Deploy and Host Hermes Agent on Sealos

Hermes Agent is an open-source AI agent with persistent memory, a web dashboard, and messaging gateways. This template deploys the official Hermes Docker runtime on Sealos Cloud with a protected dashboard and a private data volume.

About Hosting Hermes Agent

The official Hermes Docker Compose bundle contains a gateway and a dashboard. The template keeps those roles in one StatefulSet Pod so both processes share the official /opt/data home directory on persistent storage. The dashboard serves the browser UI on port 9119; the gateway's authenticated API server is available on port 8642 and through the /v1 ingress path.

Hermes stores sessions, configuration, skills, and memory under HERMES_HOME. Sealos provisions a 1 GiB persistent volume for /opt/data, an HTTPS ingress, and the application entry in Canvas. No database or object-storage service is required by the official Docker bundle.

Common Use Cases

• Personal AI workspace: Keep sessions, skills, and memory available across restarts.
• Remote agent dashboard: Use the protected web dashboard from a browser.
• Messaging gateway: Connect Telegram, Discord, Slack, WhatsApp, Signal, or email channels configured through Hermes.
• Automation and scheduling: Run recurring tasks through Hermes' built-in gateway and scheduler.

Dependencies for Hermes Agent Hosting

The template includes the official Hermes Agent image, a gateway container, a dashboard container, a persistent volume, a Service, an HTTPS Ingress, and a Canvas App entry.

Deployment Dependencies

• Hermes Agent website - Product and installation information
• Hermes Agent documentation - Configuration and feature guides
• Hermes web dashboard guide - Remote dashboard authentication
• Hermes API server guide - Bearer-authenticated API
• Hermes Agent repository - Source code and issues

Implementation Details

Architecture Components

• Gateway container: Runs the official gateway run command and exposes the authenticated API server on 8642.
• Dashboard container: Runs the official dashboard command on 9119 and provides the browser UI.
• Persistent volume: Mounts /opt/data for HERMES_HOME, sessions, skills, configuration, and memory.
• Service and Ingress: Routes the dashboard at / and the API server at /v1 over HTTPS.

Configuration

• dashboard_username and dashboard_password enable the documented username/password provider for a non-loopback dashboard.
• api_server_key is required by the official API server for every bind address.
• openai_api_key is optional and enables model-backed chat actions when an OpenAI-compatible provider is selected.
• dashboard_secret is generated by Sealos and keeps dashboard sessions stable across restarts.

Resource Profile

These are the initial personal low-load tiers for the official image. Adjust them from Canvas after observing actual model and tool workloads.

License Information

Hermes Agent is licensed under the MIT License.

Why Deploy Hermes Agent on Sealos?

Sealos is an AI-assisted Cloud Operating System built on Kubernetes. Deploying Hermes Agent on Sealos provides one-click provisioning, automatic HTTPS, persistent storage, pay-as-you-go resource usage, and Canvas operations for logs and resource changes.

• One-click deployment: Start from the App Store template without hand-writing Kubernetes manifests.
• Persistent agent state: Keep the Hermes home directory across restarts and upgrades.
• Secure public access: Use the documented dashboard credentials and API bearer key behind Sealos HTTPS.
• Operational control: Use Canvas, the AI dialog, and resource cards to inspect or tune the deployment.

Deployment Guide

1. Open the Hermes Agent template and click Deploy Now.
2. Enter a strong dashboard_username, dashboard_password, and api_server_key. Add openai_api_key when chat actions need an OpenAI-compatible provider.
3. Wait for deployment to complete, typically 2-3 minutes for the image and persistent volume. After deployment, Sealos opens the Canvas.
4. Open the generated HTTPS URL. The dashboard login page accepts the username and password entered in step 2.

Login and API Access

Dashboard login

Open the application URL and sign in at /login with dashboard_username and dashboard_password. The dashboard uses the upstream basic-auth provider for non-loopback deployments.

API clients

Send the configured bearer key when calling the API server through the /v1 path:

curl -H "Authorization: Bearer <api_server_key>" \
  "https://<your-hermes-host>/v1/models"

Keep the API key private because the Hermes API can invoke agent tools, including terminal actions.

Configuration and Scaling

Use the Hermes dashboard for model providers, skills, channels, schedules, and memory settings. Use Canvas resource cards for CPU, memory, storage, logs, and environment changes. The official bundle is intentionally single-replica because /opt/data contains the agent's local state.

Troubleshooting

Dashboard does not open

Check the dashboard container logs and confirm that dashboard_username, dashboard_password, and the generated dashboard secret are present. The dashboard must bind to 0.0.0.0 for the Sealos ingress.

API requests return 401

Use the exact api_server_key configured during deployment and send it as a Bearer token. The API server intentionally fails closed without a key.

Chat actions need a model credential

Set openai_api_key or configure another supported provider in the dashboard. The dashboard and gateway can be healthy while model calls remain unconfigured.

Getting Help

• Hermes Agent documentation
• Hermes Agent issues
• Sealos documentation
• Sealos Discord

Additional Resources

• Hermes Agent Docker Compose source
• Hermes Agent release v2026.8.3
• Sealos App Store

License

This Sealos template is provided for Sealos users under the templates repository license. Hermes Agent itself is licensed under the MIT License.