Is SuperClaude always invoked or "active", from any prompt I enter?

[jwagon]

If I have SuperClaude installed and I enter a prompt that doesn't begin with an /sc:<command> or calling a specific SC agent, does that mean SuperClaude isn't participating?

The reason I ask is that it's easy enough to begin a development task with an "sc" command or invoking an SC agent, but oftentimes there's a great deal of back and forth conversation during the same task. For example, if I begin by using /sc:troubleshoot "a bug" --type bug --fix, but then Claude does not successfully fix the bug, or--as often happens--it'll fix the original bug, but break something else--do I then need to again use the SC commands, like so: /sc:troubleshoot "you haven't fixed the bug" --type bug --fix? And if I don't, does that mean I'm no longer using SuperClaude at all?

Thanks.

[davidabreu7]

the /sc commands are always a safer bet, because it explicitly tells claude to use the superclaude instructions.

in the case of a followup discussion, for ex /sc:brainstorm or /sc:workflow, where the goal is to produce a plan I try to always do the followup commands using /sc until I get the result I want

but for implementation tasks using /sc:implement , If I want a more structured response or pass several fixes at once, I use /sc again, if its just a quick fix, I use regular prompts (without /sc)

in your case of /sc:troubleshoot I guess it falls on the same case as /sc:implement, if its a quick fix regular prompts for follow up are fine, but if you realize that it needs more context or deep analysis of the problem I would recommend using /sc:troubleshoot again
for ex: for solving a dificult bug I tend to tell claude to research the web for a fix (or another complex instruction), for that I use /sc:troubleshoot for a follow up prompt

in summary /sc commands use more tokens and fill the context window faster because you're using the full superclaude instructions, so its the developer decision to pass on every prompt the full set of instructions or not