This could be handled at the platform level but could also probably be handled by the proxy: Sometimes an app like WordPress tries to talk to the same endpoint in several subsequent page loads. I may not want to lose the option by entering an invalid endpoint (see #1), but I shouldn't get endlessly spammed by requests for the same thing.
I propose the proxy should temporarily remember what endpoints it has requested and not gotten a capability for, and for the rest of the grain session should not re-request it. A user could restart the grain if they need access to something they've already rejected.
This could be handled at the platform level but could also probably be handled by the proxy: Sometimes an app like WordPress tries to talk to the same endpoint in several subsequent page loads. I may not want to lose the option by entering an invalid endpoint (see #1), but I shouldn't get endlessly spammed by requests for the same thing.
I propose the proxy should temporarily remember what endpoints it has requested and not gotten a capability for, and for the rest of the grain session should not re-request it. A user could restart the grain if they need access to something they've already rejected.