CQL WorkbenchDocs
Docs contents
Get started

Quick start

Bring up the full stack locally with docker compose, paste your first clause, and watch it become CQL you can verify.

Prerequisites: Docker (with Compose), plus Node.js and npm for the frontend dev server.

Installation

  1. 1

    Set environment variables

    Copy .env.example and generate a JWT_SECRET. Under NODE_ENV=production, the backend refuses to start if it is empty or still the default.

    Terminal
    cp .env.example .env
    openssl rand -hex 32  # use as JWT_SECRET
  2. 2

    Start the services

    Starts nginx, backend, postgres, cql-engine and RustFS together. The first start on a fresh database volume applies every migration automatically.

    docker compose up -d
  3. 3

    Start the frontend

    The Vite dev server proxies /api to nginx on :8081.

    cd frontend && npm ci && npm run dev
    # http://localhost:5173
  4. 4

    Configure the LLM

    Sign in as an admin and add a model profile under 管理 → LLM 設定 (Admin → LLM settings). Several profiles can fail over by priority (fallback) or split traffic by weight (load_balance).

  5. 5

    Analyze your first clause

    Paste text or upload a PDF or docx. The system splits it into tasks and analyzes each part into IR. When it finishes, open the review page to accept, reject or mark each condition not applicable.

Change before go-live: the default admin is created by ensureSeedAdmin, and every password in .env.example is a development default.

Ports

ServiceAddress
Workbench (dev)http://localhost:5173
Full stack (nginx)http://localhost:8081
RustFS consolehttp://localhost:8081/rustfs/console/

The backend has no volume mount, so rebuild it after code changes:

docker compose build backend && docker compose up -d backend

Next steps