OpenClaw Telegram Bot Not Responding? 5 Fixes

Your OpenClaw agent was running fine yesterday. Today, you send it a message on Telegram and... nothing. Your OpenClaw Telegram bot is not responding. Or worse — the response comes 5-10 minutes later, long after you've given up.

This is one of the most frustrating OpenClaw issues because the agent often looks healthy. Logs show it's running. The gateway says "connected." But messages are either delayed, dropped, or silently ignored.

Here are the 5 most common causes — and the exact fix for each.

Symptom Check: Which Problem Do You Have?

Before diving into fixes, identify your symptom:

Cause 1: Telegram Bot Polling Never Started

Symptom: Fresh setup, OpenClaw Telegram bot not responding to any message.

The most basic issue: OpenClaw's Telegram channel isn't actually listening for messages. This happens when:

• The Telegram bot token is wrong or missing
• The channel config exists but isn't enabled
• The gateway started before the Telegram channel was configured

Diagnostic:

openclaw channels status

If Telegram shows "disconnected" or isn't listed at all, the channel isn't active.

Fix:

1. Verify your bot token:

curl https://api.telegram.org/bot<YOUR_TOKEN>/getMe

If this returns an error, your token is invalid. Create a new bot via @BotFather.

2. Check your channel config in openclaw.json:

{
  "channels": {
    "telegram": {
      "enabled": true,
      "bot_token": "your-token-here"
    }
  }
}

3. Restart the gateway:

openclaw gateway restart

Cause 2: OpenClaw Telegram Delayed Messages (Node 22+)

Symptom: Telegram messages delayed 5-10 minutes, or arriving in random bursts.

This is the most common cause on servers running Node.js 22 or later. The issue is a change in how Node 22 handles AbortSignal in fetch() calls.

OpenClaw's Telegram channel uses long-polling — it holds open an HTTP connection to Telegram's servers, waiting for new messages. Node 22's stricter AbortSignal handling sometimes causes these connections to silently fail without triggering a reconnect. Messages pile up on Telegram's side until OpenClaw eventually reconnects.

Diagnostic:

node -v

If you see v22.x or later, this is likely your problem.

openclaw logs --follow | grep -i "telegram\|abort\|polling"

Look for AbortError entries or gaps in polling activity.

Fix:

Option A — Switch to webhook mode (recommended for production):

{
  "channels": {
    "telegram": {
      "mode": "webhook",
      "webhook_url": "https://your-domain.com/api/telegram/webhook"
    }
  }
}

Webhook mode doesn't use long-polling, so it's immune to this issue. It's also faster — messages arrive instantly instead of waiting for the next poll cycle.

Option B — If you can't use webhooks (no public URL):

# Downgrade to Node 20 LTS
nvm install 20
nvm use 20
openclaw gateway restart

Cause 3: Telegram Bot No Messages in Groups

Symptom: Bot responds to direct messages but ignores messages in group chats — Telegram bot not working in groups.

After OpenClaw 2026.2.24, the Telegram channel requires explicit group message permissions. Without them, group messages are received by the gateway but filtered out before reaching the agent.

Diagnostic:

Send a DM to your bot — does it respond? Then send a message mentioning your bot in a group — does it ignore it?

Fix:

1. Make sure "Group Privacy" is disabled in BotFather:

   Open @BotFather → /mybots → select your bot → "Bot Settings" → "Group Privacy" → Turn off

2. Update your openclaw.json:

{
  "channels": {
    "telegram": {
      "allowed_updates": ["message", "callback_query", "inline_query"],
      "group_messages": true
    }
  }
}

3. Restart:

openclaw gateway restart

Cause 4: Telegram Bot Crash Loop (AbortError)

Symptom: OpenClaw Telegram bot stops responding every 20-30 minutes. Restarting fixes it temporarily.

This is a more severe version of Cause 2. Instead of messages being delayed, the entire Telegram polling process crashes with an AbortError, the channel disconnects, and it doesn't automatically reconnect.

Diagnostic:

openclaw logs --follow

Watch for lines like:

AbortError: The operation was aborted
  at fetch (/app/node_modules/...
Telegram channel disconnected

If this appears every 20-30 minutes, you're in a crash loop.

Fix:

The most reliable fix is switching to webhook mode (see Cause 2, Option A). If that's not possible:

1. Add a process manager that auto-restarts on crash:

# Using pm2
pm2 start "openclaw gateway start" --name openclaw --restart-delay=5000

2. Or add a health check cron job:

# Add to crontab: check every 5 minutes, restart if down
*/5 * * * * openclaw channels status --probe | grep -q "telegram.*connected" || openclaw gateway restart

These are workarounds. Webhook mode is the real fix.

Cause 5: Config Migration After Upgrade

Symptom: Bot was working, stopped after openclaw update.

OpenClaw's Telegram channel config has changed several times in 2026. Key changes:

After upgrading, your old config keys may not be recognized.

Fix:

openclaw doctor --fix
openclaw gateway restart

If openclaw doctor doesn't catch it, manually check your Telegram config against the current schema.

For a complete upgrade safety guide, see our OpenClaw migration guide.

When to Switch to Webhook Mode

If you're hitting Cause 2 or 4, webhook mode is the definitive fix. Here's the trade-off:

If your server has a domain name and SSL (which it should in production), use webhook mode.

Skip All of This

Every issue on this page has one thing in common: you're managing infrastructure instead of using your AI assistant.

ClawPod runs OpenClaw with webhook mode by default, automatic health monitoring, and zero Telegram configuration beyond pasting your bot token:

• Webhook mode — instant message delivery, no polling bugs
• 24/7 health checks — auto-restart if anything goes wrong
• No Node.js version headaches
• No config migration surprises
• Works in 30 seconds: paste token, click deploy

Get started →

Frequently Asked Questions

Why does my OpenClaw Telegram bot stop responding after a few hours?

Most likely an AbortError crash loop (Cause 4) caused by Node.js 22+ handling of long-polling connections. The bot works initially but the polling connection silently dies every 20-30 minutes. Switch to webhook mode for a permanent fix.

How do I check if my OpenClaw Telegram bot is receiving messages?

Run openclaw channels status --probe to verify the Telegram channel is connected and actively receiving. Then check openclaw logs --follow and send a test message — you should see it appear in the logs within seconds.

What is the difference between polling and webhook mode in OpenClaw?

Long-polling mode constantly asks Telegram "any new messages?" — it works without a public URL but is slower and buggy on Node 22+. Webhook mode tells Telegram to push messages to your server instantly — faster and more reliable, but requires a public URL with SSL.

My OpenClaw Telegram bot works in DMs but not in groups. Why?

After OpenClaw 2026.2.24, group messages require explicit permission. Disable "Group Privacy" in BotFather and add "group_messages": true to your Telegram channel config. See Cause 3 above for step-by-step instructions.

Having gateway connection issues instead? See Fix Gateway Disconnected 1008. For a full hosting comparison, check VPS Hosting Compared. New to OpenClaw? Read What is OpenClaw?

Last updated: March 2026