# cloudHPC [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/CFD-FEA-SERVICE/cloudhpc-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/cloudhpc

## Description
Run FDS, OpenFOAM, CalculiX and other engineering simulations on cloudHPC from your AI assistant.

## Tools
Capabilities this server exposes over MCP:

- **suggest_resources** — vCPU and RAM type recommendation, with reasoning
- **solver_guide** — How to prepare a case for a solver: input files, what cloudHPC does automatically, restart, optional settings, parallelism, logs
- **launch_simulation** — Launch a run (**asks for confirmation**)
- **sync_simulation** — Upload partial results of a running job
- **stop_simulation** — Soft or hard stop (**asks for confirmation**)
- **resume_simulation** — Resume a stopped run with the same solver, vCPU, RAM type and folder (FDS and OpenFOAM; **asks for confirmation**)
- **start_desktop** — Start a Linux or Windows remote desktop for post-processing; billed until you stop it (**asks for confirmation**)
- **open_remote_desktop** — Browser remote-desktop link of a running job
- **delete_storage** — Delete a file or folder (**asks for confirmation**)
- **api_usage** — API rate limits and calls used

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "cloudhpc": {
    "url": "https://mcp.cloudhpc.cloud/mcp"
  }
}
```

## Documentation

## What cloudHPC does

The cloudHPC MCP server lets an MCP-compatible assistant work with engineering simulations hosted by cloudHPC. It can inspect a local case, identify supported solver data, estimate model size, recommend vCPU and RAM options, upload files, start a run, monitor its status, diagnose known errors, and retrieve result archives.

Supported solvers are those available on cloudHPC. The README specifically lists FDS, OpenFOAM, snappyHexMesh, DAFoam, code_aster, CalculiX, OpenRadioss, SU2, openTELEMAC, LIGGGHTS, OpenLB, EnergyPlus, and CONTAM. Use it when simulation setup, resource selection, run management, or result retrieval would benefit from conversational access.

## How it works

The cloudHPC MCP server can run locally through the `cloudhpc-mcp` command or remotely at `https://mcp.cloudhpc.cloud/mcp`. The local mode can access case folders on the computer and provides tools for inspecting, uploading, and downloading local files. The hosted mode cannot read the computer directly; cases must be transferred with temporary upload and download links or through the cloudHPC web application.

Its tools cover case inspection, solver instructions, machine and solver listings, resource recommendations, simulation launch and status checks, partial-result synchronization, stopping and resuming runs, storage browsing, result downloads, remote desktop access, and API usage. A Linux or Windows remote desktop can be started for post-processing, but it remains billable until stopped.

Operations with financial or destructive consequences are guarded by confirmation. Launching a simulation, hard-stopping a run, starting a desktop, resuming a run, and deleting storage require approval. Local-only file tools include case inspection, folder upload, and result download; hosted use substitutes temporary file links.

## Setup and configuration

The cloudHPC MCP server requires Python 3.10 or newer, a cloudHPC account, and an API key. Install it from the Git repository with either `pipx install git+https://github.com/CFD-FEA-SERVICE/cloudhpc-mcp` or pip. The installation provides the `cloudhpc-mcp` executable.

Set `CLOUDHPC_APIKEY` in the MCP client's environment. If that variable is absent, the server reads `~/.cfscloudhpc/apikey`. Keep the API key private because it grants full access to the account.

For local clients, configure the command and environment variable as a standard stdio MCP server. Claude Desktop, Claude Code, ChatGPT desktop where MCP servers are available, Codex CLI, Gemini CLI, and VS Code Copilot agent mode are documented connection targets. The hosted endpoint requires no local installation and supports OAuth sign-in or an API key sent in a header.

## Limitations and notes

The server depends on cloudHPC availability and account permissions. The API key provides full account access, so it should not be exposed in prompts or shared configuration. Remote desktops continue to incur charges until stopped, and simulation launches, hard stops, and storage deletion can change account data or billing state.

The hosted endpoint cannot access local paths. Use the local installation when an assistant must inspect a case folder directly or write downloaded results beside it. Resume support is documented for FDS and OpenFOAM and preserves the original solver, vCPU choice, RAM type, and folder.

_Full upstream README: https://allmcps.com/mcp/cloudhpc/readme_