CloudGPU MCP
cloudgpu-app/cloudgpu-mcp
Agent GPU compute scheduling: compare billing, approve budgets, run tasks and persist results.
- GitHub stars
- 1
- tools
- 20
- auth
- Secret needed
- transport
- STDIO
- env var
- 1
- last updated
- Nov 13, 2026
Tools
20 toolsPublic tool metadata for what this MCP can expose to an agent.
get_accountBalance, and how many instances the account may run at once
Balance, and how many instances the account may run at once
No parameter schema in public metadata yet.
list_gpusGPU models with VRAM, live price per GPU-hour and availability
GPU models with VRAM, live price per GPU-hour and availability
No parameter schema in public metadata yet.
list_templatesReady-made images: JupyterLab, Ollama, ComfyUI, LLaMA-Factory, …
Ready-made images: JupyterLab, Ollama, ComfyUI, LLaMA-Factory, …
No parameter schema in public metadata yet.
compare_compute_costsCompare live total billed GPU cost for up to eight compatible candidates; no reservation or creation approval
Compare live total billed GPU cost for up to eight compatible candidates; no reservation or creation approval
No parameter schema in public metadata yet.
estimate_costLive quote for a GPU and a number of hours, plus your balance afterwards; returns a quote_id
Live quote for a GPU and a number of hours, plus your balance afterwards; returns a quote_id
No parameter schema in public metadata yet.
create_instanceRents from an approved quote, with a budget, a time limit and an idle limit
Rents from an approved quote, with a budget, a time limit and an idle limit
No parameter schema in public metadata yet.
get_statusStatus, minutes run, cost so far, auto-shutdown limits
Status, minutes run, cost so far, auto-shutdown limits
No parameter schema in public metadata yet.
list_instancesStatus, minutes run, cost so far, auto-shutdown limits
Status, minutes run, cost so far, auto-shutdown limits
No parameter schema in public metadata yet.
run_commandRuns a command on the instance; background=true for long jobs
Runs a command on the instance; background=true for long jobs
No parameter schema in public metadata yet.
tail_logsLatest log lines, whether the job finished, GPU utilisation
Latest log lines, whether the job finished, GPU utilisation
No parameter schema in public metadata yet.
uploadFiles and folders between your computer and the instance (SFTP)
Files and folders between your computer and the instance (SFTP)
No parameter schema in public metadata yet.
downloadFiles and folders between your computer and the instance (SFTP)
Files and folders between your computer and the instance (SFTP)
No parameter schema in public metadata yet.
stop_instanceDestroys the instance and stops billing
Destroys the instance and stops billing
No parameter schema in public metadata yet.
list_modelsModels on the OpenAI-compatible API, for inference without renting a GPU
Models on the OpenAI-compatible API, for inference without renting a GPU
No parameter schema in public metadata yet.
run_jobStarts a GPU, runs one command and destroys the GPU, all on CloudGPU's servers; safe to retry with the same idempotency_key
Starts a GPU, runs one command and destroys the GPU, all on CloudGPU's servers; safe to retry with the same idempotency_key
No parameter schema in public metadata yet.
get_jobJob status, exit code, cost, the last ~4 KB of output and the job's history
Job status, exit code, cost, the last ~4 KB of output and the job's history
No parameter schema in public metadata yet.
list_jobsJob status, exit code, cost, the last ~4 KB of output and the job's history
Job status, exit code, cost, the last ~4 KB of output and the job's history
No parameter schema in public metadata yet.
cancel_jobStops a job and destroys its GPU
Stops a job and destroys its GPU
No parameter schema in public metadata yet.
connect_accountGets a key without copying one: the user approves this assistant in the browser with a spending limit and a number of GPUs at once
Gets a key without copying one: the user approves this assistant in the browser with a spending limit and a number of GPUs at once
No parameter schema in public metadata yet.
finish_connectGets a key without copying one: the user approves this assistant in the browser with a spending limit and a number of GPUs at once
Gets a key without copying one: the user approves this assistant in the browser with a spending limit and a number of GPUs at once
No parameter schema in public metadata yet.
Sponsors
Promote your product hereReach 125k+ visitors monthlyPromote your product hereReach 125k+ visitors monthly
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any language
Agent CaffeineSponsoredClose your lid, come back to finished workSnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthlyWaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthlyknobsSponsoredSee your site on any phone, any day, in any languageConfiguration
CLOUDGPU_API_KEYsecretOptional CloudGPU API key. Without a key, use public discovery and connect_account for an owner-approved scoped key.
Readme
CloudGPU MCP
CloudGPU MCP server: let an AI agent compare GPU compute costs, approve source-specific billing, run a job, bring the results back and shut the GPU down. The same API key and balance also cover CloudGPU's OpenAI-compatible model API, so an agent can call a hosted model without renting anything.
"Fine-tune a 7B base model with LoRA on data/train.jsonl and put the adapter in ./outputs."
The agent checks live prices and quotes you, rents the GPU once you say yes, uploads your code and data, starts training in the background, watches the logs, downloads the adapter and destroys the instance. You get the files and the final cost.
It is a local (stdio) MCP server. It works with any MCP client that can start one: Codex, Claude Code, Cursor, Cline, opencode, goose, VS Code, Windsurf and others. Remote Streamable HTTP with OAuth is also available at https://cloudgpu.app/mcp; see remote MCP setup.
Install
For the browser-approval flow, add this local MCP server without an API key first:
claude mcp add cloudgpu -- npx -y cloudgpu-mcp
Details
- Repository
- github.com/cloudgpu-app/cloudgpu-mcp
- Website
- cloudgpu.app/docs/mcp
- Categories
- AI & LLM Tools, Finance & Commerce
- Package
cloudgpu-mcp- Transport
- STDIO
- Auth
- Required
- Registry
- MCP registry, active
