fix: clear error messages for missing API keys on provider switch

Previously, switching to zhipuai/openrouter/xai via /model would throw a
confusing 'OPENAI_API_KEY missing' error from the OpenAI SDK. Now
createClientFromConfig validates API keys before constructing the client,
throwing errors that name the correct env var (e.g. ZHIPUAI_API_KEY).

Also fixes the misleading 'as anthropic' type cast in the /model handler
to validate against MODEL_PROVIDERS and use the ModelProvider type.

src/daemon/models.ts

@@ -3,6 +3,21 @@ import { AnthropicClient, OpenAIClient, OllamaClient, LlamaCppClient, GeminiClie
 import type { ModelClient, RetryConfig, ModelTier } from '../models/index.js';
 import { logger } from '../logger.js';
 
+/**
+ * Resolve an API key from config or environment variable.
+ * Throws a clear error naming the expected env var if neither source provides a key.
+ */
+function requireApiKey(cfg: ModelConfig, envVar: string): string {
+  const key = cfg.api_key ?? process.env[envVar];
+  if (!key) {
+    throw new Error(
+      `API key required for ${cfg.provider}. ` +
+      `Set ${envVar} environment variable or provide api_key in config.`
+    );
+  }
+  return key;
+}
+
 /**
  * Create a ModelClient from a provider config entry.
  * Dispatches on the `provider` field so all tiers and fallback entries
@@ -41,19 +56,19 @@ export function createClientFromConfig(cfg: ModelConfig): ModelClient {
     case 'openrouter':
       return new OpenAIClient({
         model: cfg.model,
-        apiKey: cfg.api_key ?? process.env.OPENROUTER_API_KEY,
+        apiKey: requireApiKey(cfg, 'OPENROUTER_API_KEY'),
         baseURL: cfg.endpoint ?? 'https://openrouter.ai/api/v1',
       });
     case 'zhipuai':
       return new OpenAIClient({
         model: cfg.model,
-        apiKey: cfg.api_key ?? process.env.ZHIPUAI_API_KEY,
+        apiKey: requireApiKey(cfg, 'ZHIPUAI_API_KEY'),
         baseURL: cfg.endpoint ?? 'https://api.z.ai/api/paas/v4',
       });
     case 'xai':
       return new OpenAIClient({
         model: cfg.model,
-        apiKey: cfg.api_key ?? process.env.XAI_API_KEY,
+        apiKey: requireApiKey(cfg, 'XAI_API_KEY'),
         baseURL: cfg.endpoint ?? 'https://api.x.ai/v1',
       });
     case 'bedrock':

src/frontends/tui/minimal.ts

@@ -5,7 +5,8 @@ import type { ModelRouter, ModelTier } from '../../models/router.js';
 import type { NativeAgent } from '../../backends/native/agent.js';
 import { parseCommand, getHelpText, resolveModelAlias, getCommandCompletions, getCommandTooltip, type Command } from './commands.js';
 import { renderMarkdown } from './markdown.js';
-import type { ModelConfig } from '../../config/schema.js';
+import type { ModelConfig, ModelProvider } from '../../config/schema.js';
+import { MODEL_PROVIDERS } from '../../config/schema.js';
 import { OllamaClient, LlamaCppClient } from '../../models/index.js';
 import { createClientFromConfig } from '../../daemon/index.js';
 import { loginGitHub } from '../../auth/index.js';
@@ -232,8 +233,13 @@ export class MinimalTui {
       const provider = providerModel.slice(0, slashIdx);
       const model = providerModel.slice(slashIdx + 1);
 
+      if (!MODEL_PROVIDERS.includes(provider as ModelProvider)) {
+        console.log(`${colors.gray}Unknown provider "${provider}". Known providers: ${MODEL_PROVIDERS.join(', ')}${colors.reset}\n`);
+        return;
+      }
+
       try {
-        const client = createClientFromConfig({ provider: provider as 'anthropic', model });
+        const client = createClientFromConfig({ provider: provider as ModelProvider, model });
         router.setClient(tier, client, providerModel);
         console.log(`${colors.gray}Set ${tier} to:${colors.reset} ${providerModel}\n`);
       } catch (error) {