fix(plugin): show actual configured model in banner (Closes #24)

nemoclaw/src/index.ts

@@ -11,6 +11,7 @@
  * time.
  */
 
+import { execFileSync } from "node:child_process";
 import { handleSlashCommand } from "./commands/slash.js";
 import {
   describeOnboardEndpoint,
@@ -19,6 +20,31 @@ import {
 } from "./onboard/config.js";
 import { scanForSecrets, isMemoryPath } from "./security/secret-scanner.js";
 
+// Resolve live inference config from OpenShell as a fallback when the
+// onboard config file is not available (e.g. when running inside the
+// sandbox). Returns empty strings if the probe fails.
+function probeOpenShellInference(): { endpoint: string; provider: string; model: string } {
+  try {
+    const raw = execFileSync("openshell", ["inference", "get", "--json"], {
+      encoding: "utf-8",
+      timeout: 3000,
+      stdio: ["pipe", "pipe", "pipe"],
+    });
+    const parsed = JSON.parse(raw) as {
+      provider?: string;
+      model?: string;
+      endpoint?: string;
+    };
+    return {
+      endpoint: parsed.endpoint ?? parsed.provider ?? "",
+      provider: parsed.provider ?? "",
+      model: parsed.model ?? "",
+    };
+  } catch {
+    return { endpoint: "", provider: "", model: "" };
+  }
+}
+
 // ---------------------------------------------------------------------------
 // OpenClaw Plugin SDK compatible types (mirrors openclaw/plugin-sdk)
 // ---------------------------------------------------------------------------
@@ -151,8 +177,10 @@ export interface NemoClawConfig {
 
 function activeModelEntries(
   onboardCfg: ReturnType<typeof loadOnboardConfig>,
+  fallbackModel = "",
 ): ModelProviderEntry[] {
-  if (!onboardCfg?.model) {
+  const activeModel = onboardCfg?.model ?? fallbackModel;
+  if (!activeModel) {
     return [
       {
         id: "nvidia/nemotron-3-super-120b-a12b",
@@ -183,8 +211,8 @@ function activeModelEntries(
 
   return [
     {
-      id: `inference/${onboardCfg.model}`,
-      label: onboardCfg.model,
+      id: `inference/${activeModel}`,
+      label: activeModel,
       contextWindow: 131072,
       maxOutput: 8192,
     },
@@ -194,6 +222,7 @@ function activeModelEntries(
 function registeredProviderForConfig(
   onboardCfg: ReturnType<typeof loadOnboardConfig>,
   providerCredentialEnv: string,
+  fallbackModel = "",
 ): ProviderPlugin {
   const authLabel =
     providerCredentialEnv === "NVIDIA_API_KEY"
@@ -205,7 +234,7 @@ function registeredProviderForConfig(
     label: "Managed Inference Route",
     aliases: ["inference-local", "nemoclaw"],
     envVars: [providerCredentialEnv],
-    models: { chat: activeModelEntries(onboardCfg) },
+    models: { chat: activeModelEntries(onboardCfg, fallbackModel) },
     auth: [
       {
         type: "bearer",
@@ -264,12 +293,29 @@ export default function register(api: OpenClawPluginApi): void {
 
   // 2. Register nvidia-nim provider — use onboard config if available
   const onboardCfg = loadOnboardConfig();
-  const providerCredentialEnv = onboardCfg?.credentialEnv ?? "NVIDIA_API_KEY";
-  api.registerProvider(registeredProviderForConfig(onboardCfg, providerCredentialEnv));
 
-  const bannerEndpoint = onboardCfg ? describeOnboardEndpoint(onboardCfg) : "build.nvidia.com";
-  const bannerProvider = onboardCfg ? describeOnboardProvider(onboardCfg) : "NVIDIA Endpoints";
-  const bannerModel = onboardCfg?.model ?? "nvidia/nemotron-3-super-120b-a12b";
+  // Prefer onboard config; fall back to live OpenShell inference state when
+  // the config file is unavailable (e.g. inside the sandbox). Only resort to
+  // hardcoded defaults if both lookups fail.
+  let bannerEndpoint = onboardCfg ? describeOnboardEndpoint(onboardCfg) : "";
+  let bannerProvider = onboardCfg ? describeOnboardProvider(onboardCfg) : "";
+  let bannerModel = onboardCfg?.model ?? "";
+  let probedModel = "";
+
+  if (!bannerEndpoint || !bannerProvider || !bannerModel) {
+    const probed = probeOpenShellInference();
+    if (!bannerEndpoint) bannerEndpoint = probed.endpoint;
+    if (!bannerProvider) bannerProvider = probed.provider;
+    if (!bannerModel) bannerModel = probed.model;
+    probedModel = probed.model;
+  }
+
+  if (!bannerEndpoint) bannerEndpoint = "build.nvidia.com";
+  if (!bannerProvider) bannerProvider = "NVIDIA Endpoints";
+  if (!bannerModel) bannerModel = "nvidia/nemotron-3-super-120b-a12b";
+
+  const providerCredentialEnv = onboardCfg?.credentialEnv ?? "NVIDIA_API_KEY";
+  api.registerProvider(registeredProviderForConfig(onboardCfg, providerCredentialEnv, probedModel));
 
   // 3. Register before_tool_call hook to block secrets in memory writes (#1233)
   // NOTE: This relies on OpenClaw's before_tool_call plugin hook contract

nemoclaw/src/register.test.ts

@@ -4,15 +4,21 @@
 import { describe, it, expect, vi, beforeEach } from "vitest";
 import type { OpenClawPluginApi } from "./index.js";
 
+vi.mock("node:child_process", () => ({
+  execFileSync: vi.fn(),
+}));
+
 vi.mock("./onboard/config.js", () => ({
   loadOnboardConfig: vi.fn(),
   describeOnboardEndpoint: vi.fn(() => "build.nvidia.com"),
   describeOnboardProvider: vi.fn(() => "NVIDIA Endpoint API"),
 }));
 
+import { execFileSync } from "node:child_process";
 import register, { getPluginConfig } from "./index.js";
 import { loadOnboardConfig } from "./onboard/config.js";
 
+const mockedExecFileSync = vi.mocked(execFileSync);
 const mockedLoadOnboardConfig = vi.mocked(loadOnboardConfig);
 
 function createMockApi(): OpenClawPluginApi {
@@ -39,6 +45,7 @@ function createMockApi(): OpenClawPluginApi {
 describe("plugin registration", () => {
   beforeEach(() => {
     vi.clearAllMocks();
+    mockedExecFileSync.mockReset();
     mockedLoadOnboardConfig.mockReturnValue(null);
   });
 
@@ -77,6 +84,34 @@ describe("plugin registration", () => {
       expect.objectContaining({ id: "inference/nvidia/custom-model" }),
     ]);
   });
+
+  it("uses probed OpenShell provider and model when onboard config is unavailable", () => {
+    mockedExecFileSync.mockReturnValue(
+      JSON.stringify({
+        provider: "Ollama",
+        endpoint: "http://host.docker.internal:11434/v1",
+        model: "llama3.2:latest",
+      }),
+    );
+
+    const api = createMockApi();
+    register(api);
+
+    const providerArg = vi.mocked(api.registerProvider).mock.calls[0][0];
+    expect(providerArg.models?.chat).toEqual([
+      expect.objectContaining({
+        id: "inference/llama3.2:latest",
+        label: "llama3.2:latest",
+      }),
+    ]);
+
+    const logLines = vi.mocked(api.logger.info).mock.calls.map(([message]) => message);
+    expect(
+      logLines.some((line) => line.includes("Endpoint:  http://host.docker.internal:11434/v1")),
+    ).toBe(true);
+    expect(logLines.some((line) => line.includes("Provider:  Ollama"))).toBe(true);
+    expect(logLines.some((line) => line.includes("Model:     llama3.2:latest"))).toBe(true);
+  });
 });
 
 describe("before_tool_call secret scanner hook (#1233)", () => {