diff --git a/.github/workflows/full-test-suite.yml b/.github/workflows/full-test-suite.yml
index 29f8f9a41..22772cc09 100644
--- a/.github/workflows/full-test-suite.yml
+++ b/.github/workflows/full-test-suite.yml
@@ -52,12 +52,13 @@ jobs:
           curl -LsSf https://astral.sh/uv/0.11.19/install.sh | sh
           uv venv --python 3.12
           source .venv/bin/activate
-          uv sync --extra dev
+          uv sync --extra dev --extra sandbox
 
       - name: Test
         run: |
           source .venv/bin/activate
-          ng_dev_test
+          PYTEST_ADDOPTS='-m "not sandbox" --cov-report= --cov-fail-under=0' ng_dev_test
+          pytest --cov=. --cov-append --cov-report=term-missing --durations=10 -m sandbox
           ng_test_all +fail_on_total_and_test_mismatch=true +delete_venvs_after_each_test=true
 
   test-wheel-install:
diff --git a/.github/workflows/unit-tests.yml b/.github/workflows/unit-tests.yml
index 810f75fcc..ace2e3907 100644
--- a/.github/workflows/unit-tests.yml
+++ b/.github/workflows/unit-tests.yml
@@ -166,8 +166,11 @@ jobs:
           # Full suite: core library unit tests run here once; all server tests run in the
           # parallel `server-suite` matrix.
           if [[ "${{ needs.detect.outputs.run_full }}" == "true" ]]; then
-            echo "Running core library unit tests (server suite runs in the sharded matrix)"
-            ng_dev_test
+            echo "Running core library unit tests without sandbox marker (server suite runs in the sharded matrix)"
+            PYTEST_ADDOPTS='-m "not sandbox" --cov-report= --cov-fail-under=0' ng_dev_test
+
+            echo "Running sandbox unit tests and appending them to the coverage report"
+            pytest --cov=. --cov-append --cov-report=term-missing --durations=10 -m sandbox
 
           # Server-only: test only the changed servers (typically a small set, no sharding needed)
           elif [[ "${{ needs.detect.outputs.run_servers }}" == "true" ]]; then
diff --git a/nemo_gym/cli/main.py b/nemo_gym/cli/main.py
index 381975e43..9e74374bc 100644
--- a/nemo_gym/cli/main.py
+++ b/nemo_gym/cli/main.py
@@ -43,7 +43,10 @@ class _GymArgumentParser(argparse.ArgumentParser):
     def error(self, message: str) -> None:
         match = re.search(r"invalid choice: '([^']+)' \(choose from (.+)\)", message)
         if match:
-            typo, choices = match.group(1), re.findall(r"'([^']+)'", match.group(2))
+            typo = match.group(1)
+            choices = re.findall(r"'([^']+)'", match.group(2))
+            if not choices:
+                choices = [choice.strip() for choice in match.group(2).split(",")]
             message += _did_you_mean(typo, choices)
         super().error(message)
 
diff --git a/pyproject.toml b/pyproject.toml
index 83caf14ff..61335767e 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -490,6 +490,7 @@ markers = [
     "system: marks test working at the highest integration level (deselect with '-m \"not system\"')",
     "acceptance: marks test checking whether the developed product/model passes the user defined acceptance criteria (deselect with '-m \"not acceptance\"')",
     "docs: mark tests related to documentation (deselect with '-m \"not docs\"')",
+    "sandbox: marks sandbox unit tests (deselect with '-m \"not sandbox\"')",
     "skipduringci: marks tests that are skipped during ci as they are addressed by Jenkins jobs but should be run to test user setups",
     "pleasefixme: marks tests that are broken and need fixing",
 ]
diff --git a/tests/unit_tests/test_apptainer_provider.py b/tests/unit_tests/test_apptainer_provider.py
index fea5f8ae8..94de183a6 100644
--- a/tests/unit_tests/test_apptainer_provider.py
+++ b/tests/unit_tests/test_apptainer_provider.py
@@ -31,6 +31,9 @@
 )
 
 
+pytestmark = pytest.mark.sandbox
+
+
 FAKE_BINARY = "/usr/bin/apptainer"
 
 
diff --git a/tests/unit_tests/test_opensandbox_provider.py b/tests/unit_tests/test_opensandbox_provider.py
index c57be62a6..6e1c175ba 100644
--- a/tests/unit_tests/test_opensandbox_provider.py
+++ b/tests/unit_tests/test_opensandbox_provider.py
@@ -26,6 +26,9 @@
 from nemo_gym.sandbox.providers.base import SandboxResources, SandboxSpec, SandboxStatus
 
 
+pytestmark = pytest.mark.sandbox
+
+
 pytest.importorskip("tenacity", reason="tenacity optional sandbox dependency is not installed")
 
 from nemo_gym.sandbox.providers.opensandbox import provider as opensandbox_provider
diff --git a/tests/unit_tests/test_sandbox.py b/tests/unit_tests/test_sandbox.py
index c26f660b3..44232e599 100644
--- a/tests/unit_tests/test_sandbox.py
+++ b/tests/unit_tests/test_sandbox.py
@@ -42,6 +42,9 @@
 from responses_api_agents.mini_swe_agent_2.sandbox_environment import MiniSWESandboxEnvironment
 
 
+pytestmark = pytest.mark.sandbox
+
+
 def _has_module(module_name: str) -> bool:
     try:
         return importlib.util.find_spec(module_name) is not None