Requirements

Local development

• Python 3.11 or later.
• Node.js 22.x (web/package.json engines and the Docker UI build stage).
• Git.
• Space for source files, outputs and the SQLite database.

Docker host

• Docker Engine and Docker Compose 2.24.0 or newer, or a compatible interface such as Portainer that accepts the same Compose file.
• A persistent volume writable by the container user 1000:1000.
• Host port 2507 available, unless customized.

Inference

TraDoc needs at least one LLM service:

• local, such as LM Studio or Ollama;
• OpenAI-compatible on your network;
• remote, such as OpenAI, Anthropic, Gemini, DeepSeek, OpenRouter, Minimax, Kimi or GLM.

A local model consumes RAM or VRAM. Requirements depend on the model, its quantization, context size and concurrency.

Browser

Use a modern browser with JavaScript, localStorage and server-sent event support. On mobile, the application remains usable through the navigation drawer when the server is shared on a trusted LAN.