3 ms·
My approach is to ask Claude to plan anything beyond a trivial change and I review the plan, then let it run unsupervised to execute the plan. But I guess this
by igor47 1y ago
My approach is to ask Claude to plan anything beyond a trivial change and I review the plan, then let it run unsupervised to execute the plan. But I guess this does still leave me vulnerable to prompt injection if part of the plan is accessing external content
- ares623 1y agoJust don’t think about it too much. You’ll be fine.
- abathologist 1y agoWhat guarantees do you have it will actually follow the stated plan instead of doing something else entirely?
- igor47 11mo agoin my experience, it does tend to follow the plan pretty well.