Fix: skip old messages, drop pending updates, reduce max_tokens for faster response

bot.js

@@ -1,4 +1,4 @@
-const { Bot } = require("grammy");
+const { Bot, GrammyError, HttpError } = require("grammy");
 
 // Environment variables
 const BOT_TOKEN = process.env.TELEGRAM_BOT_TOKEN;
@@ -43,9 +43,9 @@ async function callZenAPI(messages, modelId) {
 
   let body;
   if (endpoint.includes("/messages")) {
-    body = { model: modelId, messages, max_tokens: 4096 };
+    body = { model: modelId, messages, max_tokens: 2048 };
   } else {
-    body = { model: modelId, messages, max_tokens: 4096 };
+    body = { model: modelId, messages, max_tokens: 2048 };
   }
 
   console.log("[API] Calling:", endpoint, "model:", modelId);
@@ -176,6 +176,14 @@ bot.command("new", async (ctx) => {
 bot.on("message:text", async (ctx) => {
   const chatId = ctx.chat.id;
   const message = ctx.message.text;
+  const msgTime = ctx.message.date;
+  const now = Math.floor(Date.now() / 1000);
+
+  // Skip old messages (sent more than 30 seconds ago)
+  if (now - msgTime > 30) {
+    console.log("[SKIP] Old message (", now - msgTime, "s ago):", message.substring(0, 30));
+    return;
+  }
 
   console.log("[LOG] Message from chat:", chatId, "type:", ctx.chat.type, "text:", message.substring(0, 50));
 
@@ -214,9 +222,11 @@ bot.on("message:text", async (ctx) => {
       const chunks = response.match(/.{1,4000}/gs) || [response];
       for (const chunk of chunks) {
         await ctx.reply(chunk);
+        console.log("[REPLY] Sent chunk:", chunk.length, "chars");
       }
     } else {
       await ctx.reply(response);
+      console.log("[REPLY] Sent:", response.length, "chars");
     }
   } catch (error) {
     console.error("[ERROR] Chat API failed:", error.message);
@@ -235,5 +245,7 @@ bot.start({
   onStart: () => {
     console.log("[OK] Telegram bot is running!");
     console.log("[OK] Chat ID filter:", CHAT_ID || "none (accepting all)");
+    console.log("[OK] Waiting for NEW messages only...");
   },
+  drop_pending_updates: true,
 });