Executions & revert

Every AgentSOAR action becomes an execution—a record of what was run, against which target, by whom, and what happened. This page covers the status lifecycle, why an action fails, where to find execution history, and how revert restores what an action changed.

The execution lifecycle

An execution moves through a defined set of statuses from the moment AgentSOAR accepts the request to its final outcome.

StatusMeaning
pendingThe execution is accepted and queued; AgentSOAR has captured its id before any provider call.
runningAgentSOAR is calling the provider to perform the action.
succeededThe action completed and is in effect.
failedThe action did not complete; see the failure reason.
revertedA previously succeeded action was successfully undone.
revert_failedA revert was attempted but did not complete; it can be retried.
awaiting_reauthThe credential is invalid or expired and must be re-authenticated before the action can proceed.
expired_awaiting_reauthThe re-authentication window passed without action; the run is closed out.

A run flagged awaiting_reauth waits for you to re-authenticate the credential (see Credentials & domains). If you don't act within the cutoff window, it becomes expired_awaiting_reauth.

A run shown as failed with an internal "superseded" marker is a deduplicated run, not a real failure—the same action was already applied by another execution, which holds the effect. Confirm how your deployment surfaces deduplicated runs.

Failure reasons

When an execution fails, AgentSOAR attaches a reason so you can tell a credential problem from a provider problem from a bad input:

ReasonWhat it means
credential_invalidThe provider credential was rejected—rotate or re-authenticate it.
cloud_api_errorThe provider's API returned an error while performing the action.
inventory_not_foundThe target resource could not be resolved in inventory.
input_invalidThe action's inputs failed validation.

Viewing execution history

Open Executions at /agentsoar/executions for the full history. Each row shows the capability, the target, the status, who ran it (an analyst or the Response agent), the timestamps, and the action log. You can filter by status—for example, to find everything awaiting_reauth—to focus on runs that need attention. Active containments currently in effect are also summarized on the Containment page.

How revert works

Most capabilities are revertible. Before changing anything, AgentSOAR captures revert context—the exact state it is about to modify, such as a host's original AWS security groups, the deny entries it added, or a user's prior account state. Revert uses that stored context to put things back:

  • block_ip removes the deny rules it added.
  • isolate restores the host's original network configuration.
  • power brings the host back online.
  • block_sender and block_email_domain delete the filter or blocklist entry.
  • disable_user restores the user's prior state.

A succeeded action can be reverted while it is within its revert window and the revert context is still available. A revert_failed run can be retried.

Revert depends on the original state being captured. AWS isolate is not revertible when the host was already isolated at run time, because the original security groups were no longer there to record. Plan isolation and revert deliberately.