🦞 Morning Report System cron + telegram

📋 Overview

The Morning Report system runs three automated cron jobs that check server health, security status, and available updates. Reports are delivered to Telegram with clickable inline buttons for applying updates.

Cron Schedule

🔘 Inline Buttons critical knowledge

[inline_button:Apply Updates:apply_system_updates]

Working Command

openclaw message send \
  --channel telegram \
  --target "85126311" \
  --buttons '[[{"text": "✅ Apply Updates", "callback_data": "apply_system_updates"}]]' \
  -m "Your message here"

Button JSON Format

# Single button
--buttons '[[{"text": "Button Text", "callback_data": "action_name"}]]'

# Two buttons side by side
--buttons '[[{"text": "Yes", "callback_data": "yes"}, {"text": "No", "callback_data": "no"}]]'

# Two rows of buttons
--buttons '[[{"text": "Row 1", "callback_data": "r1"}],[{"text": "Row 2", "callback_data": "r2"}]]'

Key Points

• [[...]] — outer array is rows, inner array is buttons per row
• text — what the user sees on the button
• callback_data — identifier sent back when button is clicked
• Use single quotes around the JSON to avoid shell escaping issues

📝 Telegram Message Formatting critical

🦞 <b>Morning Health Report</b>
✅ <b>SSH:</b> 211 attacks blocked

🦞 Morning Health Report

✅ SSH: 211 attacks blocked (fail2ban active)
🔴 OOM: 1 kill yesterday (opencode)
📦 Updates: 25 packages ready

What Works

• Plain text — just write normally
• Emoji — ✅🔴⚠️📦🦞 work great
• Newlines — use blank lines for spacing
• Code blocks — avoid in reports, hard to read on mobile

What Doesn't Work

• HTML tags — <b>, <i>, <code> show as text
• Markdown — **bold** shows as literal asterisks
• [inline_button:...] — not parsed, shows as text

📊 Daily Security Report Script

Location: /home/clawdija/asistent/v2/workspace/daily-security-report.sh

What It Checks

1. SSH attacks — Failed password + invalid user attempts (24h)
2. fail2ban — Total bans, currently banned IPs
3. OOM kills — Processes killed by out-of-memory (24h)
4. Package updates — Security, regular, and vendor updates pending
5. Failed services — systemd units in failed state
6. Disk usage — Root partition percentage
7. Memory usage — RAM percentage
8. Gateway status — Is OpenClaw gateway running?

Update Signal

When updates are available, the script outputs:

[UPDATES_AVAILABLE:25]

The agent reads this signal and knows to send a message with an "Apply Updates" button.

⚙️ Cron Job Configuration

Location: /home/clawdija/asistent/v2/cron/jobs.json

Morning Health Report Job

{
  "id": "c7c04792-76d1-46b6-a925-27151dfe12b8",
  "name": "morning-health-report",
  "enabled": true,
  "schedule": {
    "kind": "cron",
    "expr": "0 6 * * *"
  },
  "payload": {
    "kind": "agentTurn",
    "message": "Good morning! Run the daily security report:\n\n1. Execute ~/asistent/v2/workspace/daily-security-report.sh...\n2. Check for OpenClaw updates...\n3. Run memory-recall.sh recent 3...\n\nINLINE BUTTONS (IMPORTANT):\nIf updates available, use this EXACT command:\nopenclaw message send --channel telegram --target \"85126311\" \\\n  --buttons '[[{\"text\": \"✅ Apply Updates\", \"callback_data\": \"apply_system_updates\"}]]' \\\n  -m \"Your plain text message\"\n\nDO NOT use [inline_button:...] or HTML tags.\n\nCALLBACK HANDLING:\n- When callback_data \"apply_system_updates\" received:\n  a) React with 👍 first\n  b) Run ~/asistent/v2/workspace/system-update.sh\n  c) Report results"
  },
  "delivery": {
    "mode": "announce",
    "channel": "last"
  }
}

Key Configuration Points

• schedule.expr — Cron expression (0 6 * * * = 06:00 daily)
• sessionTarget — isolated (fresh session each run)
• delivery.mode — announce (send result to user)
• delivery.channel — last (use last active channel)

Managing Cron Jobs

# List all cron jobs
openclaw cron list

# Edit a job
openclaw cron edit <job-id> --message "New prompt..."

# Run a job manually (debug)
openclaw cron run <job-id>

# View run history
openclaw cron runs --id <job-id>

🔔 Callback Handling

When a user taps an inline button, the callback_data is sent to the agent. The agent must handle it appropriately.

Current Callbacks

Callback Response Flow

1. User taps button → Telegram sends callback to OpenClaw
2. Agent receives → callback_data in message context
3. Acknowledge first → React with 👍 to confirm receipt
4. Execute action → Run the appropriate script
5. Report results → Send summary back to user

📁 Related Files

🔧 Troubleshooting

Button Not Rendering

Symptom: Message shows [inline_button:...] as text

Solution: Use openclaw message send --buttons '[[...]]' instead of text placeholder

HTML Tags Showing

Symptom: <b> shows literally in message

Solution: Remove HTML tags, use plain text with emoji

Duplicate Reports

Symptom: User receives multiple identical reports

Solution: Check if multiple cron jobs with same schedule exist: openclaw cron list

Button Callback Not Working

Symptom: User taps button but nothing happens

Solution: Verify callback handling is in cron job prompt, check agent received the callback

Cron Job Not Running

Symptom: No report at expected time

Solution: Check gateway is running, verify cron job enabled: openclaw cron list

📖 Quick Reference

Send Message with Button

openclaw message send \
  --channel telegram \
  --target "85126311" \
  --buttons '[[{"text": "✅ Action", "callback_data": "action_name"}]]' \
  -m "Plain text message with emoji 🦞"

Run Report Manually

~/asistent/v2/workspace/daily-security-report.sh

Apply Updates Manually

~/asistent/v2/workspace/system-update.sh

Check Cron Status

openclaw cron list
openclaw cron runs --id c7c04792-76d1-46b6-a925-27151dfe12b8

Edit Cron Job

openclaw cron edit c7c04792-76d1-46b6-a925-27151dfe12b8 --message "New prompt"