cloud-infra-toolchain
Use when working with the user's cloud-infrastructure MCP stack fronted by the cloud-hub multiplexer - RunPod, Azure, Google Cloud, Kaggle, rclone, Modal, Lightning AI, and OVHcloud behind one lazy MCP slot. Covers which platform does what, adopt-vs-build-vs-vendor status, per-platform auth + the account-switch/cycle multi-account layer (incl. OVH region rotation ca/eu/us), the workflows (train on Kaggle, rent a GPU on RunPod/Modal/Lightning/OVH, deploy to Cloud Run/AKS, move data with rclone), the rclone rcd daemon prereq, the Windows `cmd /c npx` note, GPU cost-awareness + read-before-destroy operating rules, and the per-backend tool inventory. Multi-accounting can breach a platform's ToS - flag it.
- License 许可
- MIT
- 最近更新
- 2026-11-02
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- cloud-infra-toolchain
- description
- Use when working with the user's cloud-infrastructure MCP stack fronted by the cloud-hub multiplexer - RunPod, Azure, Google Cloud, Kaggle, rclone, Modal, Lightning AI, and OVHcloud behind one lazy MCP slot. Covers which platform does what, adopt-vs-build-vs-vendor status, per-platform auth + the account-switch/cycle multi-account layer (incl. OVH region rotation ca/eu/us), the workflows (train on Kaggle, rent a GPU on RunPod/Modal/Lightning/OVH, deploy to Cloud Run/AKS, move data with rclone), the rclone rcd daemon prereq, the Windows `cmd /c npx` note, GPU cost-awareness + read-before-destroy operating rules, and the per-backend tool inventory. Multi-accounting can breach a platform's ToS - flag it.
03 SKILL.md 正文
Cloud Infrastructure MCP Toolchain
Overview
The user's cloud-infra MCP servers are fronted by cloud-hub — one lazy,
name-preserving multiplexer (tools/mcp/cloud-hub/) that answers tools/list
from committed caches without spawning and lazily spawns only the owning
backend on a tools/call. Ten platforms, 324 backend tools + 54 account
tools = 378, behind a single MCP registration. Five servers are adopted
(existing MCP packages), four are built thin FastMCP wrappers, and one
(OVHcloud) is a vendored TS server built from source.
| Platform | pkg / runtime | Auth | Adopt/Build | Tools | Hard limits |
|---|---|---|---|---|---|
| RunPod | npx @runpod/mcp-server@latest (Node) |
RUNPOD_API_KEY |
adopt | 50 | GPU billed per-second; pods/endpoints cost money while running |
| Azure | npx @azure/mcp@latest server start (.NET) |
az login |
adopt | 14* | *hand-authored cache (the .NET server didn't enumerate under piped stdio); --refresh azure upgrades it |
| Google Cloud | npx @google-cloud/gcloud-mcp (Node) |
gcloud ADC + GOOGLE_CLOUD_PROJECT |
adopt | 1 | single run_gcloud_command passthrough — full gcloud surface, full gcloud risk |
| Kaggle | uvx kaggle-mcp-server (Python) |
KAGGLE_USERNAME+KAGGLE_KEY / ~/.kaggle/kaggle.json |
adopt | 51 | free-tier GPU/TPU quota (~30h/wk); kernels are public unless set private |
| rclone | npx rclone-mcp over rclone rcd |
RC daemon :5572 | adopt | 56 | needs the rcd daemon (auto-started); sync/move/delete are destructive |
Multi-account + cycling (cswap-style)
Auth-setup runbook
Workflows
Train on Kaggle (free GPU/TPU)
Rent a GPU (options — pick by workload)
Deploy an app
Move data (rclone)
rclone rcd daemon prereq
Windows launch note
Operating rules
Per-backend tool inventory
Databricks as a Codex agent host (not just Spark)
04 同仓库更多 Skills
clarify-requirements
Resolve genuine pre-execution ambiguity by scanning repo context, generating 1-6 grounded options, and asking 1-3 focused questions via AskUserQuestion. NOT for mid-task use (injection risk); NOT when the goal is already unambiguous; use to disambiguate vague prompts before any file edits.
openmm-md
Run FREE, open-source, Windows-native GPU molecular dynamics via the openmm-md MCP (OpenMM + CUDA on the local NVIDIA GPU). Use for protein / protein-ligand MD when you want a free stack or are on Windows without a Schrodinger license. Covers the full workflow (prepare, solvate, minimize, equilibrate, production, analyze, mmgbsa) AND an honest decision guide vs Schrodinger Desmond, including how to run Desmond legitimately on Linux or a cloud GPU with a real license. This is a deliberately LESSER free alternative and does NOT replace the schrodinger MCP tools. Do NOT use it to pick a free tool when a validated FEP+, WaterMap, IFD-MD, or Prime MM-GBSA result is actually required.
visual-review
Take screenshots of web pages and UI using Playwright for visual review and iteration
status
Diagnostic status report — shows metaswarm installation state, project setup, and potential issues
start
Use when starting work on any task, when the user mentions metaswarm, or when the user wants to begin tracked development work
setup
Interactive project setup — detects your project, configures metaswarm, writes project-local files