Hiya Ale! We don't have great support for custom "clientless" requests atm, but you could (depending on scale requirements and complexity) set up a python3 server like a live client passthrough,. The client constructs the requests and sends them to the API in a way that it can understand and trust(ish). I've used curl to trigger flows just fine, yet others have had trouble getting their requests to pass heavier requests past Cloud's endpoint protection tooling