A minimal, three-stage deliberation dashboard for multi-agent reasoning.
LLM Council is a research tool designed to visualize the internal deliberation process of large language models. Inspired by hierarchical decision-making, it moves beyond simple chat interfaces to provide a transparent, multi-stage pipeline for complex problem solving.
The Analyst role produces independent, high-confidence responses. Each input is analyzed for technical depth and executive summary suitability.
The Skeptic role performs peer-review audits. Responses are anonymized and subjected to logical fallacy checks, quality scoring, and gap analysis.
The Chairman role aggregates the findings and critiques into a final verdict, clearly distinguishing between consensus and remaining uncertainties.
- Information Density: Maximum signal, minimum noise.
- Glass Box UI: Structural transparency at every reasoning stage.
- Visual Silence: A neutral, IDE-inspired design system for deep focus.
- Python 3.10+
- Node.js 18+
- OpenRouter API Key
# Clone
git clone https://github.com/thesohamdatta/LLM-Council.git
# Configure
cp .env.example .env
# Start Backend
uv sync && uv run python -m backend.main
# Start Frontend
cd frontend && npm i && npm run dev