diff --git a/CHANGELOG.md b/CHANGELOG.md
index d0ee1edda5..74da8e63fd 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -37,6 +37,12 @@ and this project adheres to
 
 ### Fixed
 
+- `step:complete` no longer crashes with an unhandled `Postgrex.Error` (code
+  `22P05`) when an `output_dataclip` body contains a NUL byte. The dataclip
+  insert is rejected with a changeset error instead of raising, since
+  PostgreSQL cannot store the NUL byte (`\u0000`) in `text`/`jsonb`.
+  [#4893](https://github.com/OpenFn/lightning/issues/4893)
+
 ## [2.16.8-pre] - 2026-06-18
 
 ### Added
diff --git a/lib/lightning/runs/handlers.ex b/lib/lightning/runs/handlers.ex
index 2362eb16b5..5df29968b1 100644
--- a/lib/lightning/runs/handlers.ex
+++ b/lib/lightning/runs/handlers.ex
@@ -472,16 +472,45 @@ defmodule Lightning.Runs.Handlers do
            },
            _run_options
          ) do
-      Dataclip.new(%{
-        id: dataclip_id,
-        project_id: project_id,
-        body: output_dataclip |> Jason.decode!() |> ensure_map(),
-        type: :step_result
-      })
-      |> Repo.insert()
+      changeset =
+        Dataclip.new(%{
+          id: dataclip_id,
+          project_id: project_id,
+          body: output_dataclip |> Jason.decode!() |> ensure_map(),
+          type: :step_result
+        })
+
+      if contains_null_byte?(Ecto.Changeset.get_field(changeset, :body)) do
+        {:error,
+         Dataclip.new(%{
+           id: dataclip_id,
+           project_id: project_id,
+           type: :step_result
+         })
+         |> Ecto.Changeset.add_error(
+           :body,
+           "contains a null byte (\\u0000) which PostgreSQL cannot store"
+         )}
+      else
+        Repo.insert(changeset)
+      end
     end
 
     defp ensure_map(%{} = map), do: map
     defp ensure_map(value), do: %{"value" => value}
+
+    defp contains_null_byte?(value) when is_binary(value),
+      do: String.contains?(value, <<0>>)
+
+    defp contains_null_byte?(value) when is_map(value),
+      do:
+        Enum.any?(value, fn {k, v} ->
+          contains_null_byte?(k) or contains_null_byte?(v)
+        end)
+
+    defp contains_null_byte?(value) when is_list(value),
+      do: Enum.any?(value, &contains_null_byte?/1)
+
+    defp contains_null_byte?(_value), do: false
   end
 end
diff --git a/test/lightning/runs_test.exs b/test/lightning/runs_test.exs
index cf5c6b0ad6..c57b1058f0 100644
--- a/test/lightning/runs_test.exs
+++ b/test/lightning/runs_test.exs
@@ -385,6 +385,73 @@ defmodule Lightning.RunsTest do
       assert Jason.decode!(step.output_dataclip.body) == %{"foo" => "bar"}
     end
 
+    test "returns an error changeset when output dataclip contains a null byte" do
+      dataclip = insert(:dataclip)
+      %{triggers: [trigger], jobs: [job]} = workflow = insert(:simple_workflow)
+
+      %{runs: [run]} =
+        work_order_for(trigger, workflow: workflow, dataclip: dataclip)
+        |> insert()
+
+      step =
+        insert(:step, runs: [run], job: job, input_dataclip: dataclip)
+
+      assert {:error, %Ecto.Changeset{}} =
+               Runs.complete_step(%{
+                 step_id: step.id,
+                 reason: "success",
+                 output_dataclip: ~s({"foo": "bar\\u0000baz"}),
+                 output_dataclip_id: Ecto.UUID.generate(),
+                 run_id: run.id,
+                 project_id: workflow.project_id
+               })
+    end
+
+    test "returns an error changeset when output dataclip contains a nested null byte" do
+      dataclip = insert(:dataclip)
+      %{triggers: [trigger], jobs: [job]} = workflow = insert(:simple_workflow)
+
+      %{runs: [run]} =
+        work_order_for(trigger, workflow: workflow, dataclip: dataclip)
+        |> insert()
+
+      step =
+        insert(:step, runs: [run], job: job, input_dataclip: dataclip)
+
+      assert {:error, %Ecto.Changeset{}} =
+               Runs.complete_step(%{
+                 step_id: step.id,
+                 reason: "success",
+                 output_dataclip: ~s({"items": ["ok", "bad\\u0000value"]}),
+                 output_dataclip_id: Ecto.UUID.generate(),
+                 run_id: run.id,
+                 project_id: workflow.project_id
+               })
+    end
+
+    test "completes when a null byte is only in top-level configuration" do
+      dataclip = insert(:dataclip)
+      %{triggers: [trigger], jobs: [job]} = workflow = insert(:simple_workflow)
+
+      %{runs: [run]} =
+        work_order_for(trigger, workflow: workflow, dataclip: dataclip)
+        |> insert()
+
+      step =
+        insert(:step, runs: [run], job: job, input_dataclip: dataclip)
+
+      assert {:ok, _step} =
+               Runs.complete_step(%{
+                 step_id: step.id,
+                 reason: "success",
+                 output_dataclip:
+                   ~s({"configuration": {"token": "abc\\u0000def"}, "data": "ok"}),
+                 output_dataclip_id: Ecto.UUID.generate(),
+                 run_id: run.id,
+                 project_id: workflow.project_id
+               })
+    end
+
     # Regression for #4800: dataclip inserts no longer build the search_vector
     # synchronously (the AFTER INSERT trigger was dropped). Saving an output
     # dataclip via the handler must succeed and the row must be retrievable with