# CloudGPU rules for AI assistants

Use these rules whenever you rent or use a GPU on CloudGPU (cloudgpu.app) through the `cloudgpu` MCP tools.
Copy this file into your project as `AGENTS.md`, `.cursor/rules/cloudgpu.md`, `.clinerules/cloudgpu.md` or `CLAUDE.md` if your assistant reads one of those.

## Before renting

1. Call `get_account` and `list_gpus`. Pick a GPU with enough VRAM for the job, and don't rent more GPU than the job needs. As a rough guide, LoRA on a 7–8B model fits in 24 GB, and a full fine-tune of the same model needs 80 GB or more.
2. Call `estimate_cost` with a realistic number of hours.
3. Show the user a short quote before you rent anything:
   - the GPU and how many,
   - the price per hour,
   - the estimated hours and cost,
   - the budget and time limit you will set,
   - their balance after the job.
4. Rent only after the user says yes. Pass `user_confirmed=true` only in that case.

## Renting

- Always set `max_spend_usd` and `max_hours`. CloudGPU's servers destroy the instance when either is reached, even if you have disconnected.
- Keep `idle_minutes` on (default 60). Raise it if setup will take long, because installing packages and downloading models leave the GPU idle.
- Poll `get_status` every 30–60 seconds until the instance is `running`. This usually takes about 2 minutes.
- `failed` instances are not charged. Tell the user, then offer another GPU.

## Working on the instance

- Put code and data on the instance with `upload`. Do not upload credentials.
- Use `run_command` for short steps. For training, long installs or big downloads, use `background=true` and follow progress with `tail_logs`.
- Instances come with Python and CUDA, but not always PyTorch. Check first: `nvidia-smi -L` and `python -c "import torch; print(torch.__version__, torch.cuda.is_available())"`.
- If PyTorch is missing, plain `pip install torch` installs a current CUDA build. It is a few GB, so run it with `background=true`. Index URLs for older CUDA versions (for example `.../whl/cu124`) may have no wheel for the machine's Python, which can be as new as 3.14. Only use them if the project needs that CUDA version, and then create an environment with an older Python first (`conda create -y -n work python=3.12`, then `conda activate work && ...` in each command).
- If `nvidia-smi` shows no GPU a few minutes after start, stop and tell the user. CloudGPU detects such machines, shuts them down and refunds them automatically.
- **Do not** change the root password, edit the SSH configuration, change the firewall, or stop the platform's monitoring agent (`cgpu-*` processes). The platform must keep its access to the instance; if that access is lost, the platform shuts the instance down. The MCP server refuses such commands.
- Never use the GPU for cryptocurrency mining or anything the [acceptable use policy](https://cloudgpu.app/aup) forbids. Instances that do so are shut down.

## Finishing (always)

1. Download every result: model weights, adapters, checkpoints, logs, metrics. Destroying an instance erases its disk, and CloudGPU has no persistent storage.
2. Call `stop_instance`.
3. Tell the user what ran, where the results are on their computer, and the final cost from `stop_instance`.

If anything goes wrong mid-way, do not leave the instance running. Download whatever exists, stop the instance, then explain what happened.

## Only need a model, not a GPU?

The same API key works for CloudGPU's OpenAI-compatible model API at `https://cloudgpu.app/v1`. Call `list_models` to see the models. For inference on a hosted model this is usually quicker and cheaper than renting a GPU.