​

Routine Examples

​

Example 1: Community Pulse

​

The script

#!/usr/bin/env python3
"""ADW: Community Daily Pulse -- Daily community report via Pulse"""

import sys, os
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.dirname(
    os.path.abspath(__file__)))))
from runner import run_skill, banner, summary

def main():
    banner("Community Pulse", "Discord - Activity - Sentiment - Support | @pulse")
    results = []
    results.append(run_skill(
        "pulse-daily",
        log_name="community-daily",
        timeout=600,
        agent="pulse-community"
    ))
    summary(results, "Community Daily Pulse")

if __name__ == "__main__":
    try:
        main()
    except KeyboardInterrupt:
        print("\nCancelled.")

​

How it flows

1. Runner calls claude --print --agent pulse-community with a prompt to execute /pulse-daily
2. Claude loads the Pulse agent system prompt
3. Pulse reads the pulse-daily SKILL.md instructions
4. The skill tells Pulse to:
   • Fetch Discord messages via the Discord API
   • Fetch WhatsApp messages via the Evolution API
   • Analyze activity, sentiment, top topics, support questions
   • Read the HTML template from .claude/templates/html/
   • Fill the template with collected data
   • Save to workspace/community/reports/daily/
   • Send a Telegram notification with the summary
5. Runner logs the result and updates metrics

​

The schedule entry

daily:
  - name: "Community Pulse"
    script: community_daily.py
    time: "20:00"
    enabled: true

​

Output

workspace/community/reports/daily/[C] 2026-04-08-community-pulse.html

​

Example 2: Financial Pulse

​

The script

#!/usr/bin/env python3
"""ADW: Financial Pulse -- Daily financial snapshot via Flux"""

import sys, os
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.dirname(
    os.path.abspath(__file__)))))
from runner import run_skill, banner, summary

def main():
    banner("Financial Pulse", "Stripe - Omie - MRR - Churn | @flux")
    results = []
    results.append(run_skill(
        "fin-daily-pulse",
        log_name="financial-pulse",
        timeout=600,
        agent="flux-finance"
    ))
    summary(results, "Financial Pulse")

if __name__ == "__main__":
    try:
        main()
    except KeyboardInterrupt:
        print("\nCancelled.")

​

What the skill does (step by step)

1. Collect Stripe data — MRR from active subscriptions, today’s charges (succeeded vs failed), 30-day churn rate, 7-day refunds, new customers
2. Collect Omie data — overdue receivables, payables due in 7 days, pending invoices
3. Consolidate transactions — all financial movements for the day
4. Classify health — green (healthy), yellow (warning), or red (risk) based on churn rate, overdue amounts, and MRR trend
5. Generate alerts — payment failures, overdue accounts, missing invoices, anomalies
6. Fill HTML template — reads .claude/templates/html/custom/financial-pulse.html and replaces placeholders
7. Save to workspace/finance/reports/daily/[C] YYYY-MM-DD-financial-pulse.html
8. Notify via Telegram — sends a short summary with MRR and alert count

​

Adapting an Example for Your Use Case

​

1. Identify the components

​

2. Create or reuse a skill

​

3. Create the routine script

• The skill name in run_skill()
• The agent name
• The log name
• The banner text

​

4. Register the schedule

my-routine:         ## Description of my routine (@agent)
	$(PYTHON) $(ADW_DIR)/examples/my_routine.py

​

5. Test manually

make my-routine
# or
python3 ADWs/routines/examples/my_routine.py

​

Using the create-routine skill