Dr-Claude-Code: Skill for Claude Code

RunPod Setup Skill

This is a RIGID workflow. Follow every step in order. Do not skip phases.

---

Phase 0: Check for existing config

Before anything else, check if RunPod is already configured:

cat .raca/clusters.yaml 2>/dev/null | grep -A5 "type: runpod"

If a RunPod entry already exists, tell the user what's there and ask: "Would you like to reconfigure RunPod, or just verify the existing setup?"

Also check if the API key is already set:

echo ${RUNPOD_API_KEY:+set}

---

Phase 1: Account and API Key

Step 1.1 — Account creation

If the user doesn't have a RunPod account yet, direct them to:

"Go to https://www.runpod.io and create an account. Add a payment method before continuing — pods cannot be launched without billing configured."

Wait for the user to confirm they have an account.

Step 1.2 — Get API key

"Go to https://www.runpod.io/console/user/settings and click API Keys. Create a new key (any name). Copy it — it won't be shown again."

Security rule: Never store the API key in any file. It goes into an environment variable only.

Ask the user to paste the key, then save it to their shell profile:

# Test the key first
RUNPOD_API_KEY=<paste_here>
curl -s -H "Authorization: Bearer $RUNPOD_API_KEY" \
  "https://api.runpod.io/graphql?query={myself{id,email}}" | python3 -m json.tool

If the response contains "id" and "email", the key works. Then tell the user:

"Add this to your ~/.zshrc or ~/.bashrc so it persists across sessions:

export RUNPOD_API_KEY=<your_key>

Run source ~/.zshrc to apply it now."

---

Phase 2: Understanding GPU Types and Pricing

Explain the available GPU types before the user picks a default. Present this reference:

GPU Reference (RunPod, as of early 2026)

GPU	VRAM	Est. $/hr (community)	Est. $/hr (secure)	Best for
H200 SXM	141 GB	~$3.49	~$4.49	70B+ models, large RL
H100 SXM	80 GB	~$2.49	~$3.49	7B-70B training
H100 PCIe	80 GB	~$1.99	~$2.99	7B-70B inference
A100 SXM	80 GB	~$1.64	~$2.49	7B-70B; solid multi-GPU
L40S	48 GB	~$0.99	~$1.49	Single-GPU inference/training
RTX 4090	24 GB	~$0.44	—	1B-7B models, budget option
RTX 3090	24 GB	~$0.22	—	Smallest models, cheapest