Rate Limits Finally Mean Something

Cloudflare Workers rate limits have always been the kind of thing you ignore until you can't. A number on a dashboard that seems theoretical right up until your worker is hot and the counter is moving and suddenly you're doing arithmetic at 11pm.

What changed is MCP. Pair a Worker with an MCP server and put actual AI workloads behind it — not toy traffic, not a cron job firing twice a day, but agents hammering it with real momentum — and the rate limit panel starts telling you something useful. Days remaining at current pace. The trajectory, laid out plainly.

The decision it surfaces is clean: drop tasks and coast longer, or plus up. No ambiguity. The system is showing you exactly how fast you're burning and asking what you'd like to do about it.

That's genuinely rare for infrastructure dashboards, which typically specialize in telling you what already went wrong. This one tells you where you're headed.

MCP plus Workers is the pairing that makes this click into place. Something about the combination — stateful AI context moving through serverless execution, metered limits that are actually legible in real time — lands differently than any prior use case for either. Almost too cleanly.

Which means something will break eventually. That's a problem for a different dashboard.