An LLM council that reviews your coding agent's every move

council of models for decision

The LLM Council works together to answer your hardest questions

Multi-LLM consensus extension for Gemini CLI. Inspired by Andrej Karpathy's llm-council.

Models debating in Cursor

Multi-model debate interface inspired by Andrej Karpathy Pro/Con/Judge agents with memory, uncertainty signals, and OpenRouter + Ollama support

Multi-headed cross-model code review skill for Claude Code. 6 AI advisors (Opus + Codex), 5 peer reviewers, chairman synthesis. Based on Karpathy's LLM Council methodology.

Autonomous writing optimizer: 4 LLMs compete, rank each other, and hill-climb toward the best argument — combining karpathy/autoresearch and karpathy/llm-council. Early development.

A tool for learning and reflection in a rapidly changing world. An electric bicycle for the mind.