diff --git a/.github/workflows/test-e2e.yaml b/.github/workflows/test-e2e.yaml
index 7fec7c436a..5db817e259 100644
--- a/.github/workflows/test-e2e.yaml
+++ b/.github/workflows/test-e2e.yaml
@@ -36,7 +36,7 @@ jobs:
 
   test-local-android:
     name: Test on Android
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-latest
     if: github.repository == 'mobile-dev-inc/maestro'
     needs: build
     
diff --git a/.github/workflows/update-samples.yaml b/.github/workflows/update-samples.yaml
new file mode 100644
index 0000000000..dbd1445836
--- /dev/null
+++ b/.github/workflows/update-samples.yaml
@@ -0,0 +1,31 @@
+name: Update samples
+
+on:
+  push:
+    branches: [main]
+
+jobs:
+  main:
+    runs-on: ubuntu-latest
+    if: github.repository == 'mobile-dev-inc/maestro'
+
+    steps:
+      - name: Clone repository
+        uses: actions/checkout@v4
+
+      - name: Authenticate to Google Cloud
+        uses: google-github-actions/auth@v2
+        with:
+          # These credentials should only have write access to the bucket
+          credentials_json: ${{ secrets.GCP_MOBILEDEV_BUCKET_CREDENTIALS }}
+
+      - name: Set up Google Cloud CLI
+        uses: google-github-actions/setup-gcloud@v2
+        with:
+          version: '>= 484.0.0'
+          project_id: perf-dev-289002
+    
+      - name: Upload samples to public Google Cloud Storage bucket
+        run: |
+          cd e2e/
+          ./update_samples
diff --git a/e2e/.gitignore b/e2e/.gitignore
index d427af6407..57a77c1db6 100644
--- a/e2e/.gitignore
+++ b/e2e/.gitignore
@@ -1 +1,4 @@
 apps/
+
+samples/
+samples.zip
diff --git a/e2e/README.md b/e2e/README.md
index 5a3feeded3..fc081f3b28 100644
--- a/e2e/README.md
+++ b/e2e/README.md
@@ -2,6 +2,8 @@
 
 This directory contains glue code for testing Maestro itself.
 
+## Testing
+
 Typical workflow is:
 
 1. Start Android emulator and iOS simulator
@@ -9,4 +11,45 @@ Typical workflow is:
 3. `install_apps`
 4. `run_tests`
 
-We try to keep scripts in files, so we don't get too tightly coupled to GitHub Action.
+We try to keep shell code in separate files, so we don't get too tightly coupled
+to GitHub Actions.
+
+### Expected failures
+
+Let's say a critical bug is introduced that causes Maestro to always mark all
+tests as passed. If our e2e test suite only was only checking if all tests pass
+(i.e. `maestro test` exit code is 0), then wouldn't catch such a bug.
+
+To prevent this, all flows in this directory MUST have a `passing` or `failing`
+label, so the correct outcome can be asserted.
+
+## Samples
+
+This directory also contains samples that are downloaded by the `maestro download-samples` command,
+and some glue code to facilitate updating those samples.
+
+`maestro download-samples` provides a set of flows and apps so that users can
+quickly try out Maestro, without having to write any flows for their own app.
+
+`download-samples` downloads these files and apps from our publicly-available
+Google Cloud Storage bucket (hosted on `storage.googleapis.com`).
+
+### Intro
+
+The samples are automatically updated by the GitHub Action on every new commit
+to the `main` branch.
+
+There zip archive that is downloaded by `download-samples` consists of 2 things:
+- the Maestro workspace with flows (located in the `workspaces/wikipedia` directory)
+- the app binary files that are used in the flows (located in the `apps` directory)
+
+App binary files are heavy, so we don't store them in the repository. Instead, they are hosted
+on publicly available directory in Google Cloud Storage:
+
+### Update the samples
+
+Run the script:
+
+```console
+./update_samples
+```
diff --git a/e2e/download_apps b/e2e/download_apps
index 5a7f361cbf..6e419c6da1 100755
--- a/e2e/download_apps
+++ b/e2e/download_apps
@@ -1,5 +1,4 @@
 #!/usr/bin/env sh
-
 set -eu
 
 # Download apps from URLs listed in manifest.txt.
@@ -18,6 +17,5 @@ while read -r url; do
 	extension="${app_file##*.}"
 	if [ "$extension" = "zip" ]; then
 		unzip -qq -o -d ./apps "$app_file" -x "__MACOSX/*"
-		rm "$app_file"
 	fi
 done <manifest.txt
diff --git a/e2e/install_apps b/e2e/install_apps
index a59854844b..9903efc6bd 100755
--- a/e2e/install_apps
+++ b/e2e/install_apps
@@ -1,7 +1,12 @@
 #!/usr/bin/env sh
-
 set -eu
 
+# Install all apps from apps/ directory (that was previously created with
+# download_apps).
+#
+# Matches .apk files to install on Android devices and .app files to install on
+# iOS simulators.
+
 [ "$(basename "$PWD")" = "e2e" ] || { echo "must be run from e2e directory" && exit 1; }
 
 command -v adb >/dev/null 2>&1 || { echo "adb is required" && exit 1; }
@@ -12,10 +17,10 @@ for file in ./apps/*; do
 
 	extension="${file##*.}"
 	if [ "$extension" = "apk" ]; then
-		adb install -r "$file" >/dev/null
+		adb install -r "$file" >/dev/null || echo "adb: could not install $filename"
 	elif [ "$extension" = "app" ] && [ "$(uname)" = "Darwin" ]; then
-		xcrun simctl install booted "$file"
+		xcrun simctl install booted "$file" || echo "xcrun: could not install $filename"
 	else
-		echo "unknown file extension $extension"
+		echo "ignored unsupported file extension $extension"
 	fi
 done
diff --git a/e2e/manifest.txt b/e2e/manifest.txt
index 3609de05ec..24ed5e1e11 100644
--- a/e2e/manifest.txt
+++ b/e2e/manifest.txt
@@ -1,3 +1,3 @@
-https://storage.googleapis.com/mobile.dev/samples/e2e_apps/wikipedia.apk
-https://storage.googleapis.com/mobile.dev/samples/e2e_apps/Wikipedia.zip
-https://storage.googleapis.com/mobile.dev/samples/e2e_apps/nowinandroid.apk
+https://storage.googleapis.com/mobile.dev/cli_e2e/wikipedia.apk
+https://storage.googleapis.com/mobile.dev/cli_e2e/wikipedia.zip
+https://storage.googleapis.com/mobile.dev/cli_e2e/nowinandroid.apk
diff --git a/e2e/run_tests b/e2e/run_tests
index 04d5cbd434..00f07a5706 100755
--- a/e2e/run_tests
+++ b/e2e/run_tests
@@ -1,22 +1,35 @@
 #!/usr/bin/env sh
-
 set -eu
 
+# Runs all tests in the workspaces directory.
+
 command -v maestro >/dev/null 2>&1 || { echo "maestro is required" && exit 1; }
 
 [ "$(basename "$PWD")" = "e2e" ] || { echo "must be run from e2e directory" && exit 1; }
 
-selected_app="${1:-}"
-
 FAILED=false
-for file in ./flows/*; do
-	filename="$(basename "$file")"
 
-	if [ -z "$selected_app" ] || [ "$selected_app" = "$filename" ]; then
-		cd "$file" || exit 1
-		maestro --device emulator-5554 test . || FAILED=true
-		cd - >/dev/null || exit 1
+# Run passing tests
+for workspace_dir in ./workspaces/*; do
+	echo "run passing tests for app $workspace_dir"
+	if [ "$(basename "$workspace_dir")" = "wikipedia" ]; then
+	  # wikipedia flows have no passing/failing labels
+	  maestro test --exclude-tags ios "$workspace_dir" || FAILED=true
+	  continue
+	fi
+
+	maestro test --include-tags passing --exclude-tags ios "$workspace_dir" || FAILED=true
+done
+
+# Run failing tests
+for workspace_dir in ./workspaces/*; do
+	if [ "$(basename "$workspace_dir")" = "wikipedia" ]; then
+	  # wikipedia workspace has no failing flows
+	  continue
 	fi
+	echo "run failing tests for app $workspace_dir"
+
+	maestro test --include-tags failing --exclude-tags ios "$workspace_dir" && FAILED=true
 done
 
 if [ "$FAILED" = true ]; then
diff --git a/e2e/update_samples b/e2e/update_samples
new file mode 100755
index 0000000000..b729aec430
--- /dev/null
+++ b/e2e/update_samples
@@ -0,0 +1,31 @@
+#!/usr/bin/env sh
+set -eu
+
+# Updates the samples that are hosted in mobile.dev's GCS bucket ($SAMPLES_URL).
+# The samples are for use with `maestro download-samples` command.
+
+[ "$(basename "$PWD")" = "e2e" ] || { echo "must be run from e2e directory" && exit 1; }
+
+command -v curl >/dev/null 2>&1 || { echo "curl is required" && exit 1; }
+command -v gsutil >/dev/null 2>&1 || { echo "gsutil is required" && exit 1; }
+
+SAMPLES_URL="gs://mobile.dev/samples/samples.zip"
+
+if [ ! -d apps/ ]; then
+  ./download_apps
+fi
+
+rm -rf samples/ samples.zip
+mkdir -p samples/
+cp -r apps/wikipedia.apk apps/wikipedia.zip samples/
+cp -r workspaces/wikipedia/* samples/
+cp samples/wikipedia.apk samples/sample.apk # The name is being depended upon.
+cp samples/wikipedia.zip samples/sample.zip # The name is being depended upon.
+cd samples/
+
+zip -r -q ../samples.zip . -x "/**/.*" -x "__MACOSX"
+
+cd ..
+
+gsutil cp samples.zip "$SAMPLES_URL"
+gsutil acl ch -r -u AllUsers:R "$SAMPLES_URL"
diff --git a/e2e/flows/nowinandroid/bookmarks.yaml b/e2e/workspaces/nowinandroid/bookmarks_1.yaml
similarity index 92%
rename from e2e/flows/nowinandroid/bookmarks.yaml
rename to e2e/workspaces/nowinandroid/bookmarks_1.yaml
index e0e83929b5..536862066e 100644
--- a/e2e/flows/nowinandroid/bookmarks.yaml
+++ b/e2e/workspaces/nowinandroid/bookmarks_1.yaml
@@ -1,4 +1,6 @@
 appId: com.google.samples.apps.nowinandroid.demo.debug
+tags:
+  - passing
 ---
 - launchApp:
     clearState: true
diff --git a/e2e/workspaces/nowinandroid/bookmarks_2.yaml b/e2e/workspaces/nowinandroid/bookmarks_2.yaml
new file mode 100644
index 0000000000..cd19f5b13a
--- /dev/null
+++ b/e2e/workspaces/nowinandroid/bookmarks_2.yaml
@@ -0,0 +1,8 @@
+appId: com.google.samples.apps.nowinandroid.demo.debug
+tags:
+  - failing
+---
+- launchApp:
+    clearState: true
+- tapOn:
+    id: non-existent-id-to-fail-this-test
diff --git a/samples/android-advanced-flow.yaml b/e2e/workspaces/wikipedia/android-advanced-flow.yaml
similarity index 95%
rename from samples/android-advanced-flow.yaml
rename to e2e/workspaces/wikipedia/android-advanced-flow.yaml
index 31805bd56e..62c91abd4d 100644
--- a/samples/android-advanced-flow.yaml
+++ b/e2e/workspaces/wikipedia/android-advanced-flow.yaml
@@ -1,5 +1,6 @@
 appId: org.wikipedia
 tags:
+  - android
   - advanced
 ---
 - runFlow: subflows/onboarding-android.yaml
diff --git a/samples/android-flow.yaml b/e2e/workspaces/wikipedia/android-flow.yaml
similarity index 67%
rename from samples/android-flow.yaml
rename to e2e/workspaces/wikipedia/android-flow.yaml
index 1f85a94ef7..6f28026579 100644
--- a/samples/android-flow.yaml
+++ b/e2e/workspaces/wikipedia/android-flow.yaml
@@ -1,3 +1,5 @@
 appId: org.wikipedia
+tags:
+  - android
 ---
 - launchApp
diff --git a/samples/ios-advanced-flow.yaml b/e2e/workspaces/wikipedia/ios-advanced-flow.yaml
similarity index 98%
rename from samples/ios-advanced-flow.yaml
rename to e2e/workspaces/wikipedia/ios-advanced-flow.yaml
index 68ca89b516..fd5551a4be 100644
--- a/samples/ios-advanced-flow.yaml
+++ b/e2e/workspaces/wikipedia/ios-advanced-flow.yaml
@@ -1,5 +1,6 @@
 appId: org.wikimedia.wikipedia
 tags:
+  - ios
   - advanced
 ---
 - runFlow: subflows/onboarding-ios.yaml
@@ -18,4 +19,3 @@ tags:
 - eraseText
 - inputText: qwerty
 - assertVisible: ${output.result}
-
diff --git a/samples/ios-flow.yaml b/e2e/workspaces/wikipedia/ios-flow.yaml
similarity index 77%
rename from samples/ios-flow.yaml
rename to e2e/workspaces/wikipedia/ios-flow.yaml
index fbc91ad300..917b5b4b16 100644
--- a/samples/ios-flow.yaml
+++ b/e2e/workspaces/wikipedia/ios-flow.yaml
@@ -1,3 +1,5 @@
 appId: org.wikimedia.wikipedia
+tags:
+  - ios
 ---
 - launchApp
diff --git a/samples/scripts/getSearchQuery.js b/e2e/workspaces/wikipedia/scripts/getSearchQuery.js
similarity index 100%
rename from samples/scripts/getSearchQuery.js
rename to e2e/workspaces/wikipedia/scripts/getSearchQuery.js
diff --git a/samples/subflows/onboarding-android.yaml b/e2e/workspaces/wikipedia/subflows/onboarding-android.yaml
similarity index 100%
rename from samples/subflows/onboarding-android.yaml
rename to e2e/workspaces/wikipedia/subflows/onboarding-android.yaml
diff --git a/samples/subflows/onboarding-ios.yaml b/e2e/workspaces/wikipedia/subflows/onboarding-ios.yaml
similarity index 100%
rename from samples/subflows/onboarding-ios.yaml
rename to e2e/workspaces/wikipedia/subflows/onboarding-ios.yaml
diff --git a/samples/wikipedia-android-advanced/auth/login.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/auth/login.yml
similarity index 100%
rename from samples/wikipedia-android-advanced/auth/login.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/auth/login.yml
diff --git a/samples/wikipedia-android-advanced/auth/signup.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/auth/signup.yml
similarity index 100%
rename from samples/wikipedia-android-advanced/auth/signup.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/auth/signup.yml
diff --git a/samples/wikipedia-android-advanced/dashboard/copy-paste.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/copy-paste.yml
similarity index 100%
rename from samples/wikipedia-android-advanced/dashboard/copy-paste.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/copy-paste.yml
diff --git a/samples/wikipedia-android-advanced/dashboard/feed.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/feed.yml
similarity index 100%
rename from samples/wikipedia-android-advanced/dashboard/feed.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/feed.yml
diff --git a/samples/wikipedia-android-advanced/dashboard/main.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/main.yml
similarity index 100%
rename from samples/wikipedia-android-advanced/dashboard/main.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/main.yml
diff --git a/samples/wikipedia-android-advanced/dashboard/saved.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/saved.yml
similarity index 67%
rename from samples/wikipedia-android-advanced/dashboard/saved.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/saved.yml
index 96e9163b59..be21915c06 100644
--- a/samples/wikipedia-android-advanced/dashboard/saved.yml
+++ b/e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/saved.yml
@@ -3,5 +3,5 @@ appId: org.wikipedia
 - tapOn: "Saved"
 - tapOn: "Default list for your saved articles"
 - assertVisible: "Sun"
-- assertVisible: "Star in the Solar System"
+- assertVisible: "Star at the center of the Solar System"
 - back
diff --git a/samples/wikipedia-android-advanced/dashboard/search.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/search.yml
similarity index 74%
rename from samples/wikipedia-android-advanced/dashboard/search.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/search.yml
index e505b2d07d..5a79e8a68d 100644
--- a/samples/wikipedia-android-advanced/dashboard/search.yml
+++ b/e2e/workspaces/wikipedia/wikipedia-android-advanced/dashboard/search.yml
@@ -2,7 +2,7 @@ appId: org.wikipedia
 ---
 - tapOn: "Search Wikipedia"
 - inputText: "Sun"
-- assertVisible: "Star in the Solar System"
+- assertVisible: "Star at the center of the Solar System"
 - tapOn:
     id: ".*page_list_item_title"
 - tapOn:
diff --git a/samples/wikipedia-android-advanced/onboarding/add-language.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/onboarding/add-language.yml
similarity index 89%
rename from samples/wikipedia-android-advanced/onboarding/add-language.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/onboarding/add-language.yml
index 3d3a935322..bd498d6017 100644
--- a/samples/wikipedia-android-advanced/onboarding/add-language.yml
+++ b/e2e/workspaces/wikipedia/wikipedia-android-advanced/onboarding/add-language.yml
@@ -7,4 +7,4 @@ appId: org.wikipedia
 - inputText: "Greek"
 - assertVisible: "Ελληνικά"
 - tapOn: "Ελληνικά"
-- tapOn: "Back"
+- tapOn: "Navigate up"
diff --git a/samples/wikipedia-android-advanced/onboarding/main.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/onboarding/main.yml
similarity index 100%
rename from samples/wikipedia-android-advanced/onboarding/main.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/onboarding/main.yml
diff --git a/samples/wikipedia-android-advanced/onboarding/remove-language.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/onboarding/remove-language.yml
similarity index 91%
rename from samples/wikipedia-android-advanced/onboarding/remove-language.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/onboarding/remove-language.yml
index 62a72458ff..9dfe1c1045 100644
--- a/samples/wikipedia-android-advanced/onboarding/remove-language.yml
+++ b/e2e/workspaces/wikipedia/wikipedia-android-advanced/onboarding/remove-language.yml
@@ -10,4 +10,4 @@ appId: org.wikipedia
     id: ".*menu_delete_selected"
 - tapOn: "OK"
 - assertNotVisible: "Ελληνικά"
-- tapOn: "Back"
+- tapOn: "Navigate up"
diff --git a/samples/wikipedia-android-advanced/run-test.yml b/e2e/workspaces/wikipedia/wikipedia-android-advanced/run-test.yml
similarity index 90%
rename from samples/wikipedia-android-advanced/run-test.yml
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/run-test.yml
index b55901b797..bae39105ed 100644
--- a/samples/wikipedia-android-advanced/run-test.yml
+++ b/e2e/workspaces/wikipedia/wikipedia-android-advanced/run-test.yml
@@ -1,4 +1,6 @@
 appId: org.wikipedia
+tags:
+  - android
 ---
 - launchApp:
     clearState: true
diff --git a/samples/wikipedia-android-advanced/scripts/fetchTestUser.js b/e2e/workspaces/wikipedia/wikipedia-android-advanced/scripts/fetchTestUser.js
similarity index 100%
rename from samples/wikipedia-android-advanced/scripts/fetchTestUser.js
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/scripts/fetchTestUser.js
diff --git a/samples/wikipedia-android-advanced/scripts/generateCredentials.js b/e2e/workspaces/wikipedia/wikipedia-android-advanced/scripts/generateCredentials.js
similarity index 100%
rename from samples/wikipedia-android-advanced/scripts/generateCredentials.js
rename to e2e/workspaces/wikipedia/wikipedia-android-advanced/scripts/generateCredentials.js
diff --git a/samples/Readme.md b/samples/Readme.md
deleted file mode 100644
index 709a860151..0000000000
--- a/samples/Readme.md
+++ /dev/null
@@ -1,18 +0,0 @@
-# Maestro samples
-
-`maestro download-samples` provides a set of flows and apps so that users can quickly set up a maestro test, without having to create an app.
-
-download-samples downloads these files and apps from storage.googleapis.com.
-
-# Update the samples
-
-Although the samples are checked in, updating them requires a few manual steps:
-
-* Change the samples in this directory and merge these changes
-* Run `maestro download-samples`
-* Copy *.yaml to the samples directory created by download-samples
-* Run `(cd samples && zip -r "$OLDPWD/samples.zip" . -x "/**/.*" -x "__MACOSX")`
-* Open https://console.cloud.google.com/storage/browser/mobile.dev/samples
-* Upload samples.zip
-* Adjust the permissions of samples.zip to "Public to Internet"
-* Run `maestro download-samples` and verify that the change was successful