At some point I switched to full grug-dev mode and cut everything that wasn’t essential. No agents, no setup, no monitoring, no accounts. One request in, parallel checks from multiple places, raw results back. The goal is that you can curl it during an incident and immediately answer "is this broken everywhere, or just from here?"

There's also an OSS sidecar for people who want to self-host monitoring, but it still needs some tuning.

I'm not fully sure about the next step yet, probably a small paid tier to cover server costs and buy a coffee, but for now the public API has fairly generous limits even without an API key if you want to play with it