Quick Connect
Quick Connect is the bridge between your deployed MCP server and the AI clients that consume it. Every server on Vinkius Cloud exposes a unique URL that works with any client supporting the Model Context Protocol — Claude, Cursor, VS Code, Windsurf, Cline, and any other MCP-compatible tool.
The connection is instant: generate a token, copy the URL, paste it into your client. No SDKs, no libraries, no handshake configuration.
How it works
The Quick Connect dialog — accessible from the server banner, the Dashboard, and the token generation success screen — gives you two things:
- MCP URL — The full URL with your server ID and token embedded. Copy and paste it directly.
- Config JSON — A ready-to-paste JSON block formatted for the standard
mcpServersconfiguration format that most clients expect.
The dialog also includes a Cursor deep link button that opens Cursor directly and pre-fills the MCP configuration — zero manual pasting.
Generating a connection token
Before you can connect, you need a token. Tokens authenticate AI clients and let you control access at a granular level.
- Open your server's Connection Tokens tab.
- Click Generate Token and enter a descriptive name (e.g., "Claude Desktop — Production").
- The platform returns the plaintext token and the full MCP URL — this is the only time the plaintext is visible.
- Copy the URL or use the Quick Connect dialog to configure your client.
Token visibility
The plaintext token is shown exactly once. If you lose it, revoke the token and generate a new one. This is a security measure — tokens are stored as hashes and cannot be retrieved after creation.
Connecting AI clients
Each AI client has a slightly different configuration format, but the concept is universal: paste your MCP URL or Config JSON into the client's settings, and the connection is live. Below is the exact setup for every supported client.
{
"mcpServers": {
"your-server-name": {
"url": "https://edge.vinkius.com/{token}/mcp"
}
}
}Save and restart Claude Desktop. Your server's tools appear in Claude's tool picker — the model can now discover and call them during conversations.
Multiple servers
Add as many entries to mcpServers as you need. Each entry is a separate MCP server with its own URL and token. Claude lists all available tools from all connected servers in one unified picker.
{
"mcp": {
"servers": {
"your-server-name": {
"url": "https://edge.vinkius.com/{token}/mcp"
}
}
}
}Reload VS Code. The tools become available to GitHub Copilot, which discovers and invokes them during chat and inline completions.
Connection architecture
When an AI client connects to your MCP server, the platform establishes a stateful session over Streamable HTTP. Here's what happens technically:
Connection details
| Property | Value |
|---|---|
| Protocol | Streamable HTTP over HTTPS |
| Authentication | Token-based, embedded in the URL path |
| Reconnection | Automatic — clients reconnect if the connection drops |
| Concurrency | Multiple AI clients can connect to the same server simultaneously using the same or different tokens |
| Active connections | Visible in real time on the server's Dashboard tab and the fleet overview |
| Latency | Sub-100ms tool discovery; API call latency depends on your upstream endpoint |
Revoking access
To disconnect a specific AI client, go to the server's Connection Tokens tab, find the token, and click Revoke.
Revocation is immediate: the active connection is terminated and any subsequent MCP requests using that token fail with an authentication error. Other tokens remain unaffected — you can revoke access to one client without disrupting others.
When to revoke
Common scenarios: a team member leaves, a token is accidentally exposed, or you want to rotate credentials as part of a security policy. After revoking, generate a new token and update the affected client's configuration.
Next steps
Frequently Asked Questions
How do I connect my MCP server to Claude Desktop?
Open Claude Desktop, go to Settings → Developer → Edit Config. Paste the Config JSON block provided by Vinkius Cloud into your claude_desktop_config.json file. Restart Claude Desktop — your tools are immediately available.
Can I connect to Cursor without editing config files?
Yes. Vinkius Cloud generates a Cursor deep link for every server. Click it, and Cursor opens with the server pre-configured in your MCP settings. No manual JSON editing needed.
What is a connection token and how is it secured?
A connection token is an HMAC-SHA256 signed credential that authenticates your AI client with a specific MCP server. Tokens are shown once at creation and can be revoked in under 40ms. Each token is scoped to a single server and user.
Can I have multiple connection tokens for the same server?
Yes. You can create multiple tokens for a single server — useful for different team members, environments (dev, staging, production), or AI clients. Each token can be revoked independently.
Which AI clients are supported?
Any MCP-compatible client works, including Claude Desktop, Claude Code, Cursor, VS Code with GitHub Copilot, Windsurf, Cline, OpenAI Codex, and any custom client implementing the Model Context Protocol.
What MCP transport protocol does Vinkius Cloud use?
Vinkius Cloud uses the Streamable HTTP transport. The MCP URL follows the format https://mcp.vinkius.com/{TOKEN}/mcp, which works with all modern MCP clients that support HTTP-based transport.