MCP for AI Agents
EQVPS exposes a Model Context Protocol (MCP) server so AI agents can discover plans, order a VPS, pay from a prepaid balance, and manage servers — autonomously. The endpoint speaks Streamable HTTP (the 2026 remote-MCP standard).
Quickstart — connect in one command
Step 1 — add the server (Claude Code, one command):
claude mcp add --transport http eqvps https://mcp.eqvps.com/mcpStep 2 — tell your agent, in plain language (copy-paste):
Register an EQVPS account, show me plans, create a crypto top-up invoice for $20, then once I confirm payment, order 2 Ubuntu 24.04 VPS and give me their SSH access.⚠️ The agent does everything via tools EXCEPT funding: a human sends crypto to the top-up invoice once. After the balance is funded, the agent orders VPS fully autonomously (pays from balance).
♻️ Repeat purchases with a funded balance are fully hands-off — just ask the agent.
⏳ A freshly ordered or just-reset VM needs ~1-2 minutes to boot. Ubuntu 24.04 starts sshd on the first connection (socket-activation), so an immediate SSH attempt may briefly show "connection refused" even though provisioning succeeded — just wait a minute and retry. Don't reset or reinstall because SSH isn't instant.
🔧 Generic / OpenClaw / other MCP clients: use the JSON config block below.
Connect your agent
Add this to your MCP client config (Claude Desktop, or any remote-HTTP MCP client):
{
"mcpServers": {
"eqvps": {
"type": "streamable-http",
"url": "https://mcp.eqvps.com/mcp",
"headers": { "Authorization": "Bearer <YOUR_TOKEN>" }
}
}
}Get a token by calling register_account or login through the MCP server itself. The list_plans tool works without a token (discovery). You can also generate a token in your account Settings → API tokens for agents.
Comments
No comments yet. Be the first.