Backend / infrastructure
Every request passes one token bucket
Four API clients compete for tokens in a browser mirror of the Go limiter.
JS mirror
The production service is Go, standard library only. Run it with
go run . and read src/gatekeeper/limiter.go. This static page mirrors that code so it can be hosted without a backend.
t+00:00
Requests per second
admitted vs rejected, sampled every 500ms of simulated time
allowed
429
Bucket state
rate 9/s, burst 14
Controls
edit the simulated gateway
Traffic sits just under the limit.
Decision stream
newest first, one row per request
/metrics
same text format as the Go handler