Skip to content

Run an incident without losing the timeline

Bring incident alerts, discussion and a video call into one channel. Monitoring posts through an incoming webhook, pinned messages record decisions, and follow-up cards link to the messages that prompted them. Use the channel history to prepare the incident review.

Updated

operations

#incident · managed channel
  1. Monitoring · via webhookAlertAlert: error rate above threshold on checkout
  2. Call started in #incidentCallScreen share on, lobby and host controls ready
  3. Pinned decisionDecisionRoll back the last release, then watch the queue
  4. Card created from messageFollow-upAlert on queue depth · links back here

With Enclessa

Run an incident in an Enclessa channel

  1. Create a managed channel for incidents and give it a webhook

    Make a channel in the team that owns the service and set its mode to managed, so the room is searchable, exportable and reachable by automation.

  2. Let the alert open the incident

    Monitoring posts into the channel through the webhook, authored by a bot account that is a visible member of the room.

  3. Start the call from the room

    Type /call, or start it from the channel.

  4. Keep the decisions in the channel, not on the call

    Post each decision as it is taken and pin it.

  5. Turn the follow-up into cards that remember why

    Create the follow-up work as cards on the team’s board directly from the messages that produced them.

  6. Write the review by reading the room

    Search the channel by time, by author or by phrase, export what you need, and write the postmortem from a single ordered record.

Incidents with severity and on-call
Incidents with severity and who is on call.
Alerts in
An incoming webhook per channel, Slack-shaped payload
The call
Started in the room, with screen share and host controls
The record
Pins, threads, permalinks and full-text search
Follow-up
Board cards that link back to the message behind them
Room type
Managed, so the channel is searchable and exportable

Without one room

Four tools, one incident

  • The alert is in one place and the conversation is in another
  • The call has no thread
  • The timeline has to be rebuilt afterwards
  • The follow-up work drifts away from its reason

What it will not do

  • An incident channel has to be a managed room.
  • On-call and paging come from the Incidents module, which is in beta: alerts from Prometheus Alertmanager or Grafana open an incident in a room, a rotation decides who is paged, and the page escalates until somebody acknowledges.
  • Calls are not recorded to a downloadable file, so the call is not part of the written record.
  • Nothing pushes events out of Enclessa. Alerts come in over a webhook; if another system needs to know an incident opened, it has to ask, or you tell it.

FAQ

Incident response: questions

  • Yes. Anyone who may manage a channel can create an incoming webhook for it, which returns a URL and a bearer secret. The payload follows Slack’s shape, so most monitoring tools that already post to Slack need only a new URL.
  • Yes, in beta. The Incidents module takes alerts from Prometheus Alertmanager or Grafana, opens one incident per alert group in a room, keeps daily or weekly on-call rotations with overrides, and pages whoever is on call, escalating to the next person until somebody acknowledges.
  • Yes, provided the channel is a managed room. Managed channels are indexed with full-text search, can be exported, and keep pins, threads and permalinks — so the ordered record already exists without anybody maintaining one.

Which plan this needs

The channel, the threads and the call are on every plan, including Free. Incoming webhooks, bot accounts and task boards are on the Team plan and above, so an incident channel where humans coordinate works from day one, and the alerting and follow-up half arrives when you move up.

Compare plans

Bring your team over this afternoon

Free for up to 10 people, with encrypted direct messages and video calls. No payment card, no time limit, no server to run.

For businesses and organisations only — not available to consumers.