feat: implement AI agent for flowme-life interactions

- Add agent-run module to handle AI interactions with tools and messages.
- Create routes for proxying requests to OpenAI and Anthropic APIs.
- Implement flowme-life chat route for user queries and task management.
- Add services for retrieving and updating life records in the database.
- Implement logic for fetching today's tasks and marking tasks as done with next execution time calculation.
- Introduce tests for flowme-life functionalities.

src/routes/flowme-life/chat.ts

@@ -0,0 +1,44 @@
+
+import { schema, app, cnb, models } from '@/app.ts'
+import z from 'zod';
+import { runAgent } from '@kevisual/ai/agent'
+
+app.route({
+  path: 'flowme-life',
+  key: 'chat',
+  description: `聊天接口, 对自己的数据进行操作，参数是 question或messages，question是用户的提问，messages是对话消息列表，优先级高于 question`,
+  middleware: ['auth']
+  , metadata: {
+    args: {
+      question: z.string().describe('用户的提问'),
+      messages: z.any().optional().describe('对话消息列表，优先级高于 question'),
+    }
+  }
+}).define(async (ctx) => {
+  const question = ctx.query.question || '';
+  const _messages = ctx.query.messages;
+  const token = ctx.query.token || '';
+  if (!question && !_messages) {
+    ctx.throw(400, '缺少参数 question 或 messages');
+  }
+  const routes = ctx.app.getList().filter(r => r.path.startsWith('flowme-life') && r.key !== 'chat');
+  const messages = _messages || [
+    {
+      "role": "system" as const,
+      "content": `你是我的智能助手，协助我操作我的数据, 请根据我的提问选择合适的接口进行调用。`
+    },
+    {
+      "role": "user" as const,
+      "content": question
+    }
+  ]
+  const res = await runAgent({
+    app: app,
+    messages: messages,
+    languageModel: cnb(models['auto']),
+    // query: 'WHERE path LIKE 'flowme-life%',
+    routes,
+    token,
+  });
+  ctx.body = res
+}).addTo(app);