feat(models): add Ollama to model resolution and status

- Add OLLAMA_API_KEY to env key resolution map
- Skip model picker for Ollama (uses auto-discovered models)
- Include Ollama in status command env probe providers

src/agents/model-auth.ts

@@ -275,6 +275,7 @@ export function resolveEnvApiKey(provider: string): EnvApiKeyResult | null {
     groq: "GROQ_API_KEY",
     deepgram: "DEEPGRAM_API_KEY",
     cerebras: "CEREBRAS_API_KEY",
+    ollama: "OLLAMA_API_KEY",
     xai: "XAI_API_KEY",
     openrouter: "OPENROUTER_API_KEY",
     "vercel-ai-gateway": "AI_GATEWAY_API_KEY",

src/commands/model-picker.ts

@@ -145,6 +145,14 @@ export async function promptDefaultModel(
   );
 
   const hasPreferredProvider = preferredProvider ? providers.includes(preferredProvider) : false;
+
+  // If preferredProvider is specified but not in catalog, skip the picker.
+  // This handles auto-discovering providers (e.g., Ollama) where models aren't
+  // in the static catalog but were already configured by the auth handler.
+  if (preferredProvider && !hasPreferredProvider) {
+    return {};
+  }
+
   const shouldPromptProvider =
     !hasPreferredProvider && providers.length > 1 && models.length > PROVIDER_FILTER_THRESHOLD;
   if (shouldPromptProvider) {

src/commands/models/list.status-command.ts

@@ -131,6 +131,7 @@ export async function modelsStatusCommand(
     "cerebras",
     "xai",
     "openrouter",
+    "ollama",
     "zai",
     "mistral",
     "synthetic",