refactor(providers): drop unreachable GenerationSettings fallback

nanobot/providers/base.py

@@ -514,12 +514,8 @@ class LLMProvider(ABC):
         """Call chat_stream() with retry on transient provider failures."""
         if max_tokens is self._SENTINEL or max_tokens is None:
             max_tokens = self.generation.max_tokens
-        if max_tokens is None:
-            max_tokens = GenerationSettings.max_tokens
         if temperature is self._SENTINEL or temperature is None:
             temperature = self.generation.temperature
-        if temperature is None:
-            temperature = GenerationSettings.temperature
         if reasoning_effort is self._SENTINEL:
             reasoning_effort = self.generation.reasoning_effort
 
@@ -554,19 +550,14 @@ class LLMProvider(ABC):
         Parameters default to ``self.generation`` when not explicitly passed,
         so callers no longer need to thread temperature / max_tokens /
         reasoning_effort through every layer. Explicit ``None`` is also
-        normalized to the provider's generation defaults, with a final fallback
+        normalized to the provider's generation defaults so that downstream
-        to :class:`GenerationSettings` class defaults so that downstream
         ``_build_kwargs`` never sees ``None`` for ``max_tokens`` / ``temperature``
         (which would crash ``max(1, max_tokens)``).
         """
         if max_tokens is self._SENTINEL or max_tokens is None:
             max_tokens = self.generation.max_tokens
-        if max_tokens is None:
-            max_tokens = GenerationSettings.max_tokens
         if temperature is self._SENTINEL or temperature is None:
             temperature = self.generation.temperature
-        if temperature is None:
-            temperature = GenerationSettings.temperature
         if reasoning_effort is self._SENTINEL:
             reasoning_effort = self.generation.reasoning_effort
 

tests/providers/test_provider_retry.py

@@ -546,23 +546,6 @@ async def test_chat_with_retry_normalizes_explicit_none_max_tokens() -> None:
     assert provider.last_kwargs["temperature"] == 0.7
 
 
-@pytest.mark.asyncio
-async def test_chat_with_retry_hard_fallback_when_generation_max_tokens_is_none() -> None:
-    """Final fallback: even if provider.generation.max_tokens is None,
-    chat() must receive the GenerationSettings class default (not None)."""
-    provider = ScriptedProvider([LLMResponse(content="ok")])
-    # Bypass the dataclass type hint by constructing with None explicitly.
-    provider.generation = GenerationSettings(max_tokens=None, temperature=None)  # type: ignore[arg-type]
-
-    response = await provider.chat_with_retry(
-        messages=[{"role": "user", "content": "hi"}],
-    )
-
-    assert response.content == "ok"
-    assert provider.last_kwargs["max_tokens"] == GenerationSettings.max_tokens
-    assert provider.last_kwargs["temperature"] == GenerationSettings.temperature
-
-
 @pytest.mark.asyncio
 async def test_chat_stream_with_retry_normalizes_explicit_none_max_tokens() -> None:
     """chat_stream_with_retry must apply the same None-guard as chat_with_retry."""