EN | δΈ­ζ–‡

Get Started

Everything you need to publish your first service or connect to one shared with you β€” no public IP, no port forwarding, no VPN configuration.

⬇ Download Free 🎯 Try the Demo

For Administrators β€” Publishing Services

Follow these steps if you own or manage a computer whose services you want to share securely with remote users.

1 Create an Administrator Account and Invite Users

Create an administrator account. Once signed in, click Invite Users to send invitation emails to anyone who should be allowed to access the services you publish.

Each invited user receives a personal link to create their own account. After they accept the invitation, they can access the services you publish under that administrator account β€” no manual network setup is required on their end.

2 Install MTCode Server and Publish a Service

Download and install MTCode Server, then sign in with your administrator account.

Choose the local service you want to publish β€” such as an Ollama AI server, SSH server, web service, or the built-in MTCode-LLM or MTCode-Diffusion servers β€” then click Start Server.

The service is registered with the MTCode DirectLink platform and becomes available to authorized users through MTCode Portal, without requiring a public IP address, domain name, router configuration, or VPN.

You can publish multiple services from the same computer by registering additional ports.

Publishing a remote GPU execution server? Install MTCode GPU Server instead. Configure resource limits, invite users, and click Start GPU Server. Authorized users can then connect through MTCode RemoteGPU to upload Python projects, run them on your GPU, and retrieve results.

For Users β€” Connecting to Shared Services

Follow these steps if you received an invitation to access services published by an administrator.

1 Accept the Invitation and Create an Account

Click the link in your invitation email to create a user account, or sign in if you already have one.

In your account dashboard, pending invitations are listed automatically. Click Accept to gain access to the services shared by that administrator.

Once accepted, the shared services become available in MTCode Portal with no additional configuration.

2 Install MTCode Portal and Connect

Download and install MTCode Portal, then sign in with your user account.

All services you are authorized to access appear automatically. Each service is mapped to a local address on your computer, so it can be used as if it were running locally.

Click Connect on any service, and MTCode Portal launches the appropriate client automatically β€” a browser for web services, a terminal for SSH, Remote Desktop for RDP, or the relevant application for other supported protocols.

Connecting to a remote GPU server? Install the MTCode RemoteGPU extension in VS Code, or use MTCode Studio, which includes RemoteGPU preinstalled. Sign in with your account, select an authorized GPU server, upload your Python project, and run it remotely. Results stream back to your editor in real time, or you can submit offline jobs and download outputs later.

Where to Go Next

Use this page as the quick path. For deeper details, visit the product and architecture pages below.

How DirectLink Works

Understand the signaling-assisted peer-to-peer architecture and data path.

Read the platform overview β†’

Publish TCP Services

Learn how MTCode Server publishes local ports and built-in AI services.

Explore MTCode Server β†’

Access Shared Services

See how MTCode Portal maps remote services to local addresses.

Explore MTCode Portal β†’

Run Remote GPU Jobs

Share GPU compute with MTCode GPU Server and use it from VS Code.

Explore GPU Server β†’

Ready?

Download MTCode Server, MTCode Portal, MTCode GPU Server, or MTCode RemoteGPU for free. No credit card required.

⬇ Download Free 🎯 Try the Live Demo