diff --git a/src/specify_cli/workflows/steps/switch/__init__.py b/src/specify_cli/workflows/steps/switch/__init__.py
index 8a2e4b343e..ff597f9069 100644
--- a/src/specify_cli/workflows/steps/switch/__init__.py
+++ b/src/specify_cli/workflows/steps/switch/__init__.py
@@ -5,7 +5,11 @@
 from typing import Any
 
 from specify_cli.workflows.base import StepBase, StepContext, StepResult, StepStatus
-from specify_cli.workflows.expressions import evaluate_expression
+from specify_cli.workflows.expressions import (
+    condition_has_malformed_expression_block,
+    condition_is_never_evaluated,
+    evaluate_expression,
+)
 
 
 class SwitchStep(StepBase):
@@ -107,6 +111,31 @@ def validate(self, config: dict[str, Any]) -> list[str]:
                 f"Switch step {config.get('id', '?')!r} is missing "
                 f"'expression' field."
             )
+        # Presence is not enough. `expression` goes through the same
+        # `evaluate_expression` as a condition, so one written without braces comes
+        # back as its own source text: `expression: inputs.mode` matches no case key,
+        # falls through to `default` on every run -- or dispatches nothing at all when
+        # there is no default -- and still reports COMPLETED. `if`, `while` and
+        # `do-while` already reject that shape on their `condition`; this is the same
+        # fault on the same evaluator, one step type over.
+        #
+        # Only these two checks apply. A switch matches on strings, so a composite key
+        # such as `{{ inputs.a }}-{{ inputs.b }}` is legitimate here even though the
+        # same shape would be a fault in a boolean condition.
+        elif condition_is_never_evaluated(config["expression"]):
+            errors.append(
+                f"Switch step {config.get('id', '?')!r}: 'expression' "
+                f"{config['expression']!r} has no usable '{{ }}' block, so it is "
+                "never evaluated: the literal text is matched against the case keys, "
+                "which falls through to 'default' on every run."
+            )
+        elif condition_has_malformed_expression_block(config["expression"]):
+            errors.append(
+                f"Switch step {config.get('id', '?')!r}: 'expression' "
+                f"{config['expression']!r} opens a '{{' the interpolator cannot "
+                "close, so it falls back to the first raw '}}' and matches on a "
+                "truncated expression instead of the one written."
+            )
         # Every other control-flow step requires its branch payload: ``if``
         # requires ``then``, ``fan-out`` requires ``items`` and ``step``,
         # ``fan-in`` a non-empty ``wait_for``, ``gate`` a ``message``. Without
diff --git a/tests/test_workflows.py b/tests/test_workflows.py
index d599f3c6a4..c8325b4eed 100644
--- a/tests/test_workflows.py
+++ b/tests/test_workflows.py
@@ -3403,6 +3403,61 @@ def test_validate_invalid_cases_and_default(self):
         assert any("case 'a' must be a list" in e for e in errors)
         assert any("'default' must be a list" in e for e in errors)
 
+    def test_expression_without_a_block_is_rejected(self):
+        """`expression: inputs.mode` matches its own source text, not the input.
+
+        `evaluate_expression` only substitutes `{{ ... }}`, so the braceless form comes
+        back unchanged, matches no case key, and falls through to `default` on every
+        run while still reporting COMPLETED.
+        """
+        from specify_cli.workflows.steps.switch import SwitchStep
+        from specify_cli.workflows.base import StepContext, StepStatus
+
+        config = {
+            "id": "route",
+            "expression": "inputs.mode",
+            "cases": {"review": [{"id": "r", "type": "command", "command": "echo"}]},
+            "default": [{"id": "d", "type": "command", "command": "echo"}],
+        }
+
+        # Ground truth first: this is what the step does with it today.
+        result = SwitchStep().execute(config, StepContext(inputs={"mode": "review"}))
+        assert result.status == StepStatus.COMPLETED
+        assert result.output["matched_case"] == "__default__"
+        assert result.output["expression_value"] == "inputs.mode"
+
+        errors = [e for e in SwitchStep().validate(config) if "'expression'" in e]
+        assert len(errors) == 1
+        assert "never evaluated" in errors[0]
+
+    def test_expression_with_an_unclosable_block_is_rejected(self):
+        """Different fault, different message: the block is evaluated, but truncated."""
+        from specify_cli.workflows.steps.switch import SwitchStep
+
+        cases = {"review": [{"id": "r", "type": "command", "command": "echo"}]}
+        for expression in ("{{ inputs.x", "{{ inputs.missing | default('oops }}"):
+            config = {"id": "route", "expression": expression, "cases": cases}
+            errors = [
+                e for e in SwitchStep().validate(config) if "'expression'" in e
+            ]
+            assert len(errors) == 1, expression
+
+    def test_a_composite_key_expression_stays_accepted(self):
+        """A switch matches on strings, so more than one block is legitimate here.
+
+        This is the boundary that keeps the two condition predicates safe to reuse on
+        a non-boolean field: `{{ a }}-{{ b }}` is a composite case key, not a fault.
+        A literal `true` and the empty string are likewise ordinary case keys.
+        """
+        from specify_cli.workflows.steps.switch import SwitchStep
+
+        cases = {"a-b": [{"id": "r", "type": "command", "command": "echo"}]}
+        for expression in ("{{ inputs.a }}-{{ inputs.b }}", "{{ inputs.mode }}", "true", ""):
+            config = {"id": "route", "expression": expression, "cases": cases}
+            assert [
+                e for e in SwitchStep().validate(config) if "'expression'" in e
+            ] == [], expression
+
 
 class TestWhileStep:
     """Test the while loop step type."""