All products

Assess Rootly service health with AI

Compare errors, latency and traffic with a suitable baseline, then place active incidents and verified deployments alongside the result. Gaps stay visible in the assessment.

Assess Rootly service health with AI when an empty incident list does not explain what customers are experiencing. The service-health skill examines error rate, latency and traffic across a stated window and compares them with a similar workload or period. It distinguishes histories recorded in Rootly from measurements of current service behaviour.

A quiet endpoint at midnight can be normal. Zero traffic at peak checkout time deserves a different question. The assessment also checks whether a spike recovered and whether there is evidence of impact. Verified deployments can supply context, while a source commit by itself is not proof of a release.

What it does

  • Compare the service with its normal regional pattern.
  • Look for sustained changes and recurring small spikes.
  • Check the coverage of the incident history.
  • Review confirmed recent deployments alongside the signals.
  • Identify missing evidence before declaring recovery.

Install and use

  1. Install the skill and connect Rootly’s official read-only MCP with an authorised account for live records. Supply permitted monitoring charts or snapshots for the selected service and interval.
  2. Ask, “Assess checkout over the last thirty minutes against a comparable baseline. Separate measured traffic from what the incident records say.”
  3. Check the time window and evidence coverage before using the result in an operational update.

FAQ

Source

Made by Rootly. Original skill: service-health, published on GitHub under the Apache-2.0 licence.

View on GitHub