重构主控编排与运行时预热链路，统一工作区提示词/专家调度协议并补齐 wiki 记忆注入与写回闭环。

同时收敛启动与运维脚本默认行为（含 wiki worker）、更新 Admin 可观测性与相关测试，降低首轮时延并提高运行稳定性。

Made-with: Cursor

openclaw/extensions/openai/media-understanding-provider.ts

@@ -0,0 +1,40 @@
+import {
+  describeImageWithModel,
+  describeImagesWithModel,
+  transcribeOpenAiCompatibleAudio,
+  type AudioTranscriptionRequest,
+  type MediaUnderstandingProvider,
+} from "openclaw/plugin-sdk/media-understanding";
+import { OPENAI_DEFAULT_AUDIO_TRANSCRIPTION_MODEL } from "./default-models.js";
+
+export const DEFAULT_OPENAI_AUDIO_BASE_URL = "https://api.openai.com/v1";
+
+export async function transcribeOpenAiAudio(params: AudioTranscriptionRequest) {
+  return await transcribeOpenAiCompatibleAudio({
+    ...params,
+    provider: "openai",
+    defaultBaseUrl: DEFAULT_OPENAI_AUDIO_BASE_URL,
+    defaultModel: OPENAI_DEFAULT_AUDIO_TRANSCRIPTION_MODEL,
+  });
+}
+
+export const openaiMediaUnderstandingProvider: MediaUnderstandingProvider = {
+  id: "openai",
+  capabilities: ["image", "audio"],
+  defaultModels: {
+    image: "gpt-5.4-mini",
+    audio: OPENAI_DEFAULT_AUDIO_TRANSCRIPTION_MODEL,
+  },
+  autoPriority: { image: 10, audio: 10 },
+  describeImage: describeImageWithModel,
+  describeImages: describeImagesWithModel,
+  transcribeAudio: transcribeOpenAiAudio,
+};
+
+export const openaiCodexMediaUnderstandingProvider: MediaUnderstandingProvider = {
+  id: "openai-codex",
+  capabilities: ["image"],
+  defaultModels: { image: "gpt-5.4" },
+  describeImage: describeImageWithModel,
+  describeImages: describeImagesWithModel,
+};