Documentation

Set up Consilium and run your first deliberation — via the web app, CLI, or SDK.

All 8 modes explained: Quick, Council, Deep, Blind, Red Team, Jury, Market, Auto — phases, parameters, use cases.

The deliberation engine internals: state machine, voting algorithms, convergence detection, dissent clustering, confidence calibration.

5 providers, 15 models, complete pricing. Anthropic, OpenAI, Google, Groq (free), xAI — with judge priority and fallback system.

Full REST API: debates, deliberation, agents, personas, analytics. All endpoints with request/response schemas and SSE events.

12 commands, 20+ slash commands, 5 output formats. Interactive chat, benchmarks, sessions, decision extraction.

Sync and async clients, streaming, automatic retries, Pydantic types. Full method reference with examples.

Full TypeScript support, typed errors, SSE streaming, AsyncIterable. Complete type definitions.

System design, data flow, database schema, SSE streaming, auth flow, CI/CD pipeline, error handling.

6 pre-built templates: Code Review, Research, Risk, Healthcare, Legal, Finance — with rubrics, modes, and system prompts.

Product overview and core concepts

Frequently asked questions

Trust center, encryption, data handling

Peer-reviewed papers behind each mode

Visual walkthrough for non-technical visitors

Install, authenticate, and run your first debate

npm install and quickstart examples

pip install and async/sync usage

Cursor and Claude Code integration

All 8 modes explained with protocols

Supported models, pricing, and free tier

CLI tokens, Clerk JWT, API auth matrix

Technical deep dive for engineers

Endpoints, SSE streaming, server URLs

Cloud deployment and self-hosting