Dashu UI

A ready-to-run Next.js interface powered exclusively by the published Dashu SDK packages for core orchestration, Next.js routes, PostgreSQL, managed AI, OpenRouter, and local OpenAI-compatible models.

Quick start

git clone <your-repository-url>
cd dashu-ui
npm install
cp .env.example .env
npm run dev

Configure only these variables:

DASHU_DATABASE_URL=postgres://readonly_user:password@host:5432/analytics
DASHU_STORAGE_DATABASE_URL=
DASHU_CLOUD_URL=https://dashu.vercel.app
DASHU_CLOUD_CREDENTIAL=your-installation-token

• DASHU_DATABASE_URL is the read-only PostgreSQL database Dashu queries.
• DASHU_STORAGE_DATABASE_URL is optional. When set, users, credentials, conversations, saved queries, dashboards, and settings are stored in its dashu_ui_documents table.
• If the storage URL is empty, the same application data is stored as private JSON files in .dashu/.
• DASHU_CLOUD_URL defaults to https://dashu.vercel.app.
• DASHU_CLOUD_CREDENTIAL stays server-side and enables managed AI and Pro features. Users never enter a license key or AI credential in the UI.

The published @rophpad/dashu-provider-openrouter and @rophpad/dashu-provider-openai-compatible adapters are direct dependencies, so no extra installation is required for OpenRouter, Ollama, vLLM, LocalAI, llama.cpp, or another OpenAI-compatible endpoint. This ready-to-run four-variable configuration selects @rophpad/dashu-provider-managed; changing provider selection later only requires wiring one of the already-installed package adapters, not implementing a provider in this UI.

Open http://localhost:3000. The first visitor creates the initial user account.

Docker

docker compose up --build -d

Docker publishes port 3000 and persists JSON fallback storage in the dashu-data volume. HTTPS session cookies are detected automatically from the request or reverse proxy headers.

Validate

npm run typecheck
npm run build