diff --git a/.github/workflows/request_test_data.yml b/.github/workflows/request_test_data.yml
index 318f681..f8f62ea 100644
--- a/.github/workflows/request_test_data.yml
+++ b/.github/workflows/request_test_data.yml
@@ -7,7 +7,8 @@ name: Request Test Data
 
 on:
   pull_request_target:
-    types: [opened, synchronize, reopened]
+    # labeled: a maintainer applying the bot-test label fires the flow at once
+    types: [opened, synchronize, reopened, labeled]
     paths:
       - 'scripts/artifacts/**'
 
diff --git a/admin/scripts/check_pr_test_data.py b/admin/scripts/check_pr_test_data.py
index f0e242e..f35e3f5 100644
--- a/admin/scripts/check_pr_test_data.py
+++ b/admin/scripts/check_pr_test_data.py
@@ -33,6 +33,7 @@
 REPO_ROOT = Path(__file__).resolve().parents[2]
 MANIFEST_PATH = REPO_ROOT / "admin" / "image_manifest.json"
 MARKER = "<!-- leapp-test-data-check -->"
+TEST_LABEL = "bot-test"
 LABEL_ASK = "needs-test-data"
 LABEL_FIXTURE = "fixture-needed"
 LABELS = {
@@ -220,6 +221,22 @@ def desired_labels(fixture, ask):
     return labels
 
 
+def should_skip(login, permission, label_names):
+    """Returns (skip, reason) for authors the bot would not normally address.
+
+    A maintainer-applied bot-test label overrides the skip so the whole flow
+    can be exercised end to end on any PR. Applying labels needs triage
+    access, so contributors cannot trigger that on themselves.
+    """
+    if TEST_LABEL in label_names:
+        return False, f"'{TEST_LABEL}' label present; treating the author as external"
+    if login.endswith("[bot]"):
+        return True, f"author {login} is a bot"
+    if permission in ("admin", "write"):
+        return True, f"author {login} has {permission} access"
+    return False, f"author {login} has {permission} access"
+
+
 def find_marker_comment(token, repo, pr_number):
     for comment in paginate(token, f"{API}/repos/{repo}/issues/{pr_number}/comments"):
         if MARKER in comment.get("body", ""):
@@ -270,12 +287,12 @@ def main():
 
     pr = api_request(token, f"{API}/repos/{repo}/pulls/{pr_number}")
     login = pr["user"]["login"]
-    if login.endswith("[bot]"):
-        print(f"Author {login} is a bot; skipping.")
-        return
+    label_names = {l["name"] for l in pr.get("labels", [])}
     permission = author_permission(token, repo, login)
-    if permission in ("admin", "write"):
-        print(f"Author {login} has {permission} access; skipping.")
+    skip, reason = should_skip(login, permission, label_names)
+    print(reason)
+    if skip:
+        print("Skipping.")
         if not dry_run:
             return
         print("DRY_RUN: continuing anyway to show the decision.")
diff --git a/admin/test/scripts/test_check_pr_test_data.py b/admin/test/scripts/test_check_pr_test_data.py
index 2d6abdd..31d47c9 100644
--- a/admin/test/scripts/test_check_pr_test_data.py
+++ b/admin/test/scripts/test_check_pr_test_data.py
@@ -131,6 +131,23 @@ def test_no_em_dashes_anywhere(self):
             self.assertNotIn('—', body)
 
 
+class ShouldSkipTests(unittest.TestCase):
+    def test_write_and_admin_authors_skip(self):
+        self.assertTrue(bot.should_skip('dev', 'write', set())[0])
+        self.assertTrue(bot.should_skip('dev', 'admin', set())[0])
+
+    def test_bot_authors_skip(self):
+        self.assertTrue(bot.should_skip('dependabot[bot]', 'none', set())[0])
+
+    def test_external_authors_do_not_skip(self):
+        self.assertFalse(bot.should_skip('someone', 'read', set())[0])
+        self.assertFalse(bot.should_skip('someone', 'none', set())[0])
+
+    def test_bot_test_label_overrides_every_skip(self):
+        self.assertFalse(bot.should_skip('dev', 'admin', {bot.TEST_LABEL})[0])
+        self.assertFalse(bot.should_skip('dependabot[bot]', 'none', {bot.TEST_LABEL})[0])
+
+
 class DesiredLabelsTests(unittest.TestCase):
     def test_matrix(self):
         self.assertEqual(bot.desired_labels({}, []), set())