feat(api): integrate MCP tools support in Ollama chat route

Add support for Model Context Protocol (MCP) tools in the Ollama chat API. Detect model capabilities to use native tools or prompt engineering. Handle tool calls, execute them via MCPExecutor, and inject results back into the conversation for seamless tool-assisted responses. This enhances chat functionality by enabling external tool integration.

lib/services/model-capabilities.ts

@@ -0,0 +1,77 @@
+/**
+ * Serviço para detectar capabilities de function calling dos modelos
+ */
+
+import type { ModelCapabilities } from "@/lib/types/mcp-chat";
+
+/**
+ * Lista de modelos com suporte nativo a function calling
+ * Baseado na documentação do Ollama (atualizado em nov 2024)
+ */
+const MODELS_WITH_NATIVE_TOOLS = [
+  // Llama 3.x
+  /^llama3\.1/i,
+  /^llama3\.2/i,
+  /^llama3-/i,
+
+  // Mistral
+  /^mistral/i,
+  /^mistral-nemo/i,
+  /^mixtral/i,
+
+  // Gemma
+  /^gemma2/i,
+
+  // Qwen
+  /^qwen2\.5/i,
+
+  // Command-R
+  /^command-r/i,
+
+  // Firefunction
+  /^firefunction/i,
+];
+
+/**
+ * Detecta capabilities de function calling de um modelo
+ * @param modelName Nome do modelo (ex: "llama3.1:8b", "qwen2.5:7b")
+ * @returns Capabilities do modelo
+ */
+export function detectModelCapabilities(modelName: string): ModelCapabilities {
+  const supportsNativeTools = MODELS_WITH_NATIVE_TOOLS.some((pattern) =>
+    pattern.test(modelName)
+  );
+
+  return {
+    modelName,
+    supportsNativeTools,
+    requiresPromptEngineering: !supportsNativeTools,
+    toolCallFormat: supportsNativeTools ? "openai" : undefined,
+  };
+}
+
+/**
+ * Verifica se um modelo específico suporta function calling nativo
+ * @param modelName Nome do modelo
+ * @returns true se suporta, false caso contrário
+ */
+export function supportsNativeTools(modelName: string): boolean {
+  return detectModelCapabilities(modelName).supportsNativeTools;
+}
+
+/**
+ * Lista todos os modelos conhecidos com suporte a function calling
+ * @returns Array de padrões regex em string
+ */
+export function getSupportedModelsPatterns(): string[] {
+  return [
+    "llama3.1:*",
+    "llama3.2:*",
+    "mistral:*",
+    "mixtral:*",
+    "gemma2:*",
+    "qwen2.5:*",
+    "command-r:*",
+    "firefunction:*",
+  ];
+}