How should an AI SRE end its run when a disk filling up turns out to be nothing?

Tares lets an AI SRE end that run with the conclude tool set to no_op, which records no finding. When a disk filling up is a real problem, the agent concludes with a finding instead, carrying a verdict, a headline and a next step. The agent states the outcome on purpose, so a false alarm leaves nothing behind for a person to read.

How it works

An agent that ends with a conclusion gets the conclude tool, which ends a run on purpose. It has two outcomes:

• no_op: there is nothing to report, so no finding is recorded. Use it when the disk alert turns out to be nothing.
• A finding with a verdict about a named entity, such as the host or volume that is filling up.

A finding also carries two fields:

• headline: one line of up to 100 characters, saying what was found in the reader’s terms.
• next_step: up to 300 characters, saying what a person should do. It stays empty when nothing needs doing.

The project’s Overview shows the headline and next step on each card, and agents are told to give both. When a run has no headline, Tares takes it from the note’s heading, bold lead or first sentence. It takes the next step from the text after “Next step:”, “Next:” or “Recommendation:”.

So the choice is simple. If the disk growth is real, record a finding with a verdict, a headline and a next step. If it is not, conclude with no_op and nothing is recorded. For the wider picture of how agents investigate, see AI SRE.

Set it up

1. Open the agent form and go to the How it ends section.
2. Tick Ends every run with a conclusion (concludes: true). The agent then always gets the conclude tool, whatever its prompt says, and Tares adds the instructions for using it. Your prompt only describes the job, for example checking whether the disk growth is real.
3. Under Verdicts it can give, add your own words, each with when to give it. You can add up to 10. Each is one word (letters, digits, dashes) and listed once. Do not add no_op, because nothing to report is always possible.
4. Save the agent. In YAML the agent looks like this:

agents:
  - name: triage
    trigger: watch-web
    concludes: true
    verdicts:
      - { verdict: investigate, when: one entity looks like a problem and needs a closer look }
    prompt: |
      ...

5. Check the agent’s summary, which shows how it ends.

With a verdict list, the tool accepts only those words and tells the model what each one means. A finding must carry a verdict. If the agent gives a word that is not on the list, or leaves the verdict out, Tares sends it back to the agent to fix within the same run. Without a list, a finding can carry any verdict, or none.

If a run stops without concluding, Tares asks once more with conclude as the only choice. If you leave Ends every run with a conclusion off, the agent’s last message is its finding, with no verdict. For the full reference, see How a run ends.

What you get

• A false alarm ends cleanly. The agent calls no_op and no finding is recorded, so the project’s Overview stays free of noise.
• A real problem ends with a finding that a person can act on: a verdict about a named entity, a headline of up to 100 characters and a next step of up to 300 characters.
• Every run ends on purpose. With Ends every run with a conclusion on, the agent always has the conclude tool, and Tares asks again if a run stops without using it.
• Your own verdict words keep findings consistent. Other agents that act on those words, such as through handoffs, can rely on them.

When something else fits

• If you want a record that an alert was noise, give a finding with a verdict word for it instead of no_op. The Root cause after a handoff preset comes with resolved for something that was noise or is already over, and rca for when you found what is failing and why.
• If you do not need a verdict at all, leave the conclusion box off. The agent’s last message is then its finding, with no verdict.

Related