# Mrhy7176 Cloudgpu MCP server

Source: https://glimind.com/server/mcp-registry%2Fio.github.mrhy7176%2Fcloudgpu-mcp
Registry id: `mcp-registry/io.github.mrhy7176/cloudgpu-mcp`
Status: unknown

## Install

```json
{
  "mcpServers": {
    "mrhy7176-cloudgpu-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "cloudgpu-mcp"
      ]
    }
  }
}
```

Claude Code:

```sh
claude mcp add mrhy7176-cloudgpu-mcp-server -- npx -y cloudgpu-mcp
```

## Tools (16)

- `cancel_job`: Stops a job and destroys its GPU
- `compare_compute_costs`: Compare live total billed GPU cost for up to eight compatible candidates
- `create_instance`: Rents from an approved quote, with budget, time limit and idle limit
- `estimate_cost`: Live quote for a GPU and hours, plus balance afterwards; returns a quote_id
- `get_account`: Balance and how many instances the account may run at once
- `list_gpus`: GPU models with VRAM, live price per GPU-hour and availability
- `list_models`: Models on the OpenAI-compatible API, for inference without renting a GPU
- `list_templates`: Ready-made images: JupyterLab, Ollama, ComfyUI, LLaMA-Factory, etc.
- `prepare_job_rerun`: Read parameters from a successful owned Job; obtain fresh quote and budget approval
- `prepare_model_deployment`: Prepare setup script for existing ComfyUI template with model downloads
- `resolve_model_reference`: Resolve Hugging Face branch/tag to commit-addressed manifest, sizes and LFS SHA-256
- `resume_artifact_upload`: Resume multipart reservation with original local file and checksum
- `run_command`: Runs a command on the instance; background=true for long jobs
- `run_job`: Starts a GPU, runs one command and destroys the GPU; safe to retry with idempotency_key
- `stop_instance`: Destroys the instance and stops billing
- `tail_logs`: Latest log lines, whether job finished, GPU utilisation

Measured by Glimind (https://glimind.com): MCP handshake + tool list only, no tool is ever called. CC BY 4.0.
