Documentation › Alert and act
Incidents and postmortems
Declare, follow, resolve, learn. The timeline fills itself and the AI writes the first draft of the postmortem.
Declare an incident
- From the Problems view: Open an incident on an alert. The title, severity (taken from the priority) and link to the monitor are filled in, and the timeline starts with the group's history.
- From the Incidents page: title, severity from 1 to 5, commander, customer impact.
Each incident receives a number within the workspace.
Follow
- Status and severity can be changed; each change goes into the timeline.
- Timeline: team notes, public updates, and, while the incident is open, every state change of the linked monitors.
- Links to monitors, hosts, services.
- Charts of the linked monitors over the incident's duration.
- Notifications: creation and status changes go to the workspace's incident channels.
Inform the public
Link the incident to one or more status pages: its title, status and public updates appear there, without any internal data.
The postmortem
Postmortem tab, Draft with AI button. HYPHA gathers the timeline, the linked alerts and the error logs of the hosts concerned during the incident, then S33D's model writes in Markdown:
- summary, impact, timeline;
- root cause;
- what worked, what did not;
- actions, as a table: action, owner, due date.
The text stays editable. The drafting is recorded in the timeline. The number of AI drafts depends on the plan: see pricing.
Updated: 19 November 2026