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/venice/index.test.ts

@@ -0,0 +1,38 @@
+import { describe, expect, it } from "vitest";
+import { registerSingleProviderPlugin } from "../../test/helpers/plugins/plugin-registration.js";
+import plugin from "./index.js";
+
+describe("venice provider plugin", () => {
+  it("applies the shared xAI compat patch to Grok-backed Venice models only", async () => {
+    const provider = await registerSingleProviderPlugin(plugin);
+
+    expect(
+      provider.normalizeResolvedModel?.({
+        modelId: "venice/grok-4",
+        model: {
+          id: "grok-4",
+          compat: {
+            supportsUsageInStreaming: true,
+          },
+        },
+      } as never),
+    ).toMatchObject({
+      compat: {
+        supportsUsageInStreaming: true,
+        toolSchemaProfile: "xai",
+        nativeWebSearchTool: true,
+        toolCallArgumentsEncoding: "html-entities",
+      },
+    });
+
+    expect(
+      provider.normalizeResolvedModel?.({
+        modelId: "venice/llama-3.3-70b",
+        model: {
+          id: "llama-3.3-70b",
+          compat: {},
+        },
+      } as never),
+    ).toBeUndefined();
+  });
+});