TorpedoD/claude-council ? reverse-engineered prompt

Build me a Claude Code skill that helps me make big decisions with a small council of different viewpoints.

I want one slash command that takes a question, sends it through a few distinct advisor styles, lets them review each other, and then gives me a clear recommendation that still keeps any real disagreement visible. If the answer feels too neat, I want it to push back harder with an internal debate before wrapping up.

Please make it work in a few modes, so simple questions stay fast and harder ones get a deeper review. I also want it to save each run in a local journal so I can come back later, record what actually happened, and see patterns over time. It should output a clean report I can read plus a transcript for the full discussion.

Keep it all inside Claude Code, and if you need to check current docs or best practices online, go ahead.