions.filter((a) => a.passed).length / assertions.length : 0
+          ) : passed ? 1 : 0;
           const details = parsed?.details && typeof parsed.details === "object" && !Array.isArray(parsed.details) ? parsed.details : void 0;
           const proxyUsage = getProxyUsage?.();
           const graderRawRequest = {
@@ -8712,11 +8714,12 @@ async function invokePiAi(options) {
   const aggregateUsage = { input: 0, output: 0, cacheRead: 0, cost: 0 };
   let stepCount = 0;
   let toolCallCount = 0;
-  let result = await withRetry(
-    () => (0, import_pi_ai.complete)(model, ctx, callOptions),
-    retryConfig,
-    request.signal
-  );
+  const callPi = async () => {
+    const r = await (0, import_pi_ai.complete)(model, ctx, callOptions);
+    if (r.stopReason === "error") throw piErrorFromResult(r);
+    return r;
+  };
+  let result = await withRetry(callPi, retryConfig, request.signal);
   ctx.messages.push(result);
   stepCount = 1;
   accumulateUsage(aggregateUsage, result.usage);
@@ -8751,11 +8754,7 @@ async function invokePiAi(options) {
         timestamp: Date.now()
       });
     }
-    result = await withRetry(
-      () => (0, import_pi_ai.complete)(model, ctx, callOptions),
-      retryConfig,
-      request.signal
-    );
+    result = await withRetry(callPi, retryConfig, request.signal);
     ctx.messages.push(result);
     stepCount += 1;
     accumulateUsage(aggregateUsage, result.usage);
@@ -9018,6 +9017,18 @@ function calculateRetryDelay(attempt, config) {
 async function sleep(ms) {
   return new Promise((resolve) => setTimeout(resolve, ms));
 }
+function piErrorFromResult(r) {
+  const raw = r.errorMessage ?? "pi-ai call failed with no error message";
+  const statusMatch = raw.match(/^(\d{3})\b\s*(.*)$/);
+  if (statusMatch) {
+    const status = statusMatch[1];
+    const rest = statusMatch[2] || raw;
+    const err = new Error(`pi-ai call failed: HTTP ${status} ${rest}`);
+    err.status = Number.parseInt(status, 10);
+    return err;
+  }
+  return new Error(`pi-ai call failed: ${raw}`);
+}
 async function withRetry(fn, retryConfig, signal) {
   const config = {
     maxRetries: retryConfig?.maxRetries ?? 3,
@@ -14899,17 +14910,16 @@ function findDeprecatedCamelCaseTargetWarnings(target, location) {
   );
   return warnings;
 }
-function normalizeAzureApiVersion(value, apiFormat) {
-  const defaultVersion = apiFormat === "responses" ? DEFAULT_AZURE_RESPONSES_API_VERSION : DEFAULT_AZURE_API_VERSION;
+function normalizeAzureApiVersion(value) {
   if (!value) {
-    return defaultVersion;
+    return DEFAULT_AZURE_API_VERSION;
   }
   const trimmed = value.trim();
   if (trimmed.length === 0) {
-    return defaultVersion;
+    return DEFAULT_AZURE_API_VERSION;
   }
   const withoutPrefix = trimmed.replace(/^api[-_]?version\s*=\s*/i, "").trim();
-  return withoutPrefix.length > 0 ? withoutPrefix : defaultVersion;
+  return withoutPrefix.length > 0 ? withoutPrefix : DEFAULT_AZURE_API_VERSION;
 }
 function resolveRetryConfig(target) {
   const maxRetries = resolveOptionalNumber(target.max_retries, `${target.name} max retries`);
@@ -15166,6 +15176,11 @@ function normalizeOpenAIBaseUrl(value) {
   return trimmed.endsWith("/v1") ? trimmed : `${trimmed}/v1`;
 }
 function resolveAzureConfig(target, env) {
+  if (target.api_format !== void 0) {
+    throw new Error(
+      `The 'api_format' field is no longer supported on Azure targets ('${target.name}'). AgentV always uses Azure's Responses API. If your deployment only exposes /chat/completions, use 'provider: openai' with a deployment-scoped 'base_url' instead. See docs/targets/llm-providers for details.`
+    );
+  }
   const endpointSource = target.endpoint ?? target.resource;
   const apiKeySource = target.api_key;
   const deploymentSource = target.deployment ?? target.model;
@@ -15175,13 +15190,11 @@ function resolveAzureConfig(target, env) {
   const resourceName = resolveString(endpointSource, env, `${target.name} endpoint`);
   const apiKey = resolveString(apiKeySource, env, `${target.name} api key`);
   const deploymentName = resolveString(deploymentSource, env, `${target.name} deployment`);
-  const apiFormat = resolveApiFormat(target, env, target.name);
   const version = normalizeAzureApiVersion(
     resolveOptionalString(versionSource, env, `${target.name} api version`, {
       allowLiteral: true,
       optionalEnv: true
-    }),
-    apiFormat
+    })
   );
   const temperature = resolveOptionalNumber(temperatureSource, `${target.name} temperature`);
   const maxOutputTokens = resolveOptionalNumber(
@@ -15194,7 +15207,6 @@ function resolveAzureConfig(target, env) {
     deploymentName,
     apiKey,
     version,
-    apiFormat,
     temperature,
     maxOutputTokens,
     retry
@@ -16060,7 +16072,7 @@ function resolveOptionalNumberArray(source, description) {
   }
   return resolved.length > 0 ? resolved : void 0;
 }
-var import_node_fs11, import_node_os9, import_node_path26, import_zod4, CliHealthcheckHttpInputSchema, CliHealthcheckCommandInputSchema, CliHealthcheckInputSchema, CliTargetInputSchema, CliHealthcheckHttpSchema, CliHealthcheckCommandSchema, CliHealthcheckSchema, CliTargetConfigSchema, CLI_PLACEHOLDERS, DEPRECATED_TARGET_CAMEL_CASE_FIELDS, DEPRECATED_HEALTHCHECK_CAMEL_CASE_FIELDS, COMMON_TARGET_SETTINGS, USE_TARGET_ENV_PATTERN, BASE_TARGET_SCHEMA, DEFAULT_AZURE_API_VERSION, DEFAULT_AZURE_RESPONSES_API_VERSION, DEFAULT_OPENAI_BASE_URL, cliErrorMap;
+var import_node_fs11, import_node_os9, import_node_path26, import_zod4, CliHealthcheckHttpInputSchema, CliHealthcheckCommandInputSchema, CliHealthcheckInputSchema, CliTargetInputSchema, CliHealthcheckHttpSchema, CliHealthcheckCommandSchema, CliHealthcheckSchema, CliTargetConfigSchema, CLI_PLACEHOLDERS, DEPRECATED_TARGET_CAMEL_CASE_FIELDS, DEPRECATED_HEALTHCHECK_CAMEL_CASE_FIELDS, COMMON_TARGET_SETTINGS, USE_TARGET_ENV_PATTERN, BASE_TARGET_SCHEMA, DEFAULT_AZURE_API_VERSION, DEFAULT_OPENAI_BASE_URL, cliErrorMap;
 var init_targets = __esm({
   "src/evaluation/providers/targets.ts"() {
     "use strict";
@@ -16195,8 +16207,7 @@ var init_targets = __esm({
       subagent_mode_allowed: import_zod4.z.boolean().optional(),
       fallback_targets: import_zod4.z.array(import_zod4.z.string().min(1)).optional()
     }).passthrough();
-    DEFAULT_AZURE_API_VERSION = "2024-12-01-preview";
-    DEFAULT_AZURE_RESPONSES_API_VERSION = "v1";
+    DEFAULT_AZURE_API_VERSION = "v1";
     DEFAULT_OPENAI_BASE_URL = "https://api.openai.com/v1";
     cliErrorMap = (issue, ctx) => {
       if (issue.code === import_zod4.z.ZodIssueCode.unrecognized_keys) {