Hermes Agent + Hostinger VPS
Setup Guide

Create the Accounts You Need

Create a Hostinger account

• Go to Hostinger.
• Purchase a VPS plan.
• You do not need traditional shared web hosting.
• You specifically want a VPS because Hermes needs its own always-on server environment.

Decide how Hermes will access an AI model

Hermes is the agent software. It still needs access to an LLM such as Claude, GPT, Gemini, etc. Hermes supports multiple model providers and lets you change them later.

You can use:

• Nous Portal
• Hostinger/nexos.ai credits
• OpenRouter
• OpenAI
• Other supported providers
• A custom/self-hosted model

Create any necessary AI-provider account/API key

If using your own provider, create the account and generate the API key before configuring Hermes.

Create the Hostinger VPS

• Log into Hostinger.
• Go to VPS.
• Purchase/create the VPS that will run Hermes.
• Finish the initial VPS provisioning.
• Set a strong VPS/root password if Hostinger asks for one.
• Save the server login information somewhere secure.

Recommended Structure

For a normal business-user Hermes installation:

This keeps:

• files separate
• memory separate
• credentials separate
• integrations separate
• troubleshooting simpler
• security much cleaner

Deploy Hermes from Hostinger

Once the VPS is active:

• Open the VPS Dashboard.
• Go to Docker Manager → Catalog
• Search for Hermes Agent
• Click Select.

Hostinger will walk you through the Hermes deployment.

During Deployment

• Create an admin username.
• Create a strong admin password.
• Save both credentials securely.
• Add an AI-provider API key if you already have one.
• Add Hostinger/nexos.ai credentials if using Hostinger's AI credits.
• Add web-search/scraping credentials if desired.

Then:

• Click Deploy.

Hostinger automatically creates and starts the Hermes Docker container and the supporting services required for the deployment.

Open Hermes

After deployment finishes:

• Return to Docker Manager → Projects
• Find the Hermes project.
• Click Open.

Hostinger will open the Hermes CLI running inside the server's Docker environment.

Run the Initial Hermes Setup

The first time Hermes opens, it launches its setup process.

Choose:

Configure the AI Model

Hermes now needs to know which AI brain it should use.

• Select the AI provider.
• Enter the appropriate API key if it is not already populated.
• Select the desired model.
• Save the configuration.

Hermes also lets you change the provider/model later with:

Hermes supports multiple providers, so you are not permanently locked into the first model selected.

Configure Hermes Tools

Next, select the capabilities you want Hermes to have.

These can include things like:

• File operations
• Terminal access
• Code execution
• Web search
• Web scraping
• Browser automation
• Memory
• Skills
• Delegation/sub-agents
• Cron/scheduled tasks
• MCP integrations

You can change tool configuration later with:

Configure Web Search / Browser Capabilities

If the Hermes agent needs to research the internet:

• Enable Web Search & Scraping.
• Select the web-search/scraping provider.
• Enter the required API credentials.

Hostinger currently supports integrating Oxylabs credits directly into its Hermes deployment if that option is purchased.

If browser automation is required, make sure the appropriate browser/computer-use tools are enabled as well.

Configure Communication Channels

Decide how the user will actually communicate with Hermes.

Hermes supports channels including:

• Telegram
• Discord
• Slack
• WhatsApp
• Signal
• CLI

During setup:

• Select the desired messaging platform.
• Create/configure the necessary bot or integration.
• Enter the integration credentials.
• Connect it to Hermes.
• Send a test message.

Messaging configuration can also be changed later using:

Verify the Basic Installation BEFORE Customizing It

Before adding sophisticated workflows, confirm that the basic agent works.

Send Hermes a simple request such as:

Then test:

• Hermes responds.
• The selected LLM works.
• Hermes can use its basic tools.
• Hermes can create/read files.
• Web search works if enabled.
• Browser tools work if enabled.
• Telegram/other messaging works.
• Hermes retains a normal conversation.

Configure the Agent's Identity and Job

Now you begin turning generic Hermes into your Hermes agent.

Who the agent is

What its objectives are

• Identify high-value prospects.
• Research companies.
• Prepare meeting briefs.
• Maintain prospect information.
• Draft outreach.
• Surface opportunities requiring human attention.

What it is allowed to do

• Research the web.
• Read CRM information.
• Draft emails.
• Prepare reports.

What requires permission

• Sending external email
• Purchasing anything
• Deleting information
• Publishing publicly
• Changing CRM records

What it should never do

Create explicit boundaries for sensitive or irreversible actions.

Configure Memory

Hermes is designed to maintain persistent knowledge and improve its skills over time. It can save knowledge from previous sessions and search prior conversation history.

Set up the information you want Hermes to understand about the user, such as:

• Company
• Role
• Products/services
• ICP
• Customers
• Competitors
• Communication style
• Important terminology
• Standard processes
• Objectives
• Preferences

Install/Create Skills

Once basic Hermes works, begin adding repeatable capabilities.

Prospect Research Skill

• Receive company name.
• Research company.
• Find relevant contacts.
• Identify buying signals.
• Summarize opportunity.
• Recommend next action.

Meeting Prep Skill

• Identify attendee.
• Research attendee.
• Research company.
• Review previous conversations.
• Identify opportunities.
• Produce briefing document.

Content Skill

• Research topic.
• Find recent developments.
• Identify angle.
• Draft content.
• Match user's voice.

Connect External Systems

Only after Hermes itself is functioning correctly should you begin connecting additional systems.

Possible integrations include:

• Gmail
• Google Calendar
• Slack
• CRM
• Google Drive
• Notion
• databases
• APIs
• MCP servers
• internal company software

After every integration:

• Connect it.
• Test read access.
• Test the expected workflow.
• Verify permissions.
• Confirm Hermes cannot access anything it shouldn't.
• Only then move to the next integration.

Add Automations

Hermes includes scheduled/cron functionality for unattended tasks.

Every Morning

• Research target accounts.
• Review upcoming meetings.
• Surface high-priority opportunities.
• Send the user a morning briefing.

Every Evening

• Review completed work.
• Organize notes.
• Update internal knowledge.
• Prepare tomorrow's priorities.

Weekly

• Review pipeline.
• Research prospects.
• Identify opportunities.
• Produce performance summary.

Final QA

Before calling the deployment finished, run a standard test.

Infrastructure

• Hostinger VPS is online.
• Hermes Docker project is running.
• Admin credentials are stored securely.

AI

• AI provider connected.
• API key works.
• Correct model selected.
• Agent can successfully answer prompts.

Tools

• File operations work.
• Web search works if required.
• Browser automation works if required.
• Relevant tools are enabled.
• Unnecessary high-risk tools are disabled.

Communication

• Telegram/Slack/etc connected.
• User can message Hermes.
• Hermes can respond correctly.

Agent Configuration

• Agent knows who it is.
• Agent knows who the user is.
• Agent understands its primary objective.
• Agent understands its boundaries.
• Core company/user context is loaded.

Skills

• Core skills installed.
• Each skill has been tested.
• Failed/edge-case behavior has been tested.

Integrations

• Required accounts connected.
• Permissions verified.
• Read/write access tested.
• Sensitive actions require appropriate safeguards.

Automations

• Scheduled tasks configured.
• Test automation completed successfully.
• Delivery destination verified.