Bug Report: API Rate Limits Seem Inconsistent and Poorly Documented

We are developing an integration that relies heavily on the API, but we are running into what seems like inconsistent rate limiting. The documentation states a certain number of requests per minute, but we are getting '429 Too Many Requests' errors long before we hit that documented threshold. It's making our development process incredibly difficult. Furthermore, the response headers don't consistently include 'Retry-After' information, making it impossible to handle these errors gracefully. We need clearer, more accurate documentation on how the rate limits are calculated (is it per API key, per IP, per account?) and reliable feedback from the API when we do hit a limit. We have checked our code extensively and are confident we are not exceeding the stated limits. This is a critical issue for our integration project.