How to deploy and use DeepSeek Harness on a Hostinger VPS

Hostinger VPS DeepSeek Harness deployment allows you to run an open-source, plugin-based AI coding agent directly from a browser interface. You can install the DeepSeek Harness template in hPanel, access the web interface using server credentials, configure custom AI models, and connect Hostinger AI Router for agent workflows.

Before you start

• A Hostinger VPS running the DeepSeek Harness template on Ubuntu.
• Your VPS root password to sign in to the web interface.
• A Hostinger AI Router API key to route models through the gateway.

DeepSeek Harness overview

DeepSeek Harness (dsh) is an open-source coding agent that reads files, executes commands, and manages development tasks through a web interface. The agent operates on a plugin architecture that connects browser interfaces, command terminals, and external applications over the Agent Client Protocol (ACP).

Step 1 – Deploy DeepSeek Harness on your VPS

1. Log in to hPanel, go to VPS, and select your server.
2. Choose the DeepSeek Harness template under application templates and click Install.
3. Wait for the installation to complete. hPanel automatically configures the web interface, reverse proxy, and TLS certificate.

Step 2 – Open the Web UI and sign in

1. Open your VPS overview in hPanel to locate the application login credentials.
2. Enter the access URL (https://<your-vps-hostname>) in your web browser.
3. Sign in using admin as the username and your VPS password.

NOTE

• The DeepSeek Harness agent runs with full server permissions on your Hostinger VPS, so protect your access URL and credentials.

Step 3 – Change the model

1. Click Settings in the bottom-right corner of the web interface.
2. Open the Models tab.
3. Select your preferred AI model. Enter a DeepSeek API key to use native DeepSeek models, or configure a custom provider to connect third-party services like Hostinger AI Router.

Step 4 – Connect Hostinger AI Router

Adding Hostinger AI Router as a custom provider makes your gateway models available inside DeepSeek Harness.

1. Click Settings in the bottom-right corner of the web interface, then open the Models tab.
2. Click Add Custom Provider and enter the following values:
   • Provider ID: nexos
   • Display name: Hostinger AI Router
   • Base URL: https://api.nexos.ai/v1
   • API protocol: openai-responses
   • API key: your Hostinger AI Router key from the AI Router overview
3. Click Add Model and enter a model ID from the Hostinger AI Router models list, such as GPT 5.5.
4. Click Create Provider.

The provider models appear in the model selector for active sessions.

NOTE

• Include /v1 in the Base URL field.
• Set the API protocol field to openai-responses instead of openai-completions.
• Send unlisted model IDs directly even if they do not appear in the model selector.

Troubleshooting

• Reset your VPS password in hPanel and verify that the username is set to admin if login attempts fail.
• Verify that your API key starts with nexos- and contains no extra spaces if authentication errors occur.
• Enter the exact model ID from the Hostinger AI Router models list if a model not found error appears, as model names are case-sensitive.
• Ensure the Base URL is set to https://api.nexos.ai/v1 and the API protocol is set to openai-responses if request errors occur.

You have successfully deployed DeepSeek Harness on your Hostinger VPS and connected it to Hostinger AI Router. You can now execute coding agent tasks using custom AI models in your browser.