Skip to main content
Automation scripts are workflows that run on a schedule without manual intervention. They use cron jobs to trigger Claudio at specific times, generating reports and performing maintenance tasks.

​
How automation works

1

Cron triggers the script

A scheduled cron job executes the bash script at the configured time.
2

Script calls Claudio

The script uses Claude Code CLI (claude -p) with a detailed prompt.
3

Claudio executes the workflow

MCPs are orchestrated to gather data, generate reports, and send outputs.
4

Results are logged

Execution logs are written to scripts/logs/ for debugging and monitoring.

​
Available automation scripts

​
Weekly bot report

File: scripts/weekly-bot-report.sh Generates and emails a weekly performance report for the WhatsApp bot. Schedule:
  • Frequency: Every Monday
  • Time: 9:00 AM
  • Cron expression: 0 9 * * 1
What it does:
1

Reads funnel metrics

Pulls data from the “Full Funnel Performance” spreadsheet (ID: 1PI2NnSzDhxCrb-NY18WgEiNpQyujgSG0RkPCw4c7IM8).
2

Calculates WoW changes

Compares last week vs. previous week with absolute delta and percentage change.
3

Generates HTML email

Creates a formatted report with tables, color-coded metrics (PropHero blue #2050f6), and green/red arrows for improvements/declines.
4

Sends via Gmail

Emails the report to [email protected] with subject “Reporte Semanal Bot - Full Funnel Performance”.
Metrics included:
  1. Nuevos Leads
  2. Group Calls Booked (Direct)
  3. New Lead No Booked
  4. Group Calls Booked by Bot (Total)
  5. First Time - Group Calls Booked by Bot
  6. % First Time
  7. Rescheduled - Group Calls Booked by Bot
  8. % Rescheduled
  9. Group Calls Booked (Total)
  10. % Group Calls Booked by Bot (from Total Group Calls)
  11. Group Call Attended (Total)
  12. Group Call Attended (Booked by Bot)
  13. Show Up Rate (Total)
  14. Bot - Show Up Rate
MCPs used:
  • Google Sheets (read metrics)
  • Gmail (send report)
Manual execution: 
./scripts/weekly-bot-report.sh

​
Monthly DS & AI report

File: scripts/monthly-ds-ai-report.sh Generates a comprehensive monthly review for the DS & AI squad and publishes it to Google Drive. Schedule:
  • Frequency: First day of each month
  • Time: 10:00 AM
  • Cron expression: 0 10 1 * *
What it does:
1

Calculates date range

Automatically determines the previous full calendar month.
2

Pulls Epics from ClickUp

Retrieves current quarter Epics from list 901215396098.
3

Pulls completed tasks

Fetches tasks completed or in progress during the previous month from sprint lists.
4

Pulls Initiatives

Gets Initiatives where DS & AI is involved from list 901213053436.
5

Calculates delivery metrics

Computes Say/Do Ratio, Carry-over, Bug Rate, Cycle Time, and Completion Rate per Epic.
6

Builds structured report

Creates a report following the meeting agenda: Wins & Learnings, Metrics, Initiatives & Epics Review, Recap.
7

Publishes to Google Drive

Creates a Google Doc in folder 1M7cqCebNXSJ-kcALWC3CVS44C6sAKcFo named “DS & AI Monthly Review — [Month] [Year]”.
Report sections:
  1. Wins & Learnings (0–10 min) — Top 3 wins, biggest learnings, general updates
  2. Metrics (10–30 min) — Outcome metrics, delivery metrics, speed & quality, team health
  3. Initiatives & Epics Review (30–50 min) — Planned vs delivered, epic status, risks & blockers
  4. Recap (50–60 min) — Summary, decisions, next steps
MCPs used:
  • ClickUp (Epics, tasks, Initiatives)
  • Google Docs (create report)
Manual execution: 
./scripts/monthly-ds-ai-report.sh
Related workflow: See Sprint report for detailed metric calculations.

​
Kill bot processes

File: scripts/kill_bot_processes.sh Maintenance script to safely terminate bot processes and clean up lock files. Purpose: Use this when bots (Telegram or Slack) become unresponsive or you need to restart them. What it does:
1

Finds bot processes

Searches for running Python bot processes (telegram/bot.py, slack/bot.py).
2

Terminates gracefully

Sends SIGTERM to allow clean shutdown, waits 1 second.
3

Removes lock files

Deletes /tmp/telegram_claude_bot.lock and /tmp/slack_claude_bot.lock.
4

Forces kill if needed

If processes remain after graceful termination, sends SIGKILL.
5

Verifies cleanup

Confirms all bot processes have been terminated.
Manual execution: 
./scripts/kill_bot_processes.sh
Output: 
🗑️  Lock file de Telegram eliminado
🗑️  Lock file de Slack eliminado
✅ Todos los procesos del bot fueron terminados

​
Log files

All automation scripts write logs to scripts/logs/:
ScriptLog File
Weekly bot reportweekly-bot-report.log
Monthly DS & AI reportmonthly-ds-ai-report.log
Logs include timestamps and full output from Claudio’s execution. Example log entry: 
2026-02-28 09:00:15 - Starting weekly bot report...
[Claudio output]
2026-02-28 09:02:43 - Weekly bot report completed.

​
Setting up cron jobs

To configure these scripts to run automatically: 
crontab -e
Automated scripts require your machine to be running at the scheduled time. For production use, deploy to a server with guaranteed uptime.

​
Requirements

  • Claude Code CLI installed at the path specified in the script
  • MCPs configured (Google Sheets, Gmail, ClickUp, Google Docs)
  • Workspace path correctly set in the script
  • Write permissions for log directory

​
Customizing scripts

You can modify the prompts in the scripts to:
  • Change recipients for email reports
  • Add or remove metrics from reports
  • Adjust report formatting (colors, tables, sections)
  • Change output destinations (different folders, channels)
Edit the script files directly and update the prompt passed to claude -p.

​
Monitoring

To monitor automation health: 
tail -n 50 scripts/logs/weekly-bot-report.log

crontab -l
Run the script directly to verify it works before relying on cron.
Configure email notifications for cron failures or add monitoring via your preferred tool.

​
Troubleshooting

  • Verify cron is enabled on your system
  • Check script has execute permissions (chmod +x script.sh)
  • Ensure absolute paths are used in the script and crontab
  • Check system logs: grep CRON /var/log/syslog
  • Check the log file for error messages
  • Verify MCP configurations are valid
  • Test MCPs manually with claude -p "test prompt"
  • Ensure API tokens and credentials haven’t expired
  • Verify the scripts/logs/ directory exists
  • Check write permissions on the logs directory
  • Run the script manually to see if it creates logs

Build docs developers (and LLMs) love

Get started for freeTalk to us