Deploy and Host Hermes Agent on Railway

A one-click Railway template for Hermes Agent.

About Hosting Hermes Agent

Hermes is a self-hosted AI agent with a web dashboard, messaging integrations, skills, memory, and persistent sessions. This template deploys the pinned official Nous Research image directly. The repository contains reproducible Railway infrastructure, not a fork of Hermes.

Why Deploy Hermes Agent on Railway?

Official nousresearch/hermes-agent image pinned to a release and digest
Hermes gateway and dashboard managed by the upstream s6 supervision tree
50 GB persistent volume mounted at /opt/data
Railway HTTPS domain routed to dashboard port 9119
/api/status deployment health check
Native Hermes basic authentication

Railway supplies the HTTPS endpoint, durable storage, health-aware deployments, and restart handling without maintaining a server or a custom container image.

Common Use Cases

Run a persistent personal AI agent from a browser
Connect one agent to supported messaging platforms
Keep Hermes memory, skills, sessions, and configuration between deployments
Evaluate Hermes without installing its runtime on a local machine

Dependencies for Hermes Agent Hosting

A Railway account
Credentials for a model provider supported by Hermes

Deployment Dependencies

The template uses the official Hermes container image and a Railway persistent volume. It does not provision a database or require a custom build.

Deploy

Click Deploy on Railway above and provide these variables:

Variable	Value
`PORT`	`9119`
`HERMES_DASHBOARD`	`1`
`HERMES_GATEWAY_BOOTSTRAP_STATE`	`running`
`HERMES_DASHBOARD_BASIC_AUTH_USERNAME`	Your dashboard username, such as `admin`
`HERMES_DASHBOARD_BASIC_AUTH_PASSWORD`	A long, unique password
`HERMES_DASHBOARD_BASIC_AUTH_SECRET`	A random signing secret; generate one with `openssl rand -hex 32`

After Railway reports a successful deployment, open the generated domain, sign in, and configure your model provider and messaging integrations in the Hermes dashboard.

Basic auth is convenient for a personal deployment. For a broader internet-facing deployment, configure Hermes dashboard OAuth or OIDC after the first deploy.

Persistent state

Hermes stores configuration, secrets, sessions, memories, skills, logs, and per-profile tool state under /opt/data. The attached Railway volume preserves that data across deployments and restarts.

Do not mount a volume over /opt/hermes; that path contains the immutable application installation.

Maintain the template

Railway infrastructure is defined in .railway/railway.ts. Install the pinned Railway IaC SDK, review the plan, then apply it:

npm ci
npm run railway:plan
npm run railway:apply

Hermes itself is pinned by both release and image digest. To upgrade it, update the image reference in .railway/railway.ts, review the upstream release notes, run the plan, and verify /api/status after deployment.

The Railway template metadata can be updated with the Railway CLI:

railway templates publish hermes-agent-railw-1 \
  --category AI/ML \
  --description "Deploy a secure, persistent Hermes Agent from the official image" \
  --readme-file README.md

Run the pinned image locally

Copy the image reference from .railway/railway.ts, then run:

IMAGE='nousresearch/hermes-agent:vYYYY.M.D@sha256:IMAGE_DIGEST'
docker run --rm "$IMAGE" version

For a persistent local gateway, follow the official Hermes Docker guide.