> ## Documentation Index
> Fetch the complete documentation index at: https://docs.acornops.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Configure auto-triage

> Start durable investigations automatically for qualifying Kubernetes and VM issues

Auto-triage lets AcornOps begin a target investigation when a new issue
lifecycle reaches your configured severity. The feature is experimental and
uses the same target chat, approval, audit, and retention paths as a manual
investigation.

## Enable automatic investigations

1. Open a Kubernetes cluster or virtual machine.
2. Open **Settings** and find **Automatic investigations**.
3. Choose the minimum issue severity.
4. Choose how the investigation may use write-capable tools.
5. Add optional investigation instructions.
6. For a Kubernetes cluster, configure namespace eligibility if needed.
7. Save the settings.

Saving does not start investigations for issues that already exist. After you
enable the feature, use the separate action for current eligible issues if you
want to queue them. Future matching issue lifecycles queue automatically.

Each issue lifecycle has at most one automatic investigation. Repeated
observations of the same lifecycle do not create duplicate conversations.

## Limit Kubernetes issue eligibility

Kubernetes targets provide three eligibility settings:

| Setting                       | Behavior                                                                                                          |
| ----------------------------- | ----------------------------------------------------------------------------------------------------------------- |
| Namespace include list        | When non-empty, only issues in the listed namespaces are eligible. An empty list allows every observed namespace. |
| Namespace exclude list        | Removes listed namespaces after include filtering. Exclusion always wins.                                         |
| Include cluster-scoped issues | Controls issues that do not belong to a namespace.                                                                |

Use lowercase Kubernetes namespace names. You can provide up to 100 names in
each list.

These settings only decide which observed issues may start an automatic
investigation. They do not expand AgentK collection, Kubernetes RBAC, or tool
access. Configure AgentK namespace scope separately in Cluster Settings. See
[Connect a Kubernetes cluster](/use/connect-kubernetes#namespace-scope).

## Virtual machine eligibility

Virtual-machine automatic investigations are target-wide. Severity is the only
issue eligibility filter; Kubernetes namespace settings do not apply.

## Choose action safety

The saved action mode cannot grant more access than the target, connected
agent, workspace role, and selected diagnostic tools already permit.

* **Diagnose only** exposes no write tools.
* **Ask before changes** pauses for approval before a permitted write.
* **Allow changes** skips confirmation only when the target policy also allows
  it.
* **Follow target settings** inherits the target's effective write policy.

If required AI credentials, the target connection, or diagnostic tools are not
ready, AcornOps keeps the issue waiting and shows the setup reason instead of
starting a partial investigation.

## Review activity

The target Settings page shows active and waiting counts. Open the target's
issues to review the associated investigation, approval, findings, or retry
state. Automatic conversations also appear under **Investigations** in target
chat history.

Disabling automatic investigations prevents queued work from starting and
stops new issue lifecycles from being queued. It does not erase completed
investigation history.