Hey everyone 👋 I built InariWatch after getting paged at 2am for a failed deployment that… realistically, AI could’ve fixed in seconds. So I started working on something a bit different from traditional monitoring. Instead of just alerting you, InariWatch tries to actually resolve the issue: • It correlates alerts across GitHub, Vercel, Sentry, and Datadog • Runs AI analysis (you bring your own key — Claude, OpenAI, etc.) • Writes a fix and opens a PR • Waits for CI and auto-merges if it’s safe The goal is simple: fewer alerts, more fixes. A few things I cared about while building it: * Fully open source (MIT) * Self-hostable * No lock-in (BYOK for AI) * Safe auto-merge guardrails (CI + confidence thresholds) Would really love feedback from this community — especially: 👉 Would you trust something like this in production? 👉 What would make you feel comfortable with auto-fixes? Happy to answer anything 🙏

InariWatch

Open source AI monitoring that fixes issues while you sleep

Open source AI monitoring that fixes issues while you sleep