feat: route profiled ops runs through Glas

Assistant: codex
Assistant-Model: gpt-5.6-sol
Assistant-Session: 01a02b6f-7db1-7222-918b-e813a6bda38d

tests/test_mail_triage.py

@@ -169,6 +169,8 @@ def test_llm_connect_client_complete(monkeypatch) -> None:
     import rein_aharness.llm_connect_client as mod
 
     class FakeResp:
+        status_code = 200
+
         def raise_for_status(self) -> None:
             return None
 
@@ -187,6 +189,68 @@ def test_llm_connect_client_complete(monkeypatch) -> None:
     assert client.last_response_metadata.get("model") == "test-model"
 
 
+class _ErrorResp:
+    def __init__(self, status_code: int, body) -> None:
+        self.status_code = status_code
+        self._body = body
+
+    def json(self):
+        if isinstance(self._body, Exception):
+            raise self._body
+        return self._body
+
+
+def test_llm_connect_error_preserves_safe_provider_cause(monkeypatch) -> None:
+    import rein_aharness.llm_connect_client as mod
+
+    response = _ErrorResp(
+        502,
+        {
+            "error": "provider_api_error",
+            "provider_status": 401,
+            "provider": "openrouter",
+            "model": "model-1",
+            "message": "No auth credentials found",
+            "api_key": "sk-must-not-appear",
+            "raw_response": {"authorization": "Bearer secret"},
+        },
+    )
+    monkeypatch.setattr(mod.httpx, "post", lambda *args, **kwargs: response)
+
+    with pytest.raises(LLMConnectError) as excinfo:
+        LLMConnectClient("http://llm.test").complete("hi")
+
+    text = str(excinfo.value)
+    assert "HTTP 502" in text
+    assert "error=provider_api_error" in text
+    assert "provider_status=401" in text
+    assert "provider=openrouter" in text
+    assert "model=model-1" in text
+    assert "message=No auth credentials found" in text
+    assert "sk-must-not-appear" not in text
+    assert "authorization" not in text
+
+
+def test_llm_connect_error_message_is_bounded(monkeypatch) -> None:
+    import rein_aharness.llm_connect_client as mod
+
+    response = _ErrorResp(502, {"error": "provider_api_error", "message": "x" * 5000})
+    monkeypatch.setattr(mod.httpx, "post", lambda *args, **kwargs: response)
+    with pytest.raises(LLMConnectError) as excinfo:
+        LLMConnectClient("http://llm.test").complete("hi")
+    assert len(str(excinfo.value)) < 600
+
+
+@pytest.mark.parametrize("body", [ValueError("not json"), ["unexpected"]])
+def test_llm_connect_error_without_usable_body_reports_status(monkeypatch, body) -> None:
+    import rein_aharness.llm_connect_client as mod
+
+    response = _ErrorResp(504, body)
+    monkeypatch.setattr(mod.httpx, "post", lambda *args, **kwargs: response)
+    with pytest.raises(LLMConnectError, match="llm-connect returned HTTP 504"):
+        LLMConnectClient("http://llm.test").complete("hi")
+
+
 def test_get_client_requires_env(monkeypatch) -> None:
     monkeypatch.delenv("LLM_CONNECT_URL", raising=False)
     with pytest.raises(LLMConnectError, match="LLM_CONNECT_URL"):