Choose recurring work on Grafana Labs

Define a useful recurring result

Sat, 17 Oct 2026 02:03:35 +0100

Start with the morning decision

The checkout summary should help you decide whether errors need investigation before the working day begins. That purpose gives the prompt a clear scope: one service, a recent time window, and evidence about requests and errors.

A broad request such as “check everything” leaves the scope and output unclear. A focused summary is easier to assess because you know which question the result should answer.

Useful recurring work has an identifiable reader, a repeated question, and data that can answer it. For this example, the reader is the checkout service owner and the question is what happened over the last 24 hours.

Make the prompt self-contained

Sat, 17 Oct 2026 02:03:35 +0100

Give every run the context it needs

Each run creates a dedicated conversation. The checkout summary needs a self-contained prompt rather than references such as “the service above” or “yesterday’s query.”

This example names the scope and the evidence expected in the result:

Summarize the last 24 hours for the checkout service. Use metrics-main for metrics and logs-main for logs. Describe request volume, error rate, and notable error messages. Include the time range and links to supporting evidence. Separate missing data from healthy results. Don’t change resources.

Prompt structure

A relative window, such as “the last 24 hours,” stays relevant as the prompt repeats. Exact service and data source names reduce ambiguity, while explicit output criteria make the result easier to evaluate.

Resources

Sat, 17 Oct 2026 02:03:35 +0100

Resources

Review documentation about recurring tasks and repeatable prompts.

Documentation

Use Automations explains saved prompts and dedicated run conversations.
Automation best practices reinforces prompt scope and repeatability.