Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@cursor-yougpu-mcp what's my current balance?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP-сервер для YouGPU REST API: каталог GPU, создание/стоп/старт/удаление инстансов, баланс.
Архитектура пайплайна транскрибации и диаризации: docs/TRANSCRIPTION_ARCHITECTURE.md.
Tool | Описание |
| Баланс кошелька |
| Каталог моделей GPU |
| Офферы/цены по |
| SSH-ключи аккаунта |
| Список инстансов |
| Детали инстанса |
| Создать инстанс |
| Ждать ACTIVE + IP |
| Остановить |
| Запустить |
| Terminate (стоп биллинга GPU) |
Related MCP server: quads-mcp
mkdir -p ~/.config/yougpu
cp config.example.env ~/.config/yougpu/config.env
# set YOUGPU_API_KEY=...
npm install
npm run build
chmod +x bin/cursor-yougpu-mcp
ln -sf "$PWD/bin/cursor-yougpu-mcp" ~/.local/bin/cursor-yougpu-mcpВ ~/.cursor/mcp.json:
"yougpu": {
"command": "/Users/YOUR_USER/.local/bin/cursor-yougpu-mcp",
"args": []
}API key создаётся в console.yougpu.ru (раздел API Keys). Документация: https://docs.yougpu.ru/api/overview/
Get YouGPU wallet balance (available, reserved, hourly_rate, total).
Inputs: no named parameters.
Create a GPU instance. Requires offer_id and image_id from offers. Uses YOUGPU_SSH_KEY_IDS from config if ssh_key_ids omitted.
Inputs: image_id (required), offer_id (required), instance_name (required), ssh_key_ids, +2 more.
Terminate/delete instance permanently and stop GPU billing (POST /instances/{id}/terminate).
Inputs: instance_id (required).
Get one instance by id (summary + raw payload).
Inputs: instance_id (required).
List GPU models available in YouGPU catalog with price ranges and stock.
Inputs: no named parameters.
List YouGPU instances (active and terminated history).
Inputs: page, limit.
List in-stock offers for a GPU model id from yougpu_list_gpus. Returns offer_id, price/hour, region, images.
Inputs: gpu_id (required).
List SSH keys registered in the YouGPU account.
Inputs: no named parameters.
Start a previously stopped instance.
Inputs: instance_id (required).
Stop a running instance (keeps disk; may still reserve storage cost depending on provider).
Inputs: instance_id (required).
Poll instance until ACTIVE/INSTALLING with IP (or ERROR). Default timeout 10 minutes.
Inputs: instance_id (required), timeout_ms, poll_interval_ms.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 11 tools
Each tool targets a distinct resource or action: balance, GPU catalog, offers, SSH keys, and instance lifecycle operations. There is no ambiguity between list_instances and get_instance, or between list_gpus and list_offers, as they operate on different levels of detail.
Names follow a consistent yougpu_verb_noun pattern for most tools (list, get, create, stop, start, delete). The only deviation is 'yougpu_balance', which is a noun rather than verb-first, but it is still predictable and readable.
With 11 tools, the server is well-scoped for managing GPU instances. It includes essential operations for wallet, catalog, offers, SSH keys, and full instance lifecycle without unnecessary bloat.
The toolset covers the complete workflow: checking balance, browsing GPUs/offers, managing SSH keys, creating instances, monitoring readiness, and performing lifecycle operations (start, stop, delete). There are no critical gaps for the stated purpose.
An MCP server that provides an API to LLMs to manage their JumpCloud resources.
MCP server for Hostkey .ru (InvAPI): servers, power, order, DNS, S3, billing
MCP commerce surface for compute credits, API keys, GPU instances, and cloud storage.
MCP server for Hostkey .com (InvAPI): servers, power, order, DNS, S3, billing
