Build
Automation
Explain a schedule, write a no-code recipe, and turn a sample webhook into a contract you can trust.
How Automation agents work
You put in
Explain a schedule, write a no-code recipe, and turn a sample webhook into a contract you can trust.
Agent
Automation
You get out
A prompt you can copy, a file you can download, and the steps to run it today.
Automation breaks when the schedule is a rune and the recipe lives in someone's head. A cron line, a GitHub Actions schedule, or a chain of no-code steps will run at the wrong time for months because nobody wanted to be the person who decoded it. This shelf turns those artifacts into sentences and steps you can rebuild.
The schedule translator explains a cron expression, a systemd OnCalendar, or a workflow schedule, including the timezone trap, and it writes the expression you meant if you describe it in English. The recipe writer takes a repetitive job and returns a trigger, filters, actions, and a failure step you can recreate in n8n, Make, or Zapier. The webhook contract turns a sample payload into required fields, an idempotency key, retry behavior, and a signature check described as a defense, not as an attack.
These agents do not log into your automation account and do not store credentials. If a step would need a secret, the recipe names the secret's slot and tells you to set it in the tool, not in the chat.
Questions about Automation
Which automation tools are the recipes for?
The steps are tool-shaped but plain: trigger, filter, action, failure. You can rebuild them in n8n, Make, Zapier, or a small script. The prompt does not emit a vendor-only binary.
Will it explain a cron I already deployed?
Yes. Paste the line and the timezone the host uses. The agent says what the line does and the next runs in that zone, and it flags a line that is ambiguous.
Does the webhook agent show how to forge a signature?
No. It tells you how to verify a signature you receive, using the signing secret stored in your tool. It does not write a bypass.