Merge upstream main (afc3d9d34c): refresh before review

One conflict: upstream 6e7c7c7da9 replaced bot-mode-closed-chat-stays-closed.spec.ts with bot-mode-row-click-mirrors-registry.spec.ts while our side had rewired its mock-server import. Kept upstream's replacement and rewired the three new specs importing ./mock-server to the consolidated tests-js copy (symbols verified present).

.github/actions/detect-changes/action.yml

@@ -48,6 +48,9 @@ outputs:
   installer:
     description: Run the PowerShell installer tests on a Windows runner.
     value: ${{ steps.classify.outputs.installer }}
+  desktop_updater:
+    description: Run the Windows desktop-update hand-off (windows.ps1) integration tests.
+    value: ${{ steps.classify.outputs.desktop_updater }}
   rust:
     description: Run `cargo test` for the Tauri bootstrap installer.
     value: ${{ steps.classify.outputs.rust }}

.github/workflows/case-collision-check.yml

@@ -0,0 +1,33 @@
+name: Case Collision Check
+
+# Rejects PRs that track two files whose paths differ only by case
+# (README.md vs readme.md, src/Foo.py vs SRC/foo.py).
+#
+# Linux is case-sensitive; Windows and macOS (default) are not. A
+# case-colliding pair lives fine in a Linux checkout and silently breaks
+# every clone on a case-insensitive host — the filesystem can hold only
+# one of them, so checkout fails or whichever wins overwrites the other.
+# Git won't prevent the pair from landing (it only warns at checkout time,
+# on a case-insensitive FS, for the client doing the checkout), so the only
+# enforcement point is CI, on Linux, against the index.
+#
+# Runs unconditionally (no change-classifier gate): a collision can ship in
+# any kind of PR — docs, JS, config, not just Python — so gating on a
+# language lane would be the same "passive rule that cannot enforce a
+# policy" trap the infographic check exists to close.
+
+on:
+  workflow_call:
+
+permissions:
+  contents: read
+
+jobs:
+  check-case-collisions:
+    runs-on: ubuntu-latest
+    timeout-minutes: 5
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+
+      - name: Run case-collision checker
+        run: python3 scripts/check-case-collisions.py

.github/workflows/ci.yaml

@@ -49,6 +49,7 @@ jobs:
       uv_lock: ${{ steps.classify.outputs.uv_lock }}
       npm_lock: ${{ steps.classify.outputs.npm_lock }}
       installer: ${{ steps.classify.outputs.installer }}
+      desktop_updater: ${{ steps.classify.outputs.desktop_updater }}
       rust: ${{ steps.classify.outputs.rust }}
       docker_meta: ${{ steps.classify.outputs.docker_meta }}
       mcp_catalog: ${{ steps.classify.outputs.mcp_catalog }}
@@ -84,6 +85,11 @@ jobs:
     needs: detect
     if: needs.detect.outputs.python == 'true'
     uses: ./.github/workflows/tests-os.yml
+    with:
+      # The Windows lane spawns the real desktop-update hand-off script
+      # (tests/test_desktop_update_windows_*.py) only when that surface
+      # changed; unit-level windows_only tests always run.
+      desktop_updater: ${{ needs.detect.outputs.desktop_updater == 'true' }}
 
   lint:
     name: Python lints
@@ -122,14 +128,14 @@ jobs:
     # Tests-only PRs (~17% of commits) skip this 5-minute job — the longest
     # single job in the workflow — while still running the full pytest lanes.
     #
-    # ⛔ TEMPORARILY DISABLED (Aug 2, 2026, Teknium) — the suite is red on
-    # every PR and on main itself since the Aug 1 night engines/npm churn
-    # (#76499 → #76562 → #76575): the mock-backend Electron window never
-    # gets a title, so boot/chat/setup/interim specs all fail identically
-    # regardless of the PR's diff (verified on #76573 and the docs-only
-    # #76582). Tracking issue: #76627 (assigned: Ari). To re-enable,
-    # delete the `false &&` below — nothing else changed.
-    if: ${{ false && (needs.detect.outputs.python_prod == 'true' || needs.detect.outputs.frontend == 'true') }}
+    # Re-disabled (Sep 2026): the Sep 1 re-enable is still incredibly flaky.
+    # Keep this a bare `if: false`. The earlier
+    # `${{ false && (... || ...) }}` form on this reusable-workflow job made
+    # GitHub's workflow parser fail at startup ("An unexpected error has
+    # occurred") — every ci.yaml run repo-wide dispatched 0 jobs from
+    # 24f5a60ed1 until this line changed. To re-enable, restore:
+    #   if: ${{ needs.detect.outputs.python_prod == 'true' || needs.detect.outputs.frontend == 'true' }}
+    if: false
     uses: ./.github/workflows/e2e-desktop.yml
 
   docs-site:
@@ -171,6 +177,11 @@ jobs:
     needs: detect
     uses: ./.github/workflows/profile-artifact-check.yml
 
+  case-collision-check:
+    name: Check no case-colliding filenames
+    needs: detect
+    uses: ./.github/workflows/case-collision-check.yml
+
   lockfile-diff:
     name: package-lock.json diff
     needs: detect
@@ -233,6 +244,7 @@ jobs:
       - history-check
       - contributor-check
       - uv-lockfile
+      - case-collision-check
       - lockfile-diff
       - docker-lint
       - profile-artifact-check

.github/workflows/tests-os.yml

@@ -27,6 +27,19 @@ name: OS-specific tests
 
 on:
   workflow_call:
+    inputs:
+      desktop_updater:
+        description: >-
+          Run the Windows desktop-update hand-off integration tests
+          (tests/test_desktop_update_windows_*.py). These spawn the real
+          scripts/desktop-update/windows.ps1 and poll its loopback server, so
+          they carry process-timing noise a shared runner amplifies; the
+          caller gates them on the classifier's desktop_updater lane so a PR
+          that never touched that surface cannot be failed by it. Push /
+          dispatch runs fail open (classifier sets every lane true).
+        type: boolean
+        required: false
+        default: true
 
 permissions:
   contents: read
@@ -134,9 +147,23 @@ jobs:
           # would therefore abort the script on any non-zero exit and the
           # exit-5 branch below would be unreachable dead code — the job
           # would still fail red, but the diagnostic would never print.
+          # Desktop-update hand-off integration tests spawn the real
+          # windows.ps1; deselect them unless the PR touched that surface
+          # (see the workflow_call input). ``--ignore-glob`` keeps the file
+          # list above intact, so a renamed test file still trips the
+          # zero-tests guard rather than silently vanishing.
+          # (bash 3.2 on the macOS runner: an empty array under ``set -u`` is
+          # an unbound-variable error, hence the ``${arr[@]+...}`` idiom.)
+          EXTRA_ARGS=()
+          if [ "${{ inputs.desktop_updater }}" != "true" ]; then
+            echo "desktop_updater lane off: skipping tests/test_desktop_update_windows_*.py"
+            EXTRA_ARGS+=(--ignore-glob='*test_desktop_update_windows_*.py')
+          fi
+
           status=0
           uv run --no-sync python -m pytest \
             "$@" \
+            ${EXTRA_ARGS[@]+"${EXTRA_ARGS[@]}"} \
             -m "${{ matrix.marker }} and not integration" \
             -v --tb=short || status=$?
           if [ "$status" -eq 5 ]; then