Deploy and Host Octop on Railway

For a self-hosted AI assistant dashboard on Railway with agents, memory and channel integrations, use Octop. The template pins the official Octop 1.0.2 image and initializes an administrator with a generated password. SQLite state, workspaces and secrets persist under /data/.octop. Configure model and channel access inside your deployment; installed software alone does not supply external AI accounts or provider credits.

About Hosting Octop

The template uses the official Octop 1.0.2 image pinned by digest. The /data Railway volume stores the SQLite control-plane database, agent workspaces and secrets under /data/.octop. A first administrator is initialized automatically. Open the generated HTTPS domain and log in with username admin and OCTOP_DEFAULT_PASSWORD from the Railway service variables. An existing account is preserved across restarts; editing the initial-password variable does not reset it.

No model API key is required to deploy or access the dashboard. Add your chosen LLM provider, model and key inside Octop, then create and configure your agents. Model usage is billed by the provider separately from Railway hosting. Optional Langfuse tracing is disabled by default.

Common Use Cases

Private AI assistants with persistent workspaces and memory.
Multiple users and specialized agents managed from the native dashboard.
Scheduled reports and recurring tasks.
Optional Telegram, Discord and other supported channel integrations, configured with your own credentials.

Dependencies for Octop Hosting

One Railway service using the official ghcr.io/tencentcloud/octop image.
One persistent volume mounted at /data; one replica for the SQLite deployment.
A model provider configured after deployment to run AI tasks.
Optional credentials for any channels and connectors you choose to enable.

Deployment Dependencies

Railway HTTPS routing and one persistent volume. The model provider is configured separately after deployment; keep access secrets stable across restarts.

Implementation Details

HTTP listens on 0.0.0.0:8088; Railway routes HTTPS to that port and checks /api/health. HOME=/data and OCTOP_HOME=/data/.octop keep workspaces, account data, secrets and settings on the volume. Graceful shutdown is configured. The Railway startup supervisor forwards termination signals to the official Octop process, allows the HTTP server to shut down, then closes lingering background workers. This avoids a deployment remaining active after an operator stop. Unexpected application exits still preserve their error status; the health endpoint remains the native Octop check. The image is pinned to sha256:c15e792f405a5b5aca8f3625adcc11f3421e13f6d01da7217ea18e6f0b98553f for a reproducible starting version.

Frequently Asked Questions

Can I configure the API key later? Yes. Deployment and admin login work before adding a provider. AI tasks need a configured, working model.

Does a restart erase my agents? Their database and workspaces live on the attached volume. Keep the volume when redeploying and back it up before upgrades.

Is this a separate sandbox for every user? No. Native multi-user permissions do not turn a shared Railway container into one isolated computer per user. Use trusted accounts and configure tool permissions deliberately.

Can I use PostgreSQL? Upstream supports PostgreSQL. This template intentionally starts with the native single-service SQLite layout; an external database is not required.

Why Deploy Octop on Railway?

Railway provides HTTPS routing, service variables, persistent volumes, health checks and deployment logs. This template supplies documented defaults, generated first-login credentials and the official Octop interface without requiring manual server setup. It is a community template, not an official TencentCloud Railway integration.

Source and license: TencentCloud/Octop, MIT.