diff --git a/agent/codex_responses_adapter.py b/agent/codex_responses_adapter.py
index 6d41b2a9e8ba..1d14148c7aeb 100644
--- a/agent/codex_responses_adapter.py
+++ b/agent/codex_responses_adapter.py
@@ -1382,10 +1382,11 @@ def _normalize_codex_response(
         # Response contains only reasoning (encrypted thinking state and/or
         # human-readable summary) with no visible content or tool calls.
         #
-        # For known Codex/xAI backends, reasoning-only with status="completed"
-        # means "the model is still thinking and needs another turn" — treat
-        # it as incomplete so the Codex continuation path retries instead of
-        # falling into the empty-content retry loop.
+        # For the specially-handled backends (Codex, xAI, GitHub/Copilot),
+        # reasoning-only with status="completed" means "the model is still
+        # thinking and needs another turn" — treat it as incomplete so the
+        # Codex continuation path retries instead of falling into the
+        # empty-content retry loop.
         #
         # For all other backends (other:<base_url>, etc.), trust the provider's
         # own response.status signal. When status == "completed" and no items
@@ -1393,7 +1394,11 @@ def _normalize_codex_response(
         # state — forcing "incomplete" causes multi-minute stalls as the
         # continuation path re-issues calls (3 retries × up to 240s each).
         # See https://github.com/NousResearch/hermes-agent/issues/64434
-        if response_status == "completed" and issuer_kind not in ("codex_backend", "xai_responses"):
+        if response_status == "completed" and issuer_kind not in (
+            "codex_backend",
+            "xai_responses",
+            "github_responses",
+        ):
             finish_reason = "stop"
         else:
             finish_reason = "incomplete"
diff --git a/tests/run_agent/test_run_agent_codex_responses.py b/tests/run_agent/test_run_agent_codex_responses.py
index c0d86ddd731b..6aa381dd2321 100644
--- a/tests/run_agent/test_run_agent_codex_responses.py
+++ b/tests/run_agent/test_run_agent_codex_responses.py
@@ -2414,7 +2414,7 @@ def test_normalize_codex_response_reasoning_only_completed_is_stop_without_issue
     assert assistant_message.content == ""
 
 
-def test_normalize_codex_response_reasoning_only_is_stop_for_xai_backend(monkeypatch):
+def test_normalize_codex_response_reasoning_only_stays_incomplete_for_xai_backend(monkeypatch):
     """xAI backend also preserves incomplete for reasoning-only (same as Codex)."""
     agent = _build_agent(monkeypatch)
     from agent.codex_responses_adapter import _normalize_codex_response
@@ -2427,6 +2427,25 @@ def test_normalize_codex_response_reasoning_only_is_stop_for_xai_backend(monkeyp
     assert assistant_message.content == ""
 
 
+def test_normalize_codex_response_reasoning_only_stays_incomplete_for_github_backend(monkeypatch):
+    """GitHub/Copilot Responses backend preserves incomplete for reasoning-only.
+
+    Copilot fronts the same OpenAI model family as codex_backend and exhibits
+    the same reasoning-only "still thinking" degeneration mode, so it must
+    stay on the continuation path — only unrecognized (other:*) backends
+    trust response.status='completed' as terminal.
+    """
+    agent = _build_agent(monkeypatch)
+    from agent.codex_responses_adapter import _normalize_codex_response
+    response = _codex_reasoning_only_response()
+    assistant_message, finish_reason = _normalize_codex_response(
+        response, issuer_kind="github_responses"
+    )
+
+    assert finish_reason == "incomplete"
+    assert assistant_message.content == ""
+
+
 def test_normalize_codex_response_reasoning_with_content_is_stop(monkeypatch):
     """If a response has both reasoning and message content, it should still be 'stop'."""
     agent = _build_agent(monkeypatch)