.oxlintrc.json (1)

15-15: Consider keeping the legacy tests ignore pattern as well
Previous configs ignored tests/*; the new single entry test is fine, but if any historical folder or developer still places specs under tests/ they will suddenly be lint-checked.
Add both patterns for a painless transition.

   "docs",
-  "test"
+  "test",
+  "tests"

electron-builder-macx64.yml (1)

8-9: Duplicate !test/* entries – drop one for cleanliness
The same exclusion appears twice (lines 8 and 19). Builder tolerates duplicates but it is noisy and increases merge-conflict surface.

-  - '!test/*'   # line 8
   ...
-  - '!test/*'   # line 19

Also applies to: 19-20

electron-builder.yml (1)

8-9: Same !test/* duplication as the mac config
Two identical exclusion lines (8 and 19) – safe but untidy. Remove the second.

Also applies to: 19-20

vitest.config.renderer.ts (1)

43-45: Timeouts: consider lowering for renderer unit tests
10 s per test/hook is generous and may hide hung tests. Typical unit tests complete in < 2 s. If E2E coverage is required, move those to Playwright/Cypress suites.

test/setup.ts (1)

63-71: Use vi.resetModules() to ensure module isolation between tests

vi.clearAllMocks() resets call counts, and vi.restoreAllMocks() restores spied-on methods, but neither unloads mocked modules. Stateful singletons (e.g. Electron mocks) will therefore persist across tests.

afterEach(() => {
   // Clean up after each test
   vi.restoreAllMocks()
+  vi.resetModules()
 })

This guarantees each test receives a fresh module instance.

test/README.md (1)

5-17: Specify a language for the directory-tree code block

Markdown-lint (MD040) flags fenced blocks without a language. Add text (or bash) to improve syntax highlighting and keep the linter quiet.

-```
+```text
 test/
 ├── main/
 ...

<details>
<summary>🧰 Tools</summary>

<details>
<summary>🪛 markdownlint-cli2 (0.17.2)</summary>

5-5: Fenced code blocks should have a language specified
null

(MD040, fenced-code-language)

</details>

</details>

</blockquote></details>
<details>
<summary>vitest.config.ts (1)</summary><blockquote>

`7-9`: **Minor style nit – stray indentation**

Extra two-space indent before `'@shared'` is inconsistent with the rest of the file.  Not blocking, just keeping the config tidy.

```diff
-      '@': resolve('src/main/'),
-        '@shared': resolve('src/shared')
+      '@': resolve('src/main/'),
+      '@shared': resolve('src/shared')

test/setup.renderer.ts (2)

70-88: Global window.api stub looks good but consider freezing structure

Down-stream code may rely on Object.freeze(window.api) behaviour from preload. To mimic production more closely and avoid accidental mutation inside tests:

-Object.defineProperty(window, 'api', {
-  value: {
+Object.defineProperty(window, 'api', {
+  value: Object.freeze({
       devicePresenter: {
         getDeviceInfo: vi.fn(() => Promise.resolve({ /* … */ }))
       },
       windowPresenter: {
         minimize: vi.fn(),
         maximize: vi.fn(),
         close: vi.fn(),
         isMaximized: vi.fn(() => Promise.resolve(false))
       }
-  },
+  }),
   writable: true
 })

Optional, but reduces flakiness in larger suites.

---

103-106: Same isolation concern as main setup – reset mocked modules

Mirror the change suggested for test/setup.ts so renderer tests also start with clean modules.

afterEach(() => {
   vi.restoreAllMocks()
+  vi.resetModules()
 })

test/main/eventbus/eventbus.test.ts (2)

205-209: Prefer a deterministic promise-flush over setTimeout(0)

Using await new Promise(r => setTimeout(r, 0)) makes tests a tiny bit slower and can still race.
Consider a helper such as:

const flushPromises = () => new Promise(queueMicrotask);
await flushPromises();

or await vi.runAllTicks() if fake timers are active.

Also applies to: 221-223, 235-237

---

5-26: Add a global mock reset in afterEach

Mocks are re-created per beforeEach, but any stray vi.spyOn that forgets to mockRestore may leak across suites. Safer pattern:

afterEach(() => {
  vi.restoreAllMocks();
});

test/renderer/shell/main.test.ts (2)

8-17: Smoke test may be redundant & heavy

Importing full ESM bundles of vue, pinia, and vue-i18n only to assert that factory functions exist costs ~100 ms per run and adds no behavioural coverage.
Consider stubbing those imports or removing the test entirely.

---

62-73: Spy-only assertion adds little value

addCollection is a plain vi.fn() and you verify it was called – this does not test your icon system. Either use the real @iconify/vue API or drop the case to keep the suite meaningful.

Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro

.github/workflows/prcheck.yml (1)

35-35: Add trailing newline to satisfy YAML lint.

The file is missing a newline at EOF, causing YAML-lint to error.

-        run: npm run build
+        run: npm run build
+

🧰 Tools

🪛 YAMLlint (1.37.1)

[error] 35-35: no new line character at the end of file

(new-line-at-end-of-file)

test/setup.ts (1)

60-66: 🛠️ Refactor suggestion

path.join/resolve mock still hides platform-specific bugs

-    join: vi.fn((...args) => args.join('/')),
-    resolve: vi.fn((...args) => args.join('/'))
+    join: vi.spyOn(actual as any, 'join'),
+    resolve: vi.spyOn(actual as any, 'resolve')

Spying preserves real semantics while keeping call-count assertions intact.

test/main/presenter/llmProviderPresenter.test.ts (2)

132-150: Still performs real HTTP calls — please stub them

The test hits https://mockllm.anya2a.com/v1, making the suite slow and non-deterministic. Mock the request layer (vi.mock/msw/nock) and assert against the stubbed payload instead.

---

435-451: Possible infinite loop in concurrency section

The for await loop waits for events but a rejected stream may never yield, hanging until Vitest times out. Add a timeout guard (e.g. Promise.race with setTimeout) or abort signal.

electron-builder.yml (1)

8-19: Remove duplicate exclusion and refine test directory pattern.

The !test/* exclusion appears twice (lines 8 and 19) and only excludes top-level files, not nested directories. Consolidate to a single glob that covers all test files:

-  - '!test/*'
+  - '!test/**'

electron-builder-macx64.yml (1)

8-19: Remove duplicate exclusion and refine test directory pattern.

The exclusion !test/* is duplicated (lines 8 and 19) and doesn't match nested files. Replace with a single recursive pattern:

-  - '!test/*'
+  - '!test/**'

.oxlintrc.json (1)

15-15: Ensure ignore pattern excludes nested test files.

The pattern "test" may not ignore subdirectories. Use a glob to match all files under test/:

-    "test"
+    "test/**"

test/setup.ts (1)

44-56: fs.promises.stat still un-stubbed – async callers will fail

Several production files call fs.promises.stat. Add it to the promises mock.

   promises: {
     access: vi.fn(),
     readFile: vi.fn(),
     writeFile: vi.fn(),
     mkdir: vi.fn(),
-    readdir: vi.fn()
+    readdir: vi.fn(),
+    stat: vi.fn()
   }

vitest.config.ts (1)

6-9: Alias indentation & trailing slash inconsistency

The @shared alias is indented one level deeper than @, and src/main/ includes a trailing slash while src/shared does not. Although harmless at runtime, this inconsistency sparks avoidable merge noise across commits.

-    alias: {
-      '@': resolve('src/main/'),
-        '@shared': resolve('src/shared')
-    }
+    alias: {
+      '@': resolve('src/main'),
+      '@shared': resolve('src/shared')
+    }

test/renderer/shell/main.test.ts (1)

58-73: addCollection assertion weak

The expectation only checks the first argument shape. To guarantee both icon sets are registered, assert on the actual payload or call order.

-expect(addCollection).toHaveBeenCalledTimes(2)
-expect(addCollection).toHaveBeenCalledWith(
-  expect.objectContaining({ icons: expect.any(Object) })
-)
+expect(addCollection).toHaveBeenNthCalledWith(1, lucideIcons)
+expect(addCollection).toHaveBeenNthCalledWith(2, vscodeIcons)

test/setup.renderer.ts (2)

15-24: Router mock is missing isReady & afterEach

Many Vue-Router consumers await router.isReady() or register afterEach hooks. Omitting them will crash tests that exercise navigation guards.

   createRouter: vi.fn(() => ({
+    isReady: vi.fn(() => Promise.resolve()),
+    afterEach: vi.fn(),
     push: vi.fn(),

---

61-68: Icon component should be a valid Vue component

Returning a plain object may trigger “runtime missing template compiler” warnings. Wrap the stub with defineComponent.

-import '@iconify/vue'
+import { defineComponent } from 'vue'
 ...
-  Icon: {
-    name: 'Icon',
-    template: '<span></span>'
-  }
+  Icon: defineComponent({
+    name: 'IconStub',
+    template: '<span></span>'
+  })

test/README.md (1)

5-17: Add language tag to fenced block (MD040)

Markdown-lint complains because the directory tree block lacks a language. Mark it as text to silence CI.

-```
+```text

🧰 Tools

🪛 markdownlint-cli2 (0.17.2)

5-5: Fenced code blocks should have a language specified
null

(MD040, fenced-code-language)

test/main/eventbus/eventbus.test.ts (2)

206-212: Flushing async work with setTimeout is brittle

Instead of an empty setTimeout, rely on Vitest helpers to flush micro-tasks. This avoids test flakiness on slower runners.

-await new Promise(resolve => setTimeout(resolve, 0))
+await vi.runAllTicks()

---

334-349: Duplicate error-logging expectation pattern

The three error-handling tests share near-identical setup/teardown. Extract a helper or use describe.each to shrink the suite and reduce maintenance.

No diff provided – structural refactor suggested.

Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro