Run browser tests in parallel on per-worker accounts

.claude/agents/qa.md

@@ -12,16 +12,16 @@ claims, including regression paths, loading/empty/error states, mobile
 layouts, and denied access where relevant.
 
 - Unit/integration: the narrowest `vp run test` scope that covers the slice.
-- Browser: `just e2e-local` (`vp run test:e2e:local`) against Vite on
-  `http://localhost:5173`, not the Convex site proxy. It needs a running
+- Browser: `just journeys` (`vp run test:journeys`) runs the browser-mode
+  journeys in `apps/web/e2e` against the local backend. It needs a running
   `just dev` stack; if none is running and you were not told you own one,
   report that instead of starting it.
 - Log in with the seeded demo owner `demo@groam.example` / `GroamDemo123!`.
   Never reset or reseed data.
 - Adding coverage: read `add-web-view-test`, `add-convex-test`, or
   `add-test-id` first, and edit only test files the brief names.
 
-Use `rg` to find existing specs under `apps/web/e2e/` and tests near the
+Use `rg` to find existing journeys under `apps/web/e2e/` and tests near the
 changed code. No fabricated screenshots, skipped failures, or arbitrary sleeps.
 
 Return per criterion: pass/fail, reproducible steps, actual versus expected.

.claude/skills/add-convex-test/SKILL.md

@@ -10,7 +10,7 @@ description: >-
 # Add a Convex test
 
 Colocate `<topic>.test.ts` next to the module under
-`packages/backend/convex/modules/...` (Convex excludes `*.test.ts` from
+`packages/backend/convex/modules/<domain>/` (Convex excludes `*.test.ts` from
 deploys). Import test APIs from `vite-plus/test`, not `vitest`.
 
 ## Setup
@@ -62,6 +62,10 @@ test('adds an activity to a scheduled destination', { timeout: 15_000 }, async (
 - Tests go through `client.query` / `client.mutation` / `client.action`;
   `ctx.db` stays in modules and routes.
 - Use `{ timeout: 15_000 }` as the floor for authenticated fixtures.
+- Scheduled functions wait on fake timers; drive them with
+  `finishAllScheduledFunctions(vi.runAllTimers)` or call `vi.useRealTimers()`.
+- Console output fails the test unless declared with `expectConsole(level,
+  pattern)` from `@groam/vitest-config/console`.
 
 ## Run
 

.claude/skills/add-env-variable/SKILL.md

@@ -29,7 +29,7 @@ scripts.
 | `runtime` | installed-app runtime | `GROAM_CONVEX_URL`, `GROAM_DATA_DIR` |
 | `desktop-tooling` | `apps/desktop/scripts` | `VITE_PHOTON_URL`, `CI` |
 | `seeder` | `tooling/seeder` | `SEED_CONVEX_URL`, `SEED_USER_EMAIL` |
-| `playwright` | web E2E | `PLAYWRIGHT_BASE_URL`, seed credentials |
+| `journeys` | browser journeys (`apps/web/e2e`) | `VITE_CONVEX_URL`, `VITE_CONVEX_SITE_URL` |
 | `showcase` | `tooling/showcase` | `SHOWCASE_*` |
 | `vitest` | `tooling/vitest-config` | `CI` |
 

.claude/skills/add-test-id/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: add-test-id
 description: >-
-  Add a stable data-testid for Groam Playwright and Testing Library locators.
+  Add a stable data-testid for Groam browser-mode view tests and journeys.
   Use when a new control, dialog, row, or empty-state action needs a test
   hook — ids live in @groam/app-actions/ids, never as inline strings in one
   app.
@@ -10,7 +10,7 @@ description: >-
 # Add a test id
 
 Stable locators are the `ids` object in
-`tooling/app-actions/src/playwright/ids.ts`, published as
+`tooling/app-actions/src/testing/ids.ts`, published as
 `@groam/app-actions/ids`. The web app re-exports it as `testIds` from
 `@/lib/test-ids`.
 
@@ -31,13 +31,18 @@ Stable locators are the `ids` object in
    <Button data-testid={testIds.agentStop}>Stop</Button>
    ```
 
-3. Query the same key in tests: `screen.getByTestId(testIds.agentStop)` in
-   Testing Library (`add-web-view-test`), `page.getByTestId(ids.agentStop)`
-   in Playwright.
-4. For a repeated browser journey, add or reuse a helper in
-   `tooling/app-actions/src/playwright/<verb>.ts` (exported from its
-   `index.ts`) instead of repeating the steps in a spec under
-   `apps/web/e2e/`.
+3. Query the same key in tests: `page.getByTestId(testIds.agentStop)` in
+   view tests (`add-web-view-test`) and journeys.
+4. For a repeated browser step, add or reuse a helper in
+   `tooling/app-actions/src/vitest/<verb>.ts` (exported from its
+   `index.ts`) built on `page` locators from `vite-plus/test/browser`,
+   instead of repeating the steps in a test.
+5. View tests and journeys use the same steps. Journeys
+   (`apps/web/e2e/**/*.test.tsx`) mount the whole app with `openApp(path)`
+   from `@/testing/open-app`, signed in as the file's own account and group;
+   `arrange` from `@groam/app-actions/vitest` creates trips, ideas, issues,
+   and members through the backend, and other people act through it too.
+   Drive only the journey under test through the UI.
 
 ## Rules
 

.claude/skills/add-web-view-test/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: add-web-view-test
 description: >-
-  Add a Vitest + Testing Library test for a Groam web feature view. Use when
+  Add a Vite+ browser-mode test for a Groam web feature view. Use when
   covering a React screen, loading/empty/error state, dialog, or sidebar row
   in apps/web — mock feature hooks and agent context, never spin up Convex
   (use add-convex-test for backend behavior).
@@ -10,15 +10,18 @@ description: >-
 # Add a web view test
 
 Colocate `<name>.test.tsx` next to the component in
-`apps/web/src/features/<domain>/`. Mock the feature hooks, workspace state,
-router, and agent context; render the view; assert roles, copy, and test
-ids.
+`apps/web/src/features/<domain>/`. View tests run in real Chromium (the
+`views` browser setup of `vp test`) with the app's styles. Mock the feature
+hooks, workspace state, router, and agent context; render the view; drive it
+with `page` locators; assert with `expect.element`, which retries. Testing
+Library and jsdom are not used.
 
 ## Pattern (from `agents/agent-list/agents-list-view.test.tsx`)
 
 ```tsx
-import { cleanup, render, screen } from '@testing-library/react';
-import { afterEach, beforeEach, describe, expect, test, vi } from 'vite-plus/test';
+import { render } from '@groam/vitest-config/react';
+import { beforeEach, describe, expect, test, vi } from 'vite-plus/test';
+import { page } from 'vite-plus/test/browser';
 import { testIds } from '@/lib/test-ids';
 import { AgentsListView } from './agents-list-view';
 
@@ -51,23 +54,39 @@ beforeEach(() => {
   state.isLoading = false;
   state.runs = [];
 });
-afterEach(cleanup);
 
 describe('AgentsListView', () => {
-  test('shows a labeled loading state', () => {
+  test('shows a labeled loading state', async () => {
     state.isLoading = true;
-    render(<AgentsListView />);
-    expect(screen.getByRole('status', { name: 'Loading agent runs…' })).toBeTruthy();
+    await render(<AgentsListView />);
+    await expect.element(page.getByRole('status', { name: 'Loading agent runs…' })).toBeVisible();
   });
 
-  test('shows the empty state', () => {
-    render(<AgentsListView />);
-    expect(screen.getByText('Nothing running yet')).toBeTruthy();
-    expect(screen.queryByTestId(testIds.agentsRoster)).toBeNull();
+  test('shows the empty state', async () => {
+    await render(<AgentsListView />);
+    await expect.element(page.getByText('Nothing running yet')).toBeVisible();
+    await expect.element(page.getByTestId(testIds.agentsRoster)).not.toBeInTheDocument();
   });
 });
 ```
 
+`render` and `renderHook` come from `@groam/vitest-config/react` and unmount
+after each test. Interact like a user: `await page.getByLabelText(...).fill()`,
+`.click()`. For many input combinations, drive the view through the shared
+steps in `@groam/app-actions/vitest` (for example `createTrip(input)`) with
+`test.each`, as `trip-create/create-trip-dialog.test.tsx` does.
+
+```tsx
+test.each<[string, CreateTripFormInput]>([
+  ['only a name', { name }],
+  ['a budget', { budget: 2500, name }]
+])('creates a trip with %s', async (_, input) => {
+  await render(<CreateTripDialog createTrip={createTrip} onClose={vi.fn()} onCreated={vi.fn()} />);
+  await fillCreateTrip(input);
+  await expect.poll(() => createTrip.mock.calls.length).toBe(1);
+});
+```
+
 Reuse existing mocks before writing new ones: `@/testing/stub-match-media`,
 `@/testing/trip-planner-fixture`, `features/trips/test-workspace-mocks.tsx`,
 `features/agents/agent-test-router-mock.tsx`. Mock workspace state at
@@ -80,8 +99,12 @@ Reuse existing mocks before writing new ones: `@/testing/stub-match-media`,
   `@groam/app-actions/ids` (`add-test-id`).
 - Hook logic gets its own `hooks/use-<name>.test.ts(x)`; `packages/ui`
   primitives get a `*.test.tsx` beside the primitive.
+- Console output fails the test unless declared with `expectConsole(level,
+  pattern)` from `@groam/vitest-config/console`; never silence it with
+  `vi.spyOn(console, …)`.
 
 ## Run
 
-From `apps/web`: `vp test run src/features/<domain>/<name>.test.tsx`.
-Whole repo: `just test`.
+From `apps/web`: `vp test run --project views src/features/<domain>/<name>.test.tsx`.
+Whole repo: `just test`. Full flows against the backend are journeys
+(`apps/web/e2e`, `just journeys`); keep those few and cover combinations here.

.devcontainer/Dockerfile

@@ -12,14 +12,17 @@ COPY --from=uv /uv /uvx /usr/local/bin/
 COPY --from=just /just /usr/local/bin/
 
 # Tauri 2 Linux prerequisites (https://tauri.app/start/prerequisites/) plus the
-# extra packages from tauri-apps/tauri-action's GitHub workflow.
+# extra packages from tauri-apps/tauri-action's GitHub workflow. libnss3 and
+# libnspr4 are the only libraries Playwright's Chromium needs beyond them.
 RUN apt-get update \
   && apt-get install -y --no-install-recommends \
     build-essential \
     curl \
     file \
     libayatana-appindicator3-dev \
     libgtk-3-dev \
+    libnspr4 \
+    libnss3 \
     librsvg2-dev \
     libssl-dev \
     libwebkit2gtk-4.1-dev \

.devcontainer/post-create.sh

@@ -15,6 +15,10 @@ sudo chown -R "$(id -u):$(id -g)" "${persistent_dirs[@]}"
 
 VP_GIT_HOOKS=0 vp install --frozen-lockfile
 
+# Browser-mode tests (`vp test`) need the Chromium build matching the locked
+# Playwright provider; its system libraries come from the Dockerfile.
+vp run test:browsers:install
+
 # Interactive `claude` in bash and zsh uses the repository harness (Headroom,
 # Caveman, MCP). An alias, not a PATH shim, so the harness still finds the real
 # binary. Home is rebuilt with the container, so this reruns on every rebuild;

.github/CONTRIBUTING.md

@@ -56,20 +56,25 @@ just desktop
 
 ## Checks
 
-Run what CI runs before opening a pull request:
+Run the CI gate before opening a pull request:
 
 ```bash
 just check
 ```
 
+`just check` covers audit, typecheck, lint, unit tests, React Doctor, and
+Fallow. CI also runs the browser journeys (`just journeys`, below) and the
+showcase recording (`bun run --cwd tooling/showcase test:browser`), and builds
+the desktop app on each platform; `just check` runs none of those.
+
 Smaller loops:
 
-| Command          | What it does                       |
-| ---------------- | ---------------------------------- |
-| `just lint`      | Conventions, syncpack, `vp check`  |
-| `just typecheck` | TypeScript                         |
-| `just test`      | Unit tests                         |
-| `just e2e-local` | Playwright against Vite on `:5173` |
+| Command          | What it does                                                                  |
+| ---------------- | ----------------------------------------------------------------------------- |
+| `just lint`      | Conventions, Claude harness, syncpack, `vp check`                             |
+| `just typecheck` | TypeScript                                                                    |
+| `just test`      | Unit and view tests                                                           |
+| `just journeys`  | Browser journeys against the running local backend (start it with `just dev`) |
 
 ## Search the code graph
 
@@ -85,8 +90,6 @@ just graphify-query "trip access planner" # matches identifiers, not prose
 
 Run `just graphify-refresh` after changing source files. The generated
 `graphify-out/` directory is intentionally local and ignored by Git.
-
-
 Graphify and its isolated Python environment are part of the container image;
 nothing is installed into the host or the Bun workspace.
 
@@ -95,7 +98,7 @@ Do not hand-edit `apps/web/src/routeTree.gen.ts`. Add routes as files under
 
 ## Conventions
 
-- Import through package aliases (`@groam/ui`, `@groam/ai-contracts`, `@/features/…`),
+- Import through package aliases (`@groam/ui`, `@groam/app-actions`, `@/features/…`),
   never parent `../` paths across packages
 - Semantic color tokens only (no hex / rgb / Tailwind gradients in features)
 - Public Convex functions live under `packages/backend/convex/routes` as

.github/actions/setup-chromium/action.yml

@@ -0,0 +1,17 @@
+name: Setup Chromium
+description: Install the Chromium build that Vite+ browser-mode tests drive, cached by lockfile
+
+runs:
+  using: composite
+  steps:
+    - name: Cache Chromium
+      uses: actions/cache@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
+      with:
+        path: ~/.cache/ms-playwright
+        key: chromium-${{ runner.os }}-${{ hashFiles('bun.lock') }}
+        restore-keys: |
+          chromium-${{ runner.os }}-
+
+    - name: Install Chromium
+      shell: bash
+      run: bun run test:browsers:install --with-deps

.github/actions/setup-workspace/action.yml

@@ -15,21 +15,12 @@ runs:
         bun-version: '1.3.11'
 
     - name: Cache Bun downloads
-      uses: actions/cache@caa296126883cff596d87d8935842f9db880ef25 # v5
+      uses: actions/cache@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
       with:
         path: ~/.bun/install/cache
         key: bun-${{ runner.os }}-${{ hashFiles('bun.lock') }}
         restore-keys: bun-${{ runner.os }}-
 
-    - name: Cache Turbo
-      uses: actions/cache@caa296126883cff596d87d8935842f9db880ef25 # v5
-      with:
-        path: .turbo
-        key: turbo-${{ runner.os }}-${{ hashFiles('bun.lock') }}-${{ github.sha }}
-        restore-keys: |
-          turbo-${{ runner.os }}-${{ hashFiles('bun.lock') }}-
-          turbo-${{ runner.os }}-
-
     - name: Install dependencies
       shell: bash
       env:

.github/workflows/bootstrap.yml

@@ -14,12 +14,11 @@ env:
   GIT_CONFIG_COUNT: '1'
   GIT_CONFIG_KEY_0: init.defaultBranch
   GIT_CONFIG_VALUE_0: main
-  TURBO_TELEMETRY_DISABLED: '1'
 
 jobs:
   provision:
     name: Create Convex project and environments
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-24.04
     timeout-minutes: 10
     steps:
       - name: Checkout
@@ -55,7 +54,7 @@ jobs:
   deploy:
     name: Bring ${{ matrix.environment }} online
     needs: [provision, test]
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-24.04
     timeout-minutes: 25
     strategy:
       fail-fast: false

.github/workflows/codeql.yml

@@ -20,7 +20,7 @@ permissions:
 jobs:
   analyze:
     name: Analyze (${{ matrix.language }})
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-24.04
     timeout-minutes: 30
     strategy:
       fail-fast: false

.github/workflows/deploy-site-reusable.yml

@@ -22,12 +22,11 @@ env:
   GIT_CONFIG_COUNT: '1'
   GIT_CONFIG_KEY_0: init.defaultBranch
   GIT_CONFIG_VALUE_0: main
-  TURBO_TELEMETRY_DISABLED: '1'
 
 jobs:
   deploy:
     name: Deploy ${{ inputs.site }} to ${{ inputs.environment }}
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-24.04
     timeout-minutes: 15
     environment:
       name: ${{ inputs.environment }}

.github/workflows/deploy.yml

@@ -23,7 +23,6 @@ env:
   GIT_CONFIG_COUNT: '1'
   GIT_CONFIG_KEY_0: init.defaultBranch
   GIT_CONFIG_VALUE_0: main
-  TURBO_TELEMETRY_DISABLED: '1'
 
 jobs:
   test:
@@ -33,7 +32,7 @@ jobs:
   backend:
     name: Deploy backend and migrations to ${{ inputs.environment }}
     needs: [test]
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-24.04
     timeout-minutes: 15
     environment:
       name: ${{ inputs.environment }}
@@ -98,7 +97,7 @@ jobs:
   smoke-test:
     name: Smoke test ${{ inputs.environment }}
     needs: [backend, sites]
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-24.04
     timeout-minutes: 5
     environment:
       name: ${{ inputs.environment }}

.github/workflows/migrate.yml

@@ -23,12 +23,11 @@ env:
   GIT_CONFIG_COUNT: '1'
   GIT_CONFIG_KEY_0: init.defaultBranch
   GIT_CONFIG_VALUE_0: main
-  TURBO_TELEMETRY_DISABLED: '1'
 
 jobs:
   migrate:
     name: Run ${{ inputs.environment }} migrations
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-24.04
     timeout-minutes: 10
     environment:
       name: ${{ inputs.environment }}