council-of-high-intelligence

Why we included this project

Decisions with real downside are hard precisely because the values in play pull against each other, and this project treats that as a reason to argue rather than to guess. It sets up a council of deliberately different analytical personas inside Claude Code, Codex, Gemini CLI, and OpenCode, keeps their opening positions independent, and forces them to challenge each other before synthesizing a verdict. Each member must restate the problem, analyze blind, and declare a final stance, with enforcement checks that flag premature agreement or unsupported confidence. The verdict leads with what remains unresolved and pairs any recommendation with dissent, kill criteria, and a concrete next step, so the output stays useful even when the council cannot agree. Engineering leads and technical founders deciding on pricing or architecture are the natural audience, and the quick and duo modes shrink the ceremony when a full council would be overkill.