diff --git a/src/opensquilla/observability/replay.py b/src/opensquilla/observability/replay.py
index 66a0b7b0b..86ff39961 100644
--- a/src/opensquilla/observability/replay.py
+++ b/src/opensquilla/observability/replay.py
@@ -56,7 +56,12 @@ def format_transcript(entry: DecisionEntry) -> str:
         "  Pipeline steps:",
     ]
     for step in entry.pipeline_steps:
-        status = "OK" if step.applied else f"FAIL({step.fallback_reason})"
+        if step.applied:
+            status = "OK"
+        elif step.fallback_reason is None:
+            status = "SKIPPED"
+        else:
+            status = f"FAIL({step.fallback_reason})"
         lines.append(
             f"    - {step.step_name} [{status}] "
             f"tier={step.routed_tier} "
diff --git a/tests/test_observability/test_replay.py b/tests/test_observability/test_replay.py
new file mode 100644
index 000000000..56632431f
--- /dev/null
+++ b/tests/test_observability/test_replay.py
@@ -0,0 +1,57 @@
+from __future__ import annotations
+
+import pytest
+
+from opensquilla.observability.decision_log import DecisionEntry, PipelineStepRecord
+from opensquilla.observability.replay import format_transcript
+
+
+def _entry_with_step(step: PipelineStepRecord) -> DecisionEntry:
+    return DecisionEntry(
+        turn_id="turn-1",
+        session_key="agent:main:webchat:default",
+        prompt_hash="prompt",
+        system_prompt_hash="system",
+        tool_list_hash="tools",
+        tool_choice="auto",
+        tokens_input=1,
+        tokens_output=1,
+        model="test-model",
+        provider="test-provider",
+        latency_ms=1,
+        ts="2026-09-03T00:00:00Z",
+        pipeline_steps=[step],
+    )
+
+
+@pytest.mark.parametrize(
+    ("step", "expected_status"),
+    [
+        (PipelineStepRecord(step_name="step", applied=True), "OK"),
+        (PipelineStepRecord(step_name="step", applied=False), "SKIPPED"),
+        (
+            PipelineStepRecord(
+                step_name="step",
+                applied=False,
+                fallback_reason="boom",
+            ),
+            "FAIL(boom)",
+        ),
+        (
+            PipelineStepRecord(
+                step_name="step",
+                applied=False,
+                fallback_reason="",
+            ),
+            "FAIL()",
+        ),
+    ],
+    ids=("applied", "skipped", "failed", "empty-failure-reason"),
+)
+def test_format_transcript_renders_pipeline_step_status(
+    step: PipelineStepRecord,
+    expected_status: str,
+) -> None:
+    transcript = format_transcript(_entry_with_step(step))
+
+    assert f"    - step [{expected_status}]" in transcript