Skip to content

Incident management

From detection to resolution, automatically.

Mihari connects your monitors, telemetry, error tracking and GitHub repositories to detect, diagnose and resolve incidents. They open on their own from a failing check and resolve when it recovers.

incidents / inc-3012Acknowledged

The complete incident lifecycle. Every step connected, on one timeline.

  • Detection

    Monitors and heartbeats detect failures in real time and open the incident on their own.

  • Correlation

    The AI SRE correlates logs, metrics and error tracking data to point at the root cause.

  • Resolution

    The incident resolves when the check recovers. A proposed fix can go to GitHub as a draft pull request, validated by your CI.

  • Postmortem

    The AI SRE drafts a root-cause analysis and postmortem from the incident timeline.

Resolve on monitor recovery
Auto
Timeline per incident, end to end
1
Lifecycle states — open, acked, resolved
3
Alert channels for escalations
8

Automatic resolution

Resolve incidents before your team wakes up.

Link a monitor or heartbeat to telemetry and error tracking sources, and the AI SRE investigates the incident with that context, then posts the root cause to the timeline. When the check recovers, the incident resolves without human intervention.

  • Link monitors and heartbeats to telemetry sources (logs, metrics, traces)
  • Connect error tracking to identify the faulty code path
  • Incidents resolve on their own when the monitor recovers
monitors / api-gateway-healthDown since 14:32
5xx · current rate144/min

Linked sources

  • Telemetry · logs & metricsapi-gateway / production Error spike
  • Error trackingapi-gateway / Sentry SDK New errors
  • GitHub repositoryacme/api-gateway Connected

AI SRE · root cause

Error spike correlates with deployment v2.8.1 (deployed 14:30 UTC). NullReferenceException in UserService.cs:142 matches the HTTP 503 pattern from telemetry logs.

GitHub integration

Proposed fixes, as draft pull requests.

Link monitors and sources to a GitHub repository. When the root cause is in your code, the AI SRE proposes a patch that you can open as a draft pull request on the linked repository.

  • Link monitors, error tracking and telemetry sources to specific repositories
  • Fix proposals based on the error tracking root cause analysis
  • Draft pull requests go through your CI and code review before merge
acme / api-gateway · pull request #4218Draft

fix: add null check in user.service.ts

  • CI passing
  • Opened from the AI SRE investigation
  • Review requested · @yuki

Public status · your domain

A status page your users actually trust.

Branded, and hosted on your own domain.

status.acme.com
Acme Cloud
All systems operational
Live · 30s
  • Web app
    90d uptime · 99.998%
    Operational
  • Public API
    90d uptime · 99.991%
    Operational
  • Auth / OAuth
    90d uptime · 99.842%
    Degraded
  • Webhooks
    90d uptime · 99.996%
    Operational
  • Dashboards
    90d uptime · 99.999%
    Operational
  • Background jobs
    90d uptime · 99.988%
    Operational
OAuth elevated latency · investigating · posted 4m ago

In the product

An incident, from detection to resolution

Every step lands on one timeline: detection, escalation, paging, recovery.

incidents / checkout-api
The timeline of a resolved incident on Checkout API · POST /v3/charge in Mihari: automatic detection, escalation policy triggered, on-call engineer paged by push and email, page acknowledged, status updates and an internal note on the rollback, then recovery and resolution
Demo dataSeeded demo workspace, not a customer's data.
incidents
Mihari incident list sorted by start date: resolved and open incidents opened automatically by monitors and heartbeats, with their status, severity, source, start time and duration
Demo dataSeeded demo workspace, not a customer's data.

Frequently asked questions

Run your own post-incident review from our blameless postmortem template, in Markdown.

How does automatic incident resolution work?
An incident opens when a monitor or heartbeat fails, and resolves on its own when the check recovers. When the monitor is linked to telemetry sources (logs, metrics) and error tracking, the AI SRE investigates with that context and posts the root cause to the incident timeline. When the cause is a code change, it proposes a fix that you can open as a draft pull request on your connected GitHub repository.
Which Git providers are supported?
GitHub. Connect it through the Mihari GitHub App, then link repositories to your monitors, error tracking and telemetry sources.
Are the proposed patches safe to merge?
Mihari opens a draft pull request that requires your team's approval. The patch is validated by your CI checks, tests, and review process just like any other code change.
Can I customize the postmortem template?
Yes. Download our default postmortem template in Markdown format and customize it to match your organization's incident review process.
What telemetry and error tracking sources are supported?
Mihari supports OpenTelemetry-compatible sources for logs, metrics, and traces. For error tracking, point your Sentry SDKs at Mihari's built-in Application Errors feature.

Stop firefighting, start auto-resolving.

Connect your monitors, telemetry and GitHub repositories. Let Mihari detect, diagnose and document incidents for you.