feat: fully enable self-correction loops

- Import withSelfCorrection from self-correction.js
- Wrap chatWithAI() with self-correction wrapper
- Add /selfcorrection command to show status
- Update /start to mention self-correction and streaming
- Self-correction: 2 retries + exponential backoff + auto-simplification
- Triggers: error responses, rate limits, timeouts, 5xx errors

src/bot/index.js

@@ -10,6 +10,7 @@ import { getRTK } from '../utils/rtk.js';
 import { isDuplicate, markProcessed } from './deduplication.js';
 import { queueRequest, clearQueue, isProcessing } from './request-queue.js';
 import { sendFormatted, splitMessage, escapeMarkdown } from './message-sender.js';
+import { withSelfCorrection } from './self-correction.js';
 
 function buildSessionKey(chatId, threadId) {
   return threadId ? `${chatId}:${threadId}` : String(chatId);
@@ -314,7 +315,10 @@ export async function initBot(config, api, tools, skills, agents) {
       '📚 *Skills:* ' + svc.skills.length + ' loaded',
       '🤖 *Agents:* ' + svc.agents.length + ' available',
       '',
-      '📋 *Commands:* /tools /skills /agents /model /stats /voice /mcp /memory /cron /cancel',
+      '🔄 *Self-Correction*: 2 retries + auto-simplification',
+      '⚡ *Streaming*: Real-time text delivery',
+      '',
+      '📋 *Commands:* /tools /skills /agents /model /stats /voice /mcp /memory /cron /cancel /selfcorrection',
       '',
       'Or just chat — I will use tools when needed.',
       `Model: \`${svc.config?.api?.models?.default || 'glm-5.1'}\``,
@@ -349,6 +353,10 @@ export async function initBot(config, api, tools, skills, agents) {
     }
     await sendStreamingMessage(ctx, lines.join('\n'));
   });
+  
+  bot.command('selfcorrection', async (ctx) => {
+    await sendStreamingMessage(ctx, `🔄 *Self-Correction Loops* — FULLY ENABLED\n\nzCode CLI X now uses automatic self-correction:\n\n• **Max Retries**: 2 attempts\n• **Retry Delay**: 500ms → 1s → 1.5s (exponential backoff)\n• **Triggers**: \n  - ❌ Error responses\n  - Rate limits\n  - Timeouts\n  - 5xx server errors\n\n• **Auto-Simplification**: On retry, prompts are simplified to avoid recurring errors\n\n• **Logging**: All retries are logged with retry count and reason\n\nThis ensures robust responses even when the AI initially fails.`);
+  });
 
   bot.command('model', async (ctx) => {
     const text = ctx.match?.trim();
@@ -427,7 +435,7 @@ export async function initBot(config, api, tools, skills, agents) {
     });
   }
 
-  // ── Message text handler (with dedup + queue) ──
+  // ── Message text handler (with dedup + queue + self-correction) ──
   bot.on('message:text', async (ctx) => {
     if (isDuplicate(ctx.message.message_id)) return;
     markProcessed(ctx.message.message_id);
@@ -439,7 +447,13 @@ export async function initBot(config, api, tools, skills, agents) {
 
     await queueRequest(key, text, async () => {
       await ctx.api.sendChatAction(ctx.chat.id, 'typing');
-      const result = await chatWithAI([
+      
+      // Wrap chatWithAI with self-correction
+      const chatWithCorrection = withSelfCorrection(async (msgs) => {
+        return await chatWithAI(msgs, {});
+      });
+      
+      const result = await chatWithCorrection([
         { role: 'system', content: buildSystemPrompt(svc) },
         { role: 'user', content: text },
       ]);