holmesgpt

Why we included this project

Investigating a production outage usually means stitching together dashboards, logs, and alerts by hand. HolmesGPT automates that digging: it connects to your observability stack, Kubernetes, cloud providers, and databases, pulls the relevant metrics and logs, and works toward a root cause instead of just surfacing a status page. Its Operator mode runs continuously in the background, spots problems before customers do, and can message Slack or even open a pull request to apply a fix. It also writes findings back to Jira, PagerDuty, or OpsGenie, so the investigation results fit into your existing on-call flow. For SRE and platform teams that want an LLM agent operating inside real infrastructure rather than in a sandbox, it's worth trying against your own incident history.