I built a multi-agent Think Tank that calls out my bad decisions
Three persona-driven agents (Saul, Mike, Gus) feed a single 'Cook' synthesis layer that explicitly looks for contradictions and produces one actionable move — e.g., 'you say AI career but your logs show 90% bizdev.' The repo is primarily prompt templates and architecture docs, so it’s a practical playbook for anyone wanting to wire personal data into an opinionated multi-agent loop rather than a full platform.
