fix: address review — deterministic dates, use collection length

package.json

@@ -121,6 +121,7 @@
   },
   "devDependencies": {
     "@eslint/js": "catalog:",
+    "@faker-js/faker": "catalog:",
     "@intlify/eslint-plugin-vue-i18n": "catalog:",
     "@lobehub/i18n-cli": "catalog:",
     "@nx/eslint": "catalog:",

pnpm-workspace.yaml

@@ -7,6 +7,7 @@ catalog:
   '@astrojs/vue': ^5.0.0
   '@comfyorg/comfyui-electron-types': 0.6.2
   '@eslint/js': ^9.39.1
+  '@faker-js/faker': ^10.4.0
   '@formkit/auto-animate': ^0.9.0
   '@iconify-json/lucide': ^1.1.178
   '@iconify/json': ^2.2.380

temp/plans/fakerjs-evaluation.md

@@ -0,0 +1,34 @@
+# FakerJS Evaluation
+
+## Seed Stability
+- Are outputs identical across runs with same seed? **Partially** — all fields except `faker.date.recent()` are identical. Date-based generators use `new Date()` as a reference point, so they shift between runs even with the same seed.
+- Do other fields change when schema changes? **Yes** — inserting a new faker call before existing calls shifts the PRNG sequence, causing all downstream fields to produce different values. This is the single-stream RNG problem.
+
+## Schema Mutation Test Results
+
+Adding a field (`author: faker.person.fullName()`) before `type` in a 5-record array:
+- Record 0: `name` and `size` stable, `type` changed
+- Records 1–4: `name`, `size`, and `type` all changed
+
+This confirms that faker uses a single global PRNG stream. Any insertion/deletion of a faker call shifts all subsequent outputs.
+
+## Pros
+- Rich API for realistic test data (names, files, paths, numbers, lorem text)
+- Seeded output is deterministic within a single schema version (non-date fields)
+- Good for generating bulk data (e.g., 50 items for infinite scroll testing)
+- Lightweight, well-maintained, no native dependencies
+- Already available via pnpm catalog
+
+## Cons
+- **Schema fragility**: Adding, removing, or reordering faker calls in a generator shifts all downstream values. This will invalidate screenshot goldens whenever fixture schemas change.
+- **Date instability**: `faker.date.recent()`, `faker.date.past()`, etc. use wall-clock time as a reference, making them non-deterministic even with a seed. Must use `faker.date.between()` with fixed boundaries instead.
+- **No per-field isolation**: Unlike property-based testing libraries, faker has a single PRNG stream. There is no built-in way to isolate individual fields from schema changes.
+- **Mitigation complexity**: Workarounds exist (per-record re-seeding, `faker.seed(baseIndex + i)` per item) but add boilerplate and reduce readability.
+
+## Recommendation
+- **Use** for generating bulk test data in unit/integration tests where screenshot stability is not required (e.g., testing infinite scroll behavior, list rendering, search/filter logic).
+- **Do not use** for screenshot/golden-image test fixtures — schema changes will cascade into visual diffs. For those, prefer hand-crafted or static JSON fixtures.
+- **If adopted**, establish these conventions:
+  1. Always use `faker.date.between({ from: '2024-01-01', to: '2024-12-31' })` instead of `faker.date.recent()`.
+  2. Consider per-record seeding (`faker.seed(BASE_SEED + index)`) to isolate records from each other when schema stability matters.
+  3. Pin the `@faker-js/faker` major version to avoid cross-version PRNG changes.

temp/scripts/faker-mutation-test.ts

@@ -0,0 +1,32 @@
+/* eslint-disable no-console */
+import { faker } from '@faker-js/faker'
+
+// Original schema
+faker.seed(12345)
+const original = Array.from({ length: 5 }, () => ({
+  name: faker.system.fileName(),
+  size: faker.number.int({ min: 1000, max: 10000000 }),
+  type: faker.helpers.arrayElement(['image', 'model', 'checkpoint']),
+}))
+
+// Modified schema — extra field inserted BEFORE 'type'
+faker.seed(12345)
+const modified = Array.from({ length: 5 }, () => ({
+  name: faker.system.fileName(),
+  size: faker.number.int({ min: 1000, max: 10000000 }),
+  author: faker.person.fullName(), // NEW FIELD inserted before 'type'
+  type: faker.helpers.arrayElement(['image', 'model', 'checkpoint']),
+}))
+
+console.log('=== Schema mutation stability test ===')
+for (let i = 0; i < original.length; i++) {
+  const nameMatch = original[i].name === modified[i].name
+  const sizeMatch = original[i].size === modified[i].size
+  const typeMatch = original[i].type === modified[i].type
+  console.log(
+    `Record ${i}: name=${nameMatch ? 'STABLE' : 'CHANGED'} size=${sizeMatch ? 'STABLE' : 'CHANGED'} type=${typeMatch ? 'STABLE' : 'CHANGED'}`
+  )
+  if (!nameMatch) console.log(`  name: "${original[i].name}" -> "${modified[i].name}"`)
+  if (!sizeMatch) console.log(`  size: ${original[i].size} -> ${modified[i].size}`)
+  if (!typeMatch) console.log(`  type: "${original[i].type}" -> "${modified[i].type}"`)
+}

temp/scripts/faker-poc.ts

@@ -0,0 +1,27 @@
+/* eslint-disable no-console */
+import { faker } from '@faker-js/faker'
+
+// Seed for determinism
+faker.seed(12345)
+
+// Generate 10 mock asset records
+const assets = Array.from({ length: 10 }, () => ({
+  name: faker.system.fileName(),
+  size: faker.number.int({ min: 1000, max: 10000000 }),
+  type: faker.helpers.arrayElement(['image', 'model', 'checkpoint']),
+  path: faker.system.filePath(),
+  modified: faker.date.between({ from: '2024-01-01', to: '2024-12-31' }).toISOString(),
+}))
+
+console.log('Assets:', JSON.stringify(assets, null, 2))
+
+// Generate 50 items for infinite scroll
+faker.seed(12345) // Re-seed to test stability
+const scrollItems = Array.from({ length: 50 }, (_, i) => ({
+  id: i,
+  title: faker.lorem.words(3),
+  description: faker.lorem.sentence(),
+}))
+
+console.log(`\nGenerated ${scrollItems.length} scroll items`)
+console.log('First 3:', JSON.stringify(scrollItems.slice(0, 3), null, 2))