Until now, catching a Workers exception in production meant wiring up your own logging or bolting on a third-party error tracker. Issues folds that into the platform: it collects unhandled exceptions, deduplicates them into grouped failures with the surrounding request context, and surfaces them without an extra SDK or sampling setup. It's in open beta, so it's free to turn on and kick the tires while Cloudflare irons out the edges. The part worth noticing is the handoff. Rather than stopping at a dashboard, Issues can package a failure — stack trace, logs, distributed traces, and application context — and send it to a coding agent to dig in and propose a fix as a pull request. That's observability aimed less at a human squinting at a graph and more at an agent closing the loop from alert to patch, which is the direction Cloudflare has been pushing its Workers tooling all week. Early days and beta-shaped, but it's a concrete take on what 'self-healing' production might actually look like.