Ongrid is an open-source ops AI agent that understands your infrastructure, finds the root cause of incidents and fixes them, right from Slack, Telegram, Lark or DingTalk. Written in Go and released under AGPL-3.0, it has about 1,200 stars on GitHub and is actively developed in 2026.
When an alert fires at 3 a.m., the investigation is usually the slow part: correlating metrics, logs and traces, working out the blast radius, then finding the change that caused it. Ongrid automates that loop. The agent reasons over your existing observability stack (Prometheus, Grafana, Loki, Tempo and OpenTelemetry), your service topology and a RAG knowledge base of docs and code, then reports a grounded root cause rather than a transcript dump. Alerts can trigger investigations automatically, and with remote execution the agent can apply fixes under your control. It plugs into Kubernetes environments and the chat tools your team already lives in, and it is fully self-hosted, so production data stays on your infrastructure.
Installation is designed to be quick: a one-line installer targets mainstream Linux distributions, and Ongrid can start from an existing Prometheus, Loki and Tempo setup, adopting the observability data a team already collects instead of demanding a rip-and-replace migration before it delivers value.
Ongrid is free and open source under the AGPL-3.0 license. You install it on your own servers with a one-line installer; your only costs are infrastructure and the LLM usage you configure.
Ongrid is for SRE and DevOps teams who want an on-call copilot that investigates alerts end to end in their existing observability stack instead of another dashboard to watch.