# Know whether an automation actually worked

> Separate scheduled attempts, useful output, and successful delivery so unattended tasks can be trusted.

Canonical: https://hermeswatcher.com/workflows/reliable-recurring-automation  
Category: Development & operations  
Basis: An editorial blueprint: an original, source-informed approach. Check commands against the official docs for your version.

## The useful result

A small status record showing when the task ran, what it produced, where it went, and what failed.

## Before you begin

- One clearly bounded recurring task
- A definition of a useful output
- A place to retain run receipts and failure details

## How to approach it

1. **Define success before scheduling**: Write down the expected artifact, minimum completeness, freshness, and destination. Define what an acceptable no-change result looks like.
2. **Record the attempt**: Keep the start time, source window, configuration scope, and completion status. Enabled scheduling is only an intention to run.
3. **Validate the output**: Check the actual artifact for required sections, source references, and freshness. A process exiting successfully is not enough.
4. **Check the destination**: Verify a saved file, delivery receipt, or readback from the intended channel. Keep output success separate from delivery success.
5. **Bound retries and preserve the last good result**: Repeated failures should stop and become visible. Retrying without changing a broken configuration usually adds cost without useful progress.

## A prompt to adapt

Replace the brackets and run a bounded trial first.

```text
Inspect the latest run of [task]. Report attempt status, output validity, and delivery status separately, with timestamps and supporting artifacts. Preserve the last good output. Identify the smallest concrete cause of failure and do not broaden the task.
```

## When it does not work

If a provider or source is unavailable, report that exact block. Do not mark a schedule healthy because its timer continues to tick.

## Check the result

Can someone determine what happened from the saved evidence without trusting the agent's final success message?

## Official reference (Nous Research)

- [Scheduled tasks](https://hermes-agent.nousresearch.com/docs/user-guide/features/cron)
- [Troubleshooting](https://hermes-agent.nousresearch.com/docs/guides/troubleshooting-agent-quality)

---

HermesWatcher is an independent guide. It is not affiliated with or endorsed by Nous Research. The [official Hermes Agent documentation](https://hermes-agent.nousresearch.com/docs) and [repository](https://github.com/NousResearch/hermes-agent) from Nous Research are the authority; each page here lists the official sources it was written from. If this site and the official docs disagree, trust the official docs. Index for agents: https://hermeswatcher.com/llms.txt