I'm reading through the documentation and playing with the demo and believe there is a significant opportunity for an improvement to the user experience if responses from HtmlRequest actions could drive parameter autocompletion. Here are several ways I could think about doing such a thing:

1. subtype the html request for specifically handling graphql requests and drive the autocompletion of parameters from that response object based on the syntax of the graphql request - this could further be enhanced with stuff like graphql syntax highlighting, query autocompletion and much more
2. add a property to html response containing a json schema or perhaps an example json response and parse that to provide autocompletion - this could also apply to some file read operations
3. add a global setting area to the workflow for storing openapi specification documents and an openapi html request type action which allows the designer to choose which api and action in that api gets called and configures the action based on the specification document - it could be beneficial to centralize various api parameters and calling conventions somewhere like this in other ways for example if a workflow makes repeated calls to salesforce it would be advantageous to centrally configure certain parameters every call to salesforce would share and then use a dynamically generated action type SendToSalesforce instead of HtmlRequest

I apologize if such features already exist, but if they do then the documentation could use some improvement and they are not easily identifiable in the demo.