Deploy and Host OpenLore with Railway

About Hosting OpenLore

OpenLore serves a durable knowledge base to people and AI agents over a web UI, HTTP API, MCP, and SSH. This template deploys the official ghcr.io/aakarim/openlore:latest image built from the aakarim/OpenLore repository with Railpack.

Why Deploy OpenLore on Railway

The template creates one OpenLore service with:

an HTTP domain routed to internal port 8080 for the web UI, API, and /mcp endpoint;
a Railway TCP proxy routed to internal port 2222 for SSH;
a persistent 1 GB volume mounted at /var/lib/openlore for documents, server data, and the SSH host key;
keyless access disabled, no public docset, and an onboarding identity whose writable home is /user/onboarding.

Common Use Cases

Use OpenLore as a durable knowledge server for AI agents, an MCP endpoint, an HTTP-accessible documentation store, or an SSH-native workspace that agents and people can query with standard command-line tools.

Dependencies for OpenLore Hosting

Deployment Dependencies

Enter your complete SSH public key in OPENLORE_ONBOARDING_PUBLIC_KEY during deployment. OpenLore installs it declaratively for the onboarding identity on every start. Public-key comments are supported.

Railway assigns the SSH proxy its own public hostname and port. Find them under OpenLore → Settings → Networking → TCP Proxy, then connect with:

ssh -p  onboarding@

Railway does not reserve public port 22. To use port 22, put a raw TCP load balancer in front of the assigned Railway endpoint.

The latest image is rebuilt from OpenLore's main branch. To receive new images automatically, open OpenLore → Settings → Source → Configure Auto Updates after deployment and choose Anytime. Railway currently does not preserve image auto-update schedules when generating a reusable template, so this setting must be enabled once per deployed project.