Build, Test, Deploy

Let us build a simple daily briefing skill from scratch. This skill will run every morning at 7 AM, check the weather for your city, check your calendar for the day, and send you a summary message in Telegram. It combines three data sources into one actionable message — and once it is deployed, it runs every day without intervention.

1. 1. Create the Skill File Create a new file in your skills directory. Define the skill metadata: name, description, and trigger type. For a daily briefing, use a cron trigger set to "0 7 * * *" (every day at 7 AM).
2. 2. Write the Logic In the skill handler, call the weather API for your city, call the Google Calendar API for today's events, and format both into a clean summary. Use the AI model to generate a natural-language briefing from the raw data.
3. 3. Test Locally Run the skill manually from your terminal using the OpenClaw test command. This executes the skill logic without waiting for the cron trigger. Verify the output looks correct and all API calls succeed.
4. 4. Deploy Register the skill in your OpenClaw configuration. Restart the agent process. The skill is now live — it will fire at 7 AM tomorrow and every day after.
5. 5. Verify in Production Wait for the first scheduled run (or temporarily set the cron to run in 2 minutes). Check Telegram for the briefing message. If it arrives, your skill is deployed and running. If it does not, check the agent logs for errors.

Once your first skill is running, the pattern repeats. Every new automation follows the same structure: define a trigger, write the logic, test locally, deploy, verify. The skills compound over time. After a month, your agent might run a dozen daily automations — briefings, inbox management, CRM updates, social media monitoring — all without you doing anything beyond the initial setup.