Fix Studio telemetry account ID resolution

docs/studio-tea-telemetry.md

@@ -47,8 +47,17 @@ Studio 的产品行为数据统一上报到 TEA App `1050062`。火山引擎和
 ## 数据边界
 
 只允许上报已登记的扁平 string/number 字段。禁止上报 Prompt、消息正文、模型响应、
-源码、文件路径、自由文本错误、错误堆栈、Cookie、Token、AK/SK 或其他密钥。错误只保留
-稳定的 `error_kind`、可选 `error_code` 和适用事件的 `failed_phase`。
+源码、文件路径、错误堆栈、Cookie、Token、AK/SK 或其他密钥。错误默认保留稳定的
+`error_kind`、可选 `error_code` 和适用事件的 `failed_phase`。
+`studio_agent_deploy` 失败事件额外允许上报经过前端脱敏与长度限制的 `error_message`，
+用于定位部署、构建和 Runtime 初始化失败原因。构建阶段失败时，`error_message` 优先来自
+前端已收到的 CodePipeline 构建日志文本；没有可用构建日志时再回退到部署接口返回的
+错误摘要。构建日志较长时保留末尾内容，避免截掉通常位于日志尾部的真实失败行。
+后端在构建失败的最终同步中会有限重试拉取 CodePipeline 日志，直到保留的日志尾部包含
+构建错误 marker，降低日志服务延迟导致埋点只拿到镜像下载/解压进度的概率。
+部署主阶段按 `prepare -> upload -> build -> deploy -> publish -> update -> evaluation`
+单调推进；CodePipeline 日志事件只更新构建日志，不应把已经进入部署阶段的任务回退为
+构建阶段。
 
 `user_unique_id` 通过 TEA `config` 设置，不作为自定义事件属性重复发送。所有资源 ID
 在模块边界转换为字符串；布尔属性使用 `0/1`。
@@ -58,3 +67,10 @@ Studio 的产品行为数据统一上报到 TEA App `1050062`。火山引擎和
 本迁移仅删除 Studio 产品行为埋点使用的前端 APMPlus Web SDK 和配置传递链路。
 `veadk/tracing/telemetry/`、APMPlus OpenTelemetry exporter、Runtime trace 和问题反馈中的
 APMPlus 查询能力不在迁移范围内并继续保留。
+
+## 开发排障记录
+
+- 检查 Studio 后端使用的 AgentKit SDK 时，应优先使用仓库虚拟环境
+  `.venv/bin/python`。当前全局 `python` 命令可能不存在，且全局 `python3` 环境可能未安装
+  `agentkit` 包，直接执行 `python3 -c 'import agentkit'` 会得到
+  `ModuleNotFoundError: No module named 'agentkit'`。

frontend/server/studio_update_resources.py

@@ -20,6 +20,10 @@
 from typing import Any, Literal
 
 from frontend.server.storage.provisioning import resolve_studio_storage_for_deploy
+from veadk.cli.studio_account_id import (
+    resolve_studio_account_id_metadata,
+    studio_account_id_environment,
+)
 from veadk.utils.cloud_provider import CloudProvider
 
 SnapshotKind = Literal["codex", "openclaw", "hermes"]
@@ -31,10 +35,10 @@
 )
 
 
-def _function_config(
+def _function_state(
     function_client: Any,
     function_id: str,
-) -> dict[str, str]:
+) -> tuple[object, dict[str, str]]:
     import volcenginesdkvefaas
 
     function = function_client.get_function(
@@ -45,7 +49,7 @@ def _function_config(
         for item in (getattr(function, "envs", None) or [])
         if getattr(item, "key", None)
     }
-    return environment
+    return function, environment
 
 
 def _provision_snapshot_tool(
@@ -135,7 +139,7 @@ def reconcile_studio_update_resources(
     session_token: str,
 ) -> dict[str, str]:
     """Return environment overrides for resources missing from an older Studio."""
-    environment = _function_config(function_client, function_id)
+    function, environment = _function_state(function_client, function_id)
     overrides: dict[str, str] = {}
 
     from veadk.cli.studio_knowledge_signing import (
@@ -167,6 +171,22 @@ def reconcile_studio_update_resources(
             }
         )
 
+    account_resolution = resolve_studio_account_id_metadata(
+        environment={**environment, **overrides},
+        remote_function=function,
+        access_key=access_key,
+        secret_key=secret_key,
+        session_token=session_token,
+        region=region,
+        provider=provider,
+    )
+    overrides.update(
+        studio_account_id_environment(
+            account_resolution,
+            clear_error_on_success=True,
+        )
+    )
+
     missing_snapshot_tools = [
         item for item in _SNAPSHOT_ENVIRONMENTS if not environment.get(item[0])
     ]

frontend/src/App.tsx

@@ -2937,6 +2937,7 @@ export default function App() {
         environment,
         cloudProvider: cfg.provider,
         accountId: studio?.accountId ?? "",
+        accountIdResolutionError: studio?.accountIdResolutionError ?? "",
       });
       trackStudioEntryViewed({ authState: "anonymous" });
       setFeatures(cfg.features);

frontend/src/adk/client.ts

@@ -2637,6 +2637,7 @@ export interface StudioTelemetryContext {
   project: string;
   version: string;
   accountId?: string;
+  accountIdResolutionError?: string;
 }
 
 export interface StudioTelemetryConfig {
@@ -2715,6 +2716,10 @@ function normalizeStudioTelemetryConfig(value: unknown): StudioTelemetryConfig {
       project: typeof studio.project === "string" ? studio.project : "",
       version: typeof studio.version === "string" ? studio.version : "",
       accountId: typeof studio.accountId === "string" ? studio.accountId : "",
+      accountIdResolutionError:
+        typeof studio.accountIdResolutionError === "string"
+          ? studio.accountIdResolutionError
+          : "",
     },
   };
 }

frontend/src/automations/feishu/FeishuBotIntegration.tsx

@@ -15,6 +15,7 @@ import {
 import {
   beginAgentDeploy,
   classifyTelemetryError,
+  safeTelemetryErrorMessage,
   type AgentDeployFailedProps,
 } from "../../telemetry";
 import feishuLogo from "../../assets/feishu-logo.svg";
@@ -227,6 +228,7 @@ export function FeishuBotIntegration({ onBack }: FeishuBotIntegrationProps) {
         operation.fail({
           failedPhase: telemetryDeployPhase(latestPhaseRef.current),
           errorKind: "abort",
+          errorMessage: safeTelemetryErrorMessage("用户取消部署"),
         });
         return;
       }
@@ -242,6 +244,7 @@ export function FeishuBotIntegration({ onBack }: FeishuBotIntegrationProps) {
         ...(cancelledRef.current
           ? { errorKind: "abort" as const }
           : classifyTelemetryError(error, { phase: latestPhaseRef.current })),
+        errorMessage: safeTelemetryErrorMessage(error),
       });
       if (!mountedRef.current || cancelledRef.current) return;
       setDeploymentStatus("failed");
@@ -268,6 +271,7 @@ export function FeishuBotIntegration({ onBack }: FeishuBotIntegrationProps) {
       deploymentOperationRef.current?.fail({
         failedPhase: telemetryDeployPhase(latestPhaseRef.current),
         errorKind: "abort",
+        errorMessage: safeTelemetryErrorMessage("用户取消部署"),
       });
       if (mountedRef.current) setDeploymentStatus("cancelled");
     } catch (error) {

frontend/src/telemetry/index.ts

@@ -1,6 +1,6 @@
 import { TeaClient, type TeaClientConfig } from "./client";
 import { TelemetryRuntime } from "./runtime";
-export { classifyTelemetryError } from "./privacy";
+export { classifyTelemetryError, safeTelemetryErrorMessage } from "./privacy";
 export type {
   ClassifiedTelemetryError,
   TelemetryErrorContext,

frontend/src/telemetry/privacy.ts

@@ -16,10 +16,19 @@ export interface TelemetryErrorContext {
 
 interface ErrorShape {
   code?: unknown;
+  message?: unknown;
   name?: unknown;
   status?: unknown;
 }
 
+interface TelemetryMessageOptions {
+  preserveEnd?: boolean;
+}
+
+const DEFAULT_STRING_MAX_LENGTH = 256;
+const ERROR_MESSAGE_MAX_LENGTH = 1024;
+const REDACTED = "[REDACTED]";
+
 function stableErrorCode(value: unknown): string | undefined {
   if (typeof value !== "string" && typeof value !== "number") return undefined;
   const code = String(value).trim();
@@ -33,6 +42,58 @@ function classifiedTelemetryError(
   return errorCode === undefined ? { errorKind } : { errorKind, errorCode };
 }
 
+function truncateTelemetryString(
+  value: string,
+  maxLength: number,
+  options: TelemetryMessageOptions = {},
+): string {
+  if (value.length <= maxLength) return value;
+  if (options.preserveEnd) {
+    const prefix = "[truncated] ...";
+    return `${prefix}${value.slice(-Math.max(0, maxLength - prefix.length))}`;
+  }
+  const suffix = "... [truncated]";
+  return `${value.slice(0, Math.max(0, maxLength - suffix.length))}${suffix}`;
+}
+
+function redactTelemetryMessage(value: string): string {
+  return value
+    .replace(
+      /\b(Authorization\s*[:=]\s*)(Bearer\s+)?[^\s"',;&]+/gi,
+      (_match, prefix: string, bearer: string | undefined) =>
+        `${prefix}${bearer ?? ""}${REDACTED}`,
+    )
+    .replace(/\bBearer\s+[A-Za-z0-9._~+/=-]+/gi, `Bearer ${REDACTED}`)
+    .replace(
+      /\b([\w.-]*(?:token|password|passwd|secret|api[_-]?key|access[_-]?key|secret[_-]?key|cookie)[\w.-]*\s*[:=]\s*)(["']?)[^\s"',;&]+/gi,
+      (_match, prefix: string, quote: string) => `${prefix}${quote}${REDACTED}`,
+    );
+}
+
+/** Returns a compact, redacted error message suitable for product telemetry. */
+export function safeTelemetryErrorMessage(
+  error: unknown,
+  options: TelemetryMessageOptions = {},
+): string | undefined {
+  const shape = error !== null && typeof error === "object"
+    ? error as ErrorShape
+    : {};
+  const raw = typeof shape.message === "string"
+    ? shape.message
+    : typeof error === "string" ||
+        typeof error === "number" ||
+        typeof error === "boolean"
+    ? String(error)
+    : "";
+  const normalized = raw.replace(/\s+/g, " ").trim();
+  if (!normalized) return undefined;
+  return truncateTelemetryString(
+    redactTelemetryMessage(normalized),
+    ERROR_MESSAGE_MAX_LENGTH,
+    options,
+  );
+}
+
 /** Converts an unknown failure to an approved category without reading its text. */
 export function classifyTelemetryError(
   error: unknown,
@@ -97,6 +158,7 @@ const COMMON_KEYS = [
   "environment",
   "cloud_provider",
   "account_id",
+  "account_id_resolution_error",
   "user_role",
   "user_source",
   "page_instance_id",
@@ -120,6 +182,7 @@ const EVENT_KEYS: Record<StudioTelemetryEventName, readonly string[]> = {
     "failed_phase",
     "error_kind",
     "error_code",
+    "error_message",
   ],
   studio_sandbox_create: [
     "status",
@@ -194,7 +257,16 @@ export function sanitizeTelemetryPayload(
   const payload: TelemetryPayload = {};
   for (const [key, value] of Object.entries(input)) {
     if (!allowed.has(key) || !isTelemetryValue(value)) continue;
-    payload[key] = typeof value === "string" ? value.slice(0, 256) : value;
+    if (typeof value === "string") {
+      payload[key] = truncateTelemetryString(
+        value,
+        key === "error_message"
+          ? ERROR_MESSAGE_MAX_LENGTH
+          : DEFAULT_STRING_MAX_LENGTH,
+      );
+    } else {
+      payload[key] = value;
+    }
   }
   return payload;
 }

frontend/src/telemetry/runtime.ts

@@ -77,6 +77,7 @@ export class TelemetryRuntime {
     this.context = {
       ...context,
       accountId: context.accountId?.trim() ?? "",
+      accountIdResolutionError: context.accountIdResolutionError?.trim() ?? "",
     };
   }
 
@@ -124,6 +125,8 @@ export class TelemetryRuntime {
       environment: this.context.environment,
       cloud_provider: this.context.cloudProvider,
       account_id: this.context.accountId,
+      account_id_resolution_error:
+        this.context.accountIdResolutionError || undefined,
       page_instance_id: this.pageInstanceId,
       auth_state: props.authState,
     }));
@@ -146,6 +149,7 @@ export class TelemetryRuntime {
       failed_phase: result.failedPhase,
       error_kind: result.errorKind,
       error_code: result.errorCode,
+      error_message: result.errorMessage,
     }));
   }
 
@@ -289,6 +293,8 @@ export class TelemetryRuntime {
       environment: this.context.environment,
       cloud_provider: this.context.cloudProvider,
       account_id: this.identity.accountId,
+      account_id_resolution_error:
+        this.context.accountIdResolutionError || undefined,
       user_role: this.identity.userRole,
       user_source: this.identity.userSource,
       page_instance_id: this.pageInstanceId,

frontend/src/telemetry/schema.ts

@@ -29,6 +29,7 @@ export interface StudioTelemetryContext {
   environment: TelemetryEnvironment;
   cloudProvider: "volcengine" | "byteplus";
   accountId?: string;
+  accountIdResolutionError?: string;
 }
 
 export interface TelemetryIdentity {
@@ -105,6 +106,7 @@ export interface AgentDeployFailedProps {
     | "unknown";
   errorKind: ErrorKind;
   errorCode?: string;
+  errorMessage?: string;
 }
 
 export type SandboxKind = "codex" | "deepseek-harness" | "openclaw" | "hermes";