diff --git a/sdk/wren-pydantic/src/wren_pydantic/_tools.py b/sdk/wren-pydantic/src/wren_pydantic/_tools.py
index dd49861bda..2599814a8d 100644
--- a/sdk/wren-pydantic/src/wren_pydantic/_tools.py
+++ b/sdk/wren-pydantic/src/wren_pydantic/_tools.py
@@ -169,11 +169,26 @@ def _run_list_models(toolkit: WrenToolkit) -> list[ModelSummary]:
         raise to_model_retry(exc) from exc
 
     models = manifest.get("models") or []
-    return [
-        ModelSummary(
-            name=m["name"],
-            column_count=len(m.get("columns") or []),
-            description=(m.get("properties") or {}).get("description"),
+    # Skip non-dict rows and tolerate missing name/columns/properties so a
+    # single corrupted model entry cannot KeyError/TypeError the tool.
+    summaries: list[ModelSummary] = []
+    for m in models:
+        if not isinstance(m, dict):
+            continue
+        name = m.get("name")
+        if not name:
+            continue
+        columns = m.get("columns") or []
+        if not isinstance(columns, list):
+            columns = []
+        props = m.get("properties") or {}
+        if not isinstance(props, dict):
+            props = {}
+        summaries.append(
+            ModelSummary(
+                name=name,
+                column_count=len(columns),
+                description=props.get("description"),
+            )
         )
-        for m in models
-    ]
+    return summaries
diff --git a/sdk/wren-pydantic/tests/unit/test_run_list_models_guard.py b/sdk/wren-pydantic/tests/unit/test_run_list_models_guard.py
new file mode 100644
index 0000000000..526714a7a7
--- /dev/null
+++ b/sdk/wren-pydantic/tests/unit/test_run_list_models_guard.py
@@ -0,0 +1,25 @@
+"""_run_list_models skips non-dict / incomplete model rows."""
+
+from __future__ import annotations
+
+from unittest.mock import MagicMock
+
+from wren_pydantic._tools import _run_list_models
+
+
+def test_skips_bad_models():
+    toolkit = MagicMock()
+    toolkit._mdl_source.load_manifest.return_value = {
+        "models": [
+            None,
+            "x",
+            {"columns": []},  # no name
+            {"name": "ok", "columns": None, "properties": "nope"},
+            {"name": "t", "columns": [{"n": 1}], "properties": {"description": "d"}},
+        ]
+    }
+    out = _run_list_models(toolkit)
+    assert [m.name for m in out] == ["ok", "t"]
+    assert out[0].column_count == 0
+    assert out[1].column_count == 1
+    assert out[1].description == "d"