Everything you can do in the client portal, you can do over our HTTP API — buy a server, control power, reinstall, resize, snapshot, back up, restore, manage firewall rules and SSH keys, read metrics, and pay invoices. We also ship an official MCP server on top of that API, so a customer's AI agent can operate their VPS as an extension of itself — not by copy-pasting commands, but by calling tools directly. The full, always-current API reference lives at /user-api/docs/.
An operate key lets an agent (or a script) run your servers: read everything, provision, and manage — but it cannot spend money. Two ways to get one, no website required for either:
curl -s -X POST https://blesta.sovereignhybridcompute.com/user-api/v2/register \
-H "Content-Type: application/json" \
-d '{"email":"you@example.com","password":"a-strong-random-password","first_name":"Your","last_name":"Name","country":"US","tos_accepted":true,"scope":"operate"}'
The response carries a secret key with an shc_live_ prefix. Keep it; you send it as a bearer token on every later call. Review the Terms of Service before you set tos_accepted — that field is your agreement to them.No key — operate, read, or full — can read or change your account credentials.
Our MCP server exposes the same API as Model Context Protocol tools. Point any MCP client that speaks Streamable HTTP at our endpoint and authenticate with your operate key as a bearer token:
We run the server — there is nothing to install. This is the canonical, official MCP for SHC. (Our API docs page may still show an older in-Blesta endpoint at /user-api/mcp; that path is legacy — use mcp.sovereignhybridcompute.com.)
The safety model is enforced by the API itself, not by good intentions:
If you want your agent to be a competent operator out of the box — with the right discipline and ready-made workflows — point it at our skill-pack at /agent-skills/. It carries the operator rules (never spend without your yes, address servers by service ID, treat a billed service as not yet booted until it reports ready) and end-to-end workflows. Start with get-access: it gets the key and sets everything else up.
The API tells you what you own and how it's doing — not which physical machine in which rack your server landed on. That placement is ours to manage and not yours to worry about. You address everything by service ID; we handle the rest. This matches how we run everything else: we have no interest in your data.