Andreas tbh #4 seems pretty heavy to be a workaround. It's not so trivial to implement and will change the behaviour of the agent quite a bit. I am wondering if we should just provide a sanity check tool...I know it doesn't solve the problem but at least it can detect it. Let me think about it more, feel free to ping me on irc, we can maybe brainstorm a little more.
Andreas tbh #4 seems pretty heavy to be a workaround. It's not so trivial to implement and will change the behaviour of the agent quite a bit. I am wondering if we should just provide a sanity check tool...I know it doesn't solve the problem but at least it can detect it. Let me think about it more, feel free to ping me on irc, we can maybe brainstorm a little more.