Deploy Mobius — Personal AI OS

Deploy and Host Möbius on Railway

Möbius is a self-hosted AI agent workspace designed to do more than chat. It can build apps, work with files, maintain memory, use skills, run goals and scheduled tasks, coordinate subagents, and connect to coding agents such as Claude Code and Codex.

This template deploys Möbius as a single persistent service on Railway, giving you a personal AI operating environment that runs on infrastructure you control.

About Hosting Möbius

Möbius combines the AI agent, interface, memory, applications, files, skills, and workspace into one self-hosted environment.

Instead of treating every conversation as an isolated session, Möbius is designed around persistent work. Its agent can retain useful context, build mini-apps, manage longer-running goals, execute scheduled tasks, and delegate work to other agents.

This template uses Möbius' built-in SQLite database and persistent application storage, so no external PostgreSQL, Redis, or other database service is required.

The upstream project explicitly supports a single application container for managed hosting platforms such as Railway.

Included Architecture

                    ┌──────────────────┐
                    │       User       │
                    └────────┬─────────┘
                             │
                             ▼
                   Railway HTTPS Proxy
                             │
                             ▼
                    ┌──────────────────┐
                    │      Möbius      │
                    │                  │
                    │ Agent / Chat     │
                    │ Apps / Skills    │
                    │ Memory / Tasks   │
                    │ Files / Goals    │
                    └────────┬─────────┘
                             │
                             ▼
                       Persistent /data
                        ┌─────┴─────┐
                        ▼           ▼
                     SQLite      Workspace

Common Use Cases

• Run your own persistent AI agent
• Work with Claude Code or Codex from a web interface
• Build mini-apps directly from conversations
• Maintain AI memory across sessions
• Create reusable skills and workflows
• Run scheduled AI tasks
• Turn to-do lists into longer-running goals
• Delegate work to subagents
• Maintain files and project context alongside the agent
• Build a personal AI workspace accessible from multiple devices
• Self-host an AI operating environment on infrastructure you control

Möbius vs Traditional AI Chat

Möbius is particularly useful when you want an AI agent to become part of an ongoing workspace rather than simply answering isolated prompts.

Persistent Storage

Möbius stores persistent application state under:

/data

The built-in SQLite database is located at:

/data/db/ultimate.db

The persistent volume can contain:

• Application database
• Agent memory
• Workspace files
• User-created apps
• Skills
• Project state
• Uploaded files
• Configuration
• Other Möbius application data

Persisting the entire /data directory ensures application state survives container restarts and redeployments.

AI Agents and Providers

Möbius is designed to work with coding agents including:

• Claude Code
• OpenAI Codex

Provider access is connected after deployment rather than being hard-coded into the Railway template.

This means you do not need to provide an OpenAI or Anthropic API key before deploying the service.

After opening Möbius, connect the provider or coding-agent account you want to use and continue setup from the interface.

Apps on the Fly

One of Möbius' distinctive features is its ability to build applications directly from a conversation.

For example, you can ask the agent to create:

• A news dashboard
• Habit tracker
• Internal utility
• Data viewer
• Personal productivity tool
• Simple game
• Specialized AI application

The agent can build the application alongside the conversation and continue refining it based on your feedback.

Memory and Skills

Möbius is designed around persistent agent context.

Its workspace can retain:

• Decisions
• Preferences
• Project information
• Learned corrections
• Agent skills
• Useful context from previous work

Skills act as reusable playbooks that help the agent perform recurring work more consistently.

Goals, Tasks, and Workflows

Möbius supports longer-running work beyond individual chat messages.

Goals

Goals allow the agent to continue pursuing a larger objective across multiple steps.

Tasks

Scheduled tasks can run recurring work such as:

• Morning briefs
• Daily checks
• Reports
• Follow-up actions

Workflows and Subagents

Möbius can delegate parts of a job to other agents while keeping the execution visible through a workflow timeline.

This makes the platform useful for more complex coding, research, and automation workflows.

Getting Started

1. Deploy Möbius.
2. Wait until the application passes its health check.
3. Open the public Railway domain assigned to Möbius.
4. Complete the initial setup.
5. Connect Claude Code, Codex, or another supported provider workflow.
6. Start a conversation with your agent.
7. Create apps, goals, tasks, skills, or workflows as needed.
8. Continue using the same persistent workspace across future sessions.

Health Check

This template uses Möbius' native readiness endpoint:

/api/ready

The application listens on:

8000

Railway can use the readiness endpoint to verify that a new deployment is responding correctly before routing traffic to it.

Resource Considerations

Möbius is significantly more resource-intensive than a simple web dashboard because it can run coding agents, subprocesses, applications, workflows, and other agent tooling.

The upstream deployment configuration allows up to approximately 6 GB of memory for the application.

For Railway deployments:

• Smaller instances may work for light testing.
• Around 4 GB is a more practical starting point.
• Around 6 GB better matches the upstream deployment configuration.
• Heavier coding and agent workloads may benefit from additional memory and CPU.

Actual requirements depend on the workload being executed.

Security Considerations

Möbius is a powerful agent environment and may contain credentials, code, private files, conversations, and application data.

For production or personal environments:

• Keep the application secret persistent.
• Protect access to the public Möbius URL.
• Treat the /data volume as sensitive.
• Review integrations before connecting external services.
• Keep provider credentials private.
• Back up important workspace data.
• Review agent permissions before exposing sensitive infrastructure.

The Möbius agent can be configured with elevated privileges inside its environment, so the deployment should be treated as a trusted personal workspace rather than a publicly shared application.

Dependencies for Möbius Hosting

• Möbius — AI agent workspace and application platform
• SQLite — built-in persistent application database
• Persistent Volume — stores database, memory, apps, files, and workspace state
• Railway Public Networking — HTTPS access to Möbius
• AI Agent Provider — connected by the user after deployment
• Health Check — application readiness monitoring

No Redis, PostgreSQL, MySQL, or external database is required for this template.

Deployment Dependencies

• Möbius GitHub: https://github.com/mobius-os/mobius
• Möbius Website: https://mobius.you/
• Möbius Container: https://github.com/mobius-os/mobius/pkgs/container/mobius
• Railway Networking: https://docs.railway.com/networking
• Railway Volumes: https://docs.railway.com/volumes

Why Deploy Möbius on Railway?

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.

By deploying Möbius on Railway, you get a persistent self-hosted AI agent workspace that can build apps, maintain memory, run goals and scheduled tasks, coordinate agent workflows, and keep your projects and files together on infrastructure you control.