Show IH: I built a self-hosted on-call tool that tells you WHY the alert fired

I used to work in banking. Every alert meant opening my laptop, pulling up Grafana, then Dynatrace, then checking recent deploys — manually piecing together what happened. By the time I understood the cause, 20 minutes were gone.

Wachd automates that chase. When an alert fires, it pulls the last N commits from your repo, error logs from Loki/Datadog/Splunk, and metric history from

Prometheus/Grafana — sanitizes everything before it touches AI — then sends your on-call engineer a plain-English summary: probable cause, suggested action, and a link to similar past incidents.

Fully self-hosted. Deploys with a single Helm chart. Works completely offline with Ollama for air-gapped environments. Claude and OpenAI supported if you prefer cloud AI.

Why now: OpsGenie shuts down April 2027. Most alternatives are cloud SaaS with per-user pricing. Wachd is the open-source option for teams that can't send incident data to a third party.

Current state: v0.4.17, one external contributor with 8 merged PRs, 16 GitHub stars.

Looking for engineers dealing with alert fatigue or migrating off OpsGenie to break it — happy to walk you through setup personally.

Founders Program — 10 lifetime SMB licenses Giving 10 teams a free lifetime SMB license in exchange for real production feedback. Claim a slot on Discord: discord.gg/XAgXgnqxa

Also looking for 10 engineers to test this personally — I'll walk you through setup, 30 minutes, coffee gift card on me. DM me if interested.

GitHub: https://github.com/wachd/wachd Demo: https://youtu.be/VQAx-Kxhcoc