Skip to main content

How a change runs

A command that changes something, such as acknowledging an incident, asks MoleSignal through a managed tool, as Mole Agent and agent tools do. The organization’s Tool Policy sets how each tool runs, at Mole Agent → Settings → Tools & capabilities: Without a policy of its own, a tool runs by its risk: the default modes are confirmation for create_annotation and submit_search_job, and single approval for acknowledge_incident, resolve_incident, and run_synthetic_monitor. --reason says why, for the reviewers and the audit log. Confirming needs the MCP protocol revision 2026-07-28, which the CLI speaks when MoleSignal does. molesignal status reports the revision in use.

List incidents

The list shows the active incidents, newest first, with their status, severity, age, services, and summary. --status resolved or --status all lists others, --since 7d only those opened since then, and -n sets how many. molesignal incidents show prints one incident: its rule, times, services, labels, notes such as the runbook, and sampled traces. --rca adds the root-cause analysis, and --web opens the incident in MoleSignal:

Acknowledge and resolve incidents

Both go through the Tool Policy and take alerts.acknowledge.

See who is on call

The list shows who is on call for each enabled on-call schedule. --schedule keeps one schedule, and --at asks about another time, such as --at "2026-10-27 09:00".

Carry out approved changes

molesignal approvals lists the approval requests that wait for reviewers and those approved but not yet carried out, from Mole Agent and from agent tools and commands. --all lists decided ones too.
molesignal approvals show <approval> prints one request: its action, target, reason, impact, and reviews. molesignal approvals execute <approval> carries out an approved change; MoleSignal may ask to confirm it first:

Annotate dashboards

molesignal annotate adds an annotation at a moment or over a stretch of time, on every dashboard or on one:
The annotation takes dashboards.edit.

Run synthetic checks

molesignal synthetics lists the checks with their kind, state, and lifecycle. molesignal synthetics results <check> lists a check’s latest runs at each location, with their outcome, duration, and the reason a run failed. A check is named by its id or by its name. molesignal synthetics run runs a check now at each of its locations. With --wait, the command waits for the runs and exits with 1 when one failed or did not finish within --timeout (10 minutes by default), so that a pipeline can stop a release on it:
A run passes when its outcome is healthy or flaky. Running a check takes synthetics.manage; to run checks from CI without a reviewer, set run_synthetic_monitor to Automatic in the Tool Policy.
Last modified on October 26, 2026