During yesterday’s Robot Assistant Webinar, MacSparky mentioned the concept of gotchas — and it stuck with me. I made a quick note in Drafts, my Robot Assistant picked it up, turned it into an action, and suggested we add a Gotchassection to every skill. I accepted, but honestly I’m still not fully sure how to use it in practice.
As I understand it, gotchas are about catching known skill flaws and knowing how to correct them — but how does that actually work during skill execution?
Can anyone expand on this with a concrete example or two?
I wasn’t on the call, but: after any edits to a skill, I have agents run through scenarios to see where the skill runs into issues and refine the language. Then run through scenarios again. It sounds like Gotchas could be a place to store little abrasions as they’re noticed by either you or the agent, for a similar refinement run-through.
Is It is like a small repository where you store stuff that is to be addressed at a later stage?
Maybe for batching stuff to be adjusted, optimized, or rectified…