refactor(cli): group inference and onboard support modules (NVIDIA#3195)

## Summary
Groups the remaining inference and onboarding support modules into the
feature folders from the placement map. This retargets the combined
R4/R5 reorg onto `main` now that the lower stack has landed, while
keeping the large legacy `onboard.ts` flow in place.

## Changes
- Move inference config, health, provider model, prompt, NIM, local
inference, web-search, and onboard probe helpers under
`src/lib/inference/**`.
- Move Ollama-specific helpers under `src/lib/inference/ollama/**` and
vLLM helpers under `src/lib/inference/vllm.ts`.
- Move onboard support modules under `src/lib/onboard/**`.
- Move persisted onboard session state under
`src/lib/state/onboard-session.ts`.
- Update imports, tests, bin shims, installer references, E2E path
probes, scripts, and source-shape checks for the new paths.
- Keep `src/lib/onboard.ts` as the legacy flow entrypoint for future
focused splitting.

## Type of Change
- [x] Code change (feature, bug fix, or refactor)
- [ ] Code change with doc updates
- [ ] Doc only (prose changes, no code sample modifications)
- [ ] Doc only (includes code sample changes)

## Verification
- [x] `npx prek run --all-files` passes
- [x] `npm test` passes
- [x] Tests added or updated for new or changed behavior
- [x] No secrets, API keys, or credentials committed
- [ ] Docs updated for user-facing behavior changes
- [ ] `make docs` builds without warnings (doc changes only)
- [ ] Doc pages follow the [style
guide](https://github.com/NVIDIA/NemoClaw/blob/main/docs/CONTRIBUTING.md)
(doc changes only)
- [ ] New doc pages include SPDX header and frontmatter (new pages only)

---
<!-- DCO sign-off required by CI. Run: git config user.name && git
config user.email -->
Signed-off-by: Carlos Villela <cvillela@nvidia.com>

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **Refactor**
* Project modules reorganized into a new internal folder structure;
import wiring updated across the codebase.
* **Install / Onboarding**
* Installer preflight now uses the relocated onboarding preflight
implementation, affecting which host checks run during install.
* **Tests**
* Test suites and fixtures updated to reference the restructured modules
so tests continue to run against the new layout.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

Author: cv

.coderabbit.yaml

@@ -221,7 +221,7 @@ reviews:
         gh workflow run nightly-e2e.yaml --ref <branch> -f jobs=overlayfs-autofix-e2e
         ```
 
-    - path: "src/lib/preflight.ts"
+    - path: "src/lib/onboard/preflight.ts"
       instructions: *e2e-overlayfs
 
     - path: "src/lib/deploy.ts"

bin/lib/usage-notice.js

@@ -1,9 +1,9 @@
 // SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 // SPDX-License-Identifier: Apache-2.0
 
-// Thin re-export shim — the implementation lives in src/lib/usage-notice.ts,
-// compiled to dist/lib/usage-notice.js.
-const usageNotice = require("../../dist/lib/usage-notice");
+// Thin re-export shim — the implementation lives in src/lib/onboard/usage-notice.ts,
+// compiled to dist/lib/onboard/usage-notice.js.
+const usageNotice = require("../../dist/lib/onboard/usage-notice");
 
 if (require.main === module) {
   usageNotice.cli().catch((error) => {

scripts/check-legacy-migrated-paths.ts

@@ -27,12 +27,12 @@ const REMOVED_SHIM_MOVES: Record<string, string> = {
   "bin/lib/chat-filter.js": "src/lib/chat-filter.ts",
   "bin/lib/config-io.js": "src/lib/state/config-io.ts",
   "bin/lib/debug.js": "src/lib/diagnostics/debug.ts",
-  "bin/lib/inference-config.js": "src/lib/inference-config.ts",
-  "bin/lib/local-inference.js": "src/lib/local-inference.ts",
-  "bin/lib/nim.js": "src/lib/nim.ts",
-  "bin/lib/onboard-session.js": "src/lib/onboard-session.ts",
+  "bin/lib/inference-config.js": "src/lib/inference/config.ts",
+  "bin/lib/local-inference.js": "src/lib/inference/local.ts",
+  "bin/lib/nim.js": "src/lib/inference/nim.ts",
+  "bin/lib/onboard-session.js": "src/lib/state/onboard-session.ts",
   "bin/lib/platform.js": "src/lib/platform.ts",
-  "bin/lib/preflight.js": "src/lib/preflight.ts",
+  "bin/lib/preflight.js": "src/lib/onboard/preflight.ts",
   "bin/lib/registry.js": "src/lib/state/registry.ts",
   "bin/lib/resolve-openshell.js": "src/lib/adapters/openshell/resolve.ts",
   "bin/lib/runtime-recovery.js": "src/lib/runtime-recovery.ts",

scripts/install.sh

@@ -1464,7 +1464,7 @@ verify_nemoclaw() {
 # 5. Onboard
 # ---------------------------------------------------------------------------
 run_installer_host_preflight() {
-  local preflight_module="${NEMOCLAW_SOURCE_ROOT}/dist/lib/preflight.js"
+  local preflight_module="${NEMOCLAW_SOURCE_ROOT}/dist/lib/onboard/preflight.js"
   if ! command_exists node || [[ ! -f "$preflight_module" ]]; then
     return 0
   fi

scripts/ts-migration-assist.ts

@@ -22,17 +22,17 @@ const SPECIAL_REWRITES: Record<string, Array<[string, string]>> = {
   "bin/lib/onboard.js": [
     ['require("./runner")', 'require("../../bin/lib/runner")'],
     ['require("./sandbox-build-context")', 'require("../../bin/lib/sandbox-build-context")'],
-    ['require("./local-inference")', 'require("../../bin/lib/local-inference")'],
-    ['require("./inference-config")', 'require("../../bin/lib/inference-config")'],
+    ['require("./inference/local")', 'require("../../bin/lib/local-inference")'],
+    ['require("./inference/config")', 'require("../../bin/lib/inference-config")'],
     ['require("./platform")', 'require("../../bin/lib/platform")'],
     ['require("./resolve-openshell")', 'require("../../bin/lib/resolve-openshell")'],
     ['require("./credentials/store")', 'require("../../bin/lib/credentials")'],
     ['require("./state/registry")', 'require("../../bin/lib/registry")'],
-    ['require("./nim")', 'require("../../bin/lib/nim")'],
-    ['require("./onboard-session")', 'require("../../bin/lib/onboard-session")'],
+    ['require("./inference/nim")', 'require("../../bin/lib/nim")'],
+    ['require("./state/onboard-session")', 'require("../../bin/lib/onboard-session")'],
     ['require("./policies")', 'require("../../bin/lib/policies")'],
-    ['require("./usage-notice")', 'require("../../bin/lib/usage-notice")'],
-    ['require("./preflight")', 'require("../../bin/lib/preflight")'],
+    ['require("./onboard/usage-notice")', 'require("../../bin/lib/usage-notice")'],
+    ['require("./onboard/preflight")', 'require("../../bin/lib/preflight")'],
   ],
   "bin/nemoclaw.js": [
     ['require("./lib/runner")', 'require("../bin/lib/runner")'],
@@ -42,11 +42,11 @@ const SPECIAL_REWRITES: Record<string, Array<[string, string]>> = {
     ['require("./lib/registry")', 'require("../bin/lib/registry")'],
     ['require("./lib/nim")', 'require("../bin/lib/nim")'],
     ['require("./lib/policies")', 'require("../bin/lib/policies")'],
-    ['require("./lib/inference-config")', 'require("../bin/lib/inference-config")'],
+    ['require("./lib/inference/config")', 'require("../bin/lib/inference-config")'],
     ['require("./lib/version")', 'require("../bin/lib/version")'],
-    ['require("./lib/onboard-session")', 'require("../bin/lib/onboard-session")'],
+    ['require("./lib/state/onboard-session")', 'require("../bin/lib/onboard-session")'],
     ['require("./lib/runtime-recovery")', 'require("../bin/lib/runtime-recovery")'],
-    ['require("./lib/usage-notice")', 'require("../bin/lib/usage-notice")'],
+    ['require("./lib/onboard/usage-notice")', 'require("../bin/lib/usage-notice")'],
     ['require("./lib/services")', 'require("../bin/lib/services")'],
     ['require("./lib/debug")', 'require("./lib/diagnostics/debug")'],
     ['require("./lib/debug-command")', 'require("./lib/diagnostics/debug-command")'],

src/lib/actions/onboard.ts

@@ -2,8 +2,8 @@
 // SPDX-License-Identifier: Apache-2.0
 
 import { listAgents } from "../agent/defs";
-import { runDeprecatedOnboardAliasCommand, runOnboardCommand } from "../onboard-command";
-import { NOTICE_ACCEPT_ENV, NOTICE_ACCEPT_FLAG } from "../usage-notice";
+import { runDeprecatedOnboardAliasCommand, runOnboardCommand } from "../onboard/legacy-command";
+import { NOTICE_ACCEPT_ENV, NOTICE_ACCEPT_FLAG } from "../onboard/usage-notice";
 
 const { onboard: runOnboard } = require("../onboard") as {
   onboard: (options?: unknown) => Promise<void>;

src/lib/actions/sandbox/connect.ts

@@ -7,8 +7,8 @@ import { spawnSync } from "node:child_process";
 import os from "node:os";
 
 import { CLI_NAME } from "../../cli/branding";
-import { parseGatewayInference } from "../../inference-config";
-import { ensureOllamaAuthProxy } from "../../onboard-ollama-proxy";
+import { parseGatewayInference } from "../../inference/config";
+import { ensureOllamaAuthProxy } from "../../inference/ollama/proxy";
 import {
   captureOpenshell,
   getOpenshellBinary,

src/lib/actions/sandbox/destroy.ts

@@ -12,8 +12,8 @@ import {
   type DestroySandboxOptions,
   normalizeDestroySandboxOptions,
 } from "../../domain/lifecycle/options";
-import * as onboardSession from "../../onboard-session";
-import type { Session } from "../../onboard-session";
+import * as onboardSession from "../../state/onboard-session";
+import type { Session } from "../../state/onboard-session";
 import { OPENSHELL_PROBE_TIMEOUT_MS } from "../../adapters/openshell/timeouts";
 import { DASHBOARD_PORT } from "../../core/ports";
 import * as registry from "../../state/registry";
@@ -95,7 +95,7 @@ function cleanupSandboxServices(
     // branch a single-sandbox destroy would leave models loaded on the GPU.
     const sb = registry.getSandbox(sandboxName);
     if (sb?.provider?.includes("ollama")) {
-      const { unloadOllamaModels } = require("../../onboard-ollama-proxy");
+      const { unloadOllamaModels } = require("../../inference/ollama/proxy");
       unloadOllamaModels();
     }
   }
@@ -225,7 +225,7 @@ export async function destroySandbox(
     }
   }
 
-  const nim = require("../../nim") as {
+  const nim = require("../../inference/nim") as {
     stopNimContainer: (sandboxName: string, opts?: { silent?: boolean }) => void;
     stopNimContainerByName: (name: string) => void;
   };
@@ -246,7 +246,7 @@ export async function destroySandbox(
   // through `stopAll()` or directly into `unloadOllamaModels()` based on
   // whether host services are being torn down).
   if (sb?.provider?.includes("ollama")) {
-    const { killStaleProxy } = require("../../onboard-ollama-proxy");
+    const { killStaleProxy } = require("../../inference/ollama/proxy");
     killStaleProxy();
   }
 

src/lib/actions/sandbox/doctor.ts

@@ -10,8 +10,8 @@ import path from "node:path";
 import { CLI_DISPLAY_NAME, CLI_NAME } from "../../cli/branding";
 import { isErrnoException } from "../../core/errno";
 import { recoverNamedGatewayRuntime } from "../../gateway-runtime-action";
-import { probeProviderHealth } from "../../inference-health";
-import { parseGatewayInference } from "../../inference-config";
+import { probeProviderHealth } from "../../inference/health";
+import { parseGatewayInference } from "../../inference/config";
 import { stripAnsi } from "../../adapters/openshell/client";
 import { captureOpenshell } from "../../adapters/openshell/runtime";
 import { OPENSHELL_PROBE_TIMEOUT_MS } from "../../adapters/openshell/timeouts";

src/lib/actions/sandbox/gateway-state.ts

@@ -16,8 +16,8 @@ import {
 const { pruneKnownHostsEntries } = require("../../onboard") as {
   pruneKnownHostsEntries: (contents: string) => string;
 };
-import * as onboardSession from "../../onboard-session";
-import type { Session } from "../../onboard-session";
+import * as onboardSession from "../../state/onboard-session";
+import type { Session } from "../../state/onboard-session";
 import { stripAnsi } from "../../adapters/openshell/client";
 import {
   captureOpenshell,