Merge 80c3cffbff into 4fade39c90

hermes_cli/auth.py

@@ -356,6 +356,14 @@ PROVIDER_REGISTRY: Dict[str, ProviderConfig] = {
         api_key_env_vars=(),
         base_url_env_var="BEDROCK_BASE_URL",
     ),
+    "ollama": ProviderConfig(
+        id="ollama",
+        name="Ollama",
+        auth_type="api_key",
+        inference_base_url="http://127.0.0.1:11434/v1",
+        api_key_env_vars=("OLLAMA_API_KEY",),
+        base_url_env_var="OLLAMA_BASE_URL",
+    ),
 }
 
 
@@ -1111,9 +1119,10 @@ def resolve_provider(
         "aws": "bedrock", "aws-bedrock": "bedrock", "amazon-bedrock": "bedrock", "amazon": "bedrock",
         "go": "opencode-go", "opencode-go-sub": "opencode-go",
         "kilo": "kilocode", "kilo-code": "kilocode", "kilo-gateway": "kilocode",
-        # Local server aliases — route through the generic custom provider
+        # Local server aliases — route named built-ins directly and keep the
+        # generic custom provider for endpoints without first-class registry entries.
         "lmstudio": "custom", "lm-studio": "custom", "lm_studio": "custom",
-        "ollama": "custom", "ollama_cloud": "ollama-cloud",
+        "ollama": "ollama", "ollama_cloud": "ollama-cloud",
         "vllm": "custom", "llamacpp": "custom",
         "llama.cpp": "custom", "llama-cpp": "custom",
     }

hermes_cli/providers.py

@@ -167,6 +167,12 @@ HERMES_OVERLAYS: Dict[str, HermesOverlay] = {
         transport="openai_chat",
         base_url_env_var="OLLAMA_BASE_URL",
     ),
+    "ollama": HermesOverlay(
+        transport="openai_chat",
+        extra_env_vars=("OLLAMA_API_KEY",),
+        base_url_override="http://127.0.0.1:11434/v1",
+        base_url_env_var="OLLAMA_BASE_URL",
+    ),
 }
 
 
@@ -295,7 +301,7 @@ ALIASES: Dict[str, str] = {
     "lmstudio": "lmstudio",
     "lm-studio": "lmstudio",
     "lm_studio": "lmstudio",
-    "ollama": "custom",  # bare "ollama" = local; use "ollama-cloud" for cloud
+    "ollama": "ollama",
     "vllm": "local",
     "llamacpp": "local",
     "llama.cpp": "local",
@@ -316,6 +322,7 @@ _LABEL_OVERRIDES: Dict[str, str] = {
     "local": "Local endpoint",
     "bedrock": "AWS Bedrock",
     "ollama-cloud": "Ollama Cloud",
+    "ollama": "Ollama",
 }
 
 

tests/hermes_cli/test_ollama_cloud_auth.py

@@ -21,6 +21,62 @@ from unittest.mock import patch, MagicMock
 class TestOllamaCloudCredentials:
     """runtime_provider should use OLLAMA_API_KEY for ollama.com endpoints."""
 
+
+class TestOllamaProviderAliasRegression:
+    """Regression tests for the built-in ollama provider alias."""
+
+    def test_resolve_provider_full_keeps_ollama_canonical(self):
+        """`ollama` should resolve to the built-in ollama provider, not ollama-cloud."""
+        from hermes_cli.providers import resolve_provider_full
+        from hermes_cli.auth import resolve_provider
+
+        resolved = resolve_provider_full("ollama", user_providers={}, custom_providers=[])
+
+        assert resolved is not None
+        assert resolved.id == "ollama"
+        assert resolved.name == "Ollama"
+        assert resolve_provider("ollama") == "ollama"
+
+    def test_switch_model_explicit_ollama_provider_does_not_map_to_ollama_cloud(self, monkeypatch):
+        """`/model ... --provider ollama` should stay on the built-in ollama provider."""
+        from hermes_cli.model_switch import switch_model
+
+        monkeypatch.setattr(
+            "hermes_cli.runtime_provider.resolve_runtime_provider",
+            lambda requested: {
+                "provider": requested,
+                "api_key": "ollama-test-key",
+                "base_url": "http://127.0.0.1:11434/v1",
+                "api_mode": "chat_completions",
+            },
+        )
+        monkeypatch.setattr("hermes_cli.models.validate_requested_model", lambda *a, **k: {
+            "accepted": True,
+            "persist": True,
+            "recognized": True,
+            "message": None,
+        })
+        monkeypatch.setattr("hermes_cli.model_switch.get_model_info", lambda *a, **k: None)
+        monkeypatch.setattr("hermes_cli.model_switch.get_model_capabilities", lambda *a, **k: None)
+
+        result = switch_model(
+            raw_input="qwen3.5:cloud",
+            current_provider="openai-codex",
+            current_model="gpt-5.4",
+            current_base_url="https://chatgpt.com/backend-api/codex",
+            current_api_key="dummy",
+            explicit_provider="ollama",
+            user_providers={},
+            custom_providers=[],
+        )
+
+        assert result.success is True
+        assert result.target_provider == "ollama"
+        assert result.provider_label == "Ollama"
+        assert result.new_model == "qwen3.5:cloud"
+        assert result.base_url == "http://127.0.0.1:11434/v1"
+
+
     def test_ollama_api_key_used_for_ollama_endpoint(self, monkeypatch, tmp_path):
         """When base_url contains ollama.com, OLLAMA_API_KEY is in the candidate chain."""
         monkeypatch.setenv("OLLAMA_API_KEY", "test-ollama-key-12345")