fix: address issues in model pull and chat API

- Refactored model pull logic to improve error handling and user feedback.
- Enhanced chat API to ensure better performance and reliability during message fetching.
- Updated UI components for a more intuitive user experience when interacting with models and messages.

hooks/use-ollama-check.ts

@@ -0,0 +1,41 @@
+import { useState, useEffect } from 'react';
+import { invoke } from '@tauri-apps/api/core';
+
+export type OllamaStatus = 'checking' | 'not_installed' | 'installed_stopped' | 'running';
+
+export function useOllamaCheck() {
+  const [status, setStatus] = useState<OllamaStatus>('checking');
+
+  const check = async () => {
+    // Don't set checking here if we want to silently revalidate, 
+    // but for initial load or explicit re-check it is fine.
+    // Only set checking if it's not already in a known state to avoid flicker?
+    // For now simple is better.
+    try {
+      const installed = await invoke<boolean>('check_ollama_installed');
+      if (!installed) {
+        setStatus('not_installed');
+        return;
+      }
+
+      const running = await invoke<boolean>('check_ollama_running');
+      if (!running) {
+        setStatus('installed_stopped');
+        return;
+      }
+
+      setStatus('running');
+    } catch (error) {
+      console.error('Failed to check ollama:', error);
+      // If invoke fails (e.g. backend not ready), we might want to handle gracefully
+      setStatus('not_installed');
+    }
+  };
+
+  useEffect(() => {
+    check();
+  }, []);
+
+  return { status, check };
+}
+