Thinking about our need to mock the request/response from kai/rpc-server because we haven't integrated it has become apparent that we may actually want to fake this most of the time so that we don't actually need to talk to the LLM unless we really want to.
We need a mechanism in our extension to put ourselves in a dev mode of sorts where we skip the long running calls so that we can iterate faster.
Thinking about our need to mock the request/response from kai/rpc-server because we haven't integrated it has become apparent that we may actually want to fake this most of the time so that we don't actually need to talk to the LLM unless we really want to.
We need a mechanism in our extension to put ourselves in a dev mode of sorts where we skip the long running calls so that we can iterate faster.