renja-g/RiftRelay

AI Summary: RiftRelay currently lacks rate limiting enforcement immediately after startup, allowing a burst of parallel requests to bypass pacing until limits are dynamically learned. The issue highlights that multiple simultaneous requests will all pass through during this "cold start" period. The proposed solution is to implement default rate limits, such as "20 requests every 1 second" and "100 requests every 2 minutes," to prevent initial request flooding.