It's 2 AM. Your phone buzzes. Sentry caught an error spike. You open your laptop, check GitHub — CI is red. Check Vercel — deploy failed. Check your uptime monitor — endpoint is down. Three tabs, three dashboards, three tools telling you something is wrong, but none of them talking to each other.
You spend 20 minutes figuring out it was a bad deploy that introduced a null pointer. You revert the PR, redeploy, go back to sleep. Total damage: 40 minutes of your night, a broken SLA, and a vague sense that this will happen again next week.
We built InariWatch because we were tired of that story.
## What is InariWatch?
InariWatch is an open-source DevOps monitoring platform that watches your entire stack — GitHub CI, Vercel deployments, Sentry errors, Datadog metrics, uptime endpoints, npm packages, Postgres — and surfaces everything in one place.
But monitoring is the easy part. The real problem is what happens after the alert fires.
## The problem with existing tools
Most monitoring tools are great at detecting problems. They're terrible at understanding them.
When Sentry fires an alert, it doesn't know your CI is also red. When Vercel tells you a deploy failed, it doesn't know that Sentry started seeing errors 3 minutes before the deploy. These tools live in silos. You're the one connecting the dots at 2 AM.
The result: alert fatigue. Engineers start ignoring notifications because there are too many of them, they lack context, and acting on them requires too much manual work.
## How InariWatch is different
InariWatch treats your stack as a single system, not a collection of independent tools.
When an alert comes in, InariWatch correlates it with signals from your other integrations. It knows that a Sentry error spike happening within 5 minutes of a Vercel deploy failure is probably the same incident — not two separate issues.
From there, if you have GitHub connected, InariWatch can go further: it reads the diff that caused the problem, generates a fix using AI, runs it through a self-review pass (a second AI call that critiques the fix before it's committed), checks safety gates — CI status, confidence score, number of lines changed — and if everything passes, opens a pull request automatically.
You wake up to a merged fix instead of a pager alert.
## Safety first
We know "AI merges code automatically" sounds terrifying. That's why we built the safety gate system:
- CI must be green on the fix before anything merges - Confidence score ≥ 90% — the AI must be highly confident in its fix - Self-review score ≥ 70/100 — a second AI call reviews and critiques the fix - ≤ 50 lines changed — large changes always require human review - Auto-merge must be explicitly enabled per project — it's off by default
After a merge, InariWatch monitors your stack for 10 minutes. If error rates go up or uptime drops, it automatically opens a revert PR.
You're always in control.
## Open source
InariWatch is fully open source. The CLI runs locally — no account needed. The web dashboard is the cloud layer for teams that want centralized visibility, notifications, and the auto-remediation pipeline.
We built this because we wanted it to exist. If you're tired of context-switching between dashboards at 3 AM, we think you'll want it too.
---
You can self-host it today or use the cloud version at app.inariwatch.com. The code is on GitHub.