Deploy and Host Agent Zero on Railway

Agent Zero is an open-source, general-purpose AI agent. It runs tasks by writing and executing code in its own Linux environment, searches the web, keeps memory between chats and can use tools and sub-agents.

About Hosting Agent Zero

This template runs the official agent0ai/agent-zero image with a persistent volume at /a0/usr for settings, memory, chats and your files. The agent can run shell commands inside its container, so the web UI is protected by a login that is generated for you: open the service in Railway, go to Variables and read AUTH_LOGIN and AUTH_PASSWORD, then sign in with them. Do not remove these variables.

After signing in, open Settings and choose an LLM provider and enter its API key (OpenAI, Anthropic, OpenRouter, a local Ollama endpoint, etc.). No model key is included.

Common Use Cases

Automate research, scripting and data tasks with an agent that has a real shell
Build long-running assistants with persistent memory
Experiment with multi-agent workflows

Dependencies for Agent Zero Hosting

One service with a volume; an API key from an LLM provider (entered in the UI)

Deployment Dependencies

Why Deploy Agent Zero on Railway?

Railway gives you a public HTTPS URL, a persistent volume and automatic restarts with no server to maintain, and keeps the agent in an isolated container away from your own machine.