fix(llm): normalize model environment values

src/llm/providers/astraflow.py

@@ -16,6 +16,7 @@ from llm.core.interface import (
 )
 from llm.core.types import LLMInput, LLMOutput, ModelInfo, ProviderType, ToolCall
 from llm.providers.constants import EMPTY_FILTERED_RESPONSE_ERROR
+from llm.providers.model_config import read_model_env
 from llm.providers.openai_compat import to_openai_message_param
 
 ASTRAFLOW_BASE_URL = "https://api.umodelverse.ai/v1"
@@ -54,8 +55,8 @@ class _AstraflowBaseProvider(LLMProvider):
     ) -> None:
         self.api_key = api_key or os.environ.get(self.api_key_env) or ""
         self.base_url = base_url or os.environ.get(self.base_url_env, self.default_base_url)
-        env_model = os.environ.get(self.model_env)
-        fallback_model = os.environ.get(self.fallback_model_env) if self.fallback_model_env else None
+        env_model = read_model_env(self.model_env)
+        fallback_model = read_model_env(self.fallback_model_env) if self.fallback_model_env else None
         self.default_model = default_model or env_model or fallback_model or DEFAULT_ASTRAFLOW_MODEL
         self.client = OpenAI(api_key=self.api_key, base_url=self.base_url, _enforce_credentials=False)
         self._models = [

src/llm/providers/atlas.py

@@ -16,6 +16,7 @@ from llm.core.interface import (
 )
 from llm.core.types import LLMInput, LLMOutput, ModelInfo, ProviderType, ToolCall
 from llm.providers.constants import EMPTY_FILTERED_RESPONSE_ERROR
+from llm.providers.model_config import read_model_env
 from llm.providers.openai_compat import to_openai_message_param
 
 ATLAS_BASE_URL = "https://api.atlascloud.ai/v1"
@@ -68,7 +69,7 @@ class AtlasProvider(LLMProvider):
             or ""
         )
         self.base_url = base_url or os.environ.get(self.base_url_env, self.default_base_url)
-        env_model = os.environ.get(self.model_env)
+        env_model = read_model_env(self.model_env)
         self.default_model = default_model or env_model or DEFAULT_ATLAS_MODEL
         self.client = OpenAI(api_key=self.api_key, base_url=self.base_url, _enforce_credentials=False)
         self._models = [

src/llm/providers/model_config.py

@@ -0,0 +1,11 @@
+"""Model environment parsing shared by the resolver and provider adapters."""
+
+import os
+
+
+def read_model_env(name: str) -> str | None:
+    """Read a model ID, treating surrounding whitespace and quotes as config syntax."""
+    value = os.environ.get(name, "").strip()
+    if len(value) >= 2 and value[0] == value[-1] and value[0] in {"'", '"'}:
+        value = value[1:-1].strip()
+    return value or None

src/llm/providers/ollama.py

@@ -19,6 +19,7 @@ from llm.core.types import (
     ProviderType,
     ToolCall,
 )
+from llm.providers.model_config import read_model_env
 
 
 class OllamaProvider(LLMProvider):
@@ -30,7 +31,7 @@ class OllamaProvider(LLMProvider):
         default_model: str | None = None,
     ) -> None:
         self.base_url = base_url or os.environ.get("OLLAMA_BASE_URL", "http://localhost:11434")
-        self.default_model = default_model or os.environ.get("OLLAMA_MODEL", "llama3.2")
+        self.default_model = default_model or read_model_env("OLLAMA_MODEL") or "llama3.2"
         self._models = [
             ModelInfo(
                 name="llama3.2",

src/llm/providers/resolver.py

@@ -10,6 +10,7 @@ from llm.core.types import ProviderType
 from llm.providers.astraflow import AstraflowCNProvider, AstraflowProvider
 from llm.providers.atlas import AtlasProvider
 from llm.providers.claude import ClaudeProvider
+from llm.providers.model_config import read_model_env
 from llm.providers.ollama import OllamaProvider
 from llm.providers.openai import OpenAIProvider
 
@@ -78,19 +79,19 @@ def _resolve_default_model(
     provider_type: ProviderType, provider_cls: type[LLMProvider], saved_model: str | None,
 ) -> str | None:
     for name in _MODEL_ENV_VARS[provider_cls]:
-        model = os.environ.get(name)
+        model = read_model_env(name)
         if model:
             return model
 
     env_provider = _strip_env_value(os.environ.get("LLM_PROVIDER", "")).lower()
     if not env_provider or env_provider == provider_type.value:
-        env_model = _strip_env_value(os.environ.get("LLM_MODEL", ""))
+        env_model = read_model_env("LLM_MODEL")
         if env_model:
             return env_model
     return saved_model or None
 
 
-def get_provider(provider_type: ProviderType | str | None = None, **kwargs: str) -> LLMProvider:
+def get_provider(provider_type: ProviderType | str | None = None, **kwargs: str | None) -> LLMProvider:
     """Create a provider, honoring the model selected by llm-select.
 
     Built-in model defaults prefer explicit kwargs, then existing provider-specific