What problem does it solve? High-stakes decisions made from a single AI response risk blind spots and bias. This Skill convenes multiple LLMs to answer independently, rank each other's anonymized responses, and produce a chairman-synthesized final answer, giving you a structured multi-perspective deliberation. ## Core Features & Use Cases - Three-phase deliberation: independent responses, cross-model anonymized ranking, and chairman synthesis, with all phases persisted to disk verbatim. - Provider-agnostic configuration: works with Anthropic, OpenAI, OpenRouter, Fireworks, or any custom OpenAI-compatible endpoint selected via environment variables. - Wiki persistence: optionally writes the full transcript to a wiki page so the deliberation becomes searchable later. - Use Case: Before committing to a contested architecture decision, run the council on your plan so three to five models critique it independently, rank the critiques, and a chairman model delivers the final synthesized recommendation. ## Quick Start Ask the AI to run the LLM Council on your question, for example: run a council deliberation on whether we should adopt episodic memory in our agents and persist the transcript to the wiki.
