Hey Product Hunt 👋 I'm Dev, Co-founder of Steadwing - the team behind OpenAlerts. We built OpenAlerts because AI agents fail silently - and that's a huge problem. If you're running agentic frameworks, you've probably experienced this: → Your LLM provider returns a 401 and your agent just… stops → A session gets stuck in a loop and nobody notices for hours → A tool call fails mid-session and your agent starts hallucinating There was no dedicated alerting layer for this. Traditional monitoring tools don't understand agent-specific failures. So we built one. OpenAlerts watches your agent in real-time and alerts you the moment something breaks - via Telegram, Discord, Slack, WhatsApp, or Signal. What makes it different: → Built specifically for agentic frameworks, not retrofitted from generic monitoring → 8 real-time alert rules covering LLM errors, stuck sessions, gateway outages, queue buildup, and more → Optional LLM-enriched alerts that explain what went wrong and suggest a fix → Built-in live dashboard for event timelines, logs, and health status → Currently supports OpenClaw, OpenManus, and Nanobot → Zero config to start - if you have a paired channel, it just works It's fully open source (Apache-2.0), framework-agnostic by design, and more adapter plugins are on the way. I'll be here all day answering questions and hearing your feedback. Would love to know what frameworks you'd like supported next 🚀

OpenAlerts

Monitoring & Alerting layer for Agentic frameworks

Monitoring & Alerting layer for Agentic frameworks