Enables management of RunPod cloud infrastructure including GPU pods (with NVIDIA GPU support), serverless endpoints, templates, network volumes, and container registry authentications through the RunPod REST API.
Enables management of RunPod cloud infrastructure including GPU pods (with PyTorch container support), serverless endpoints, templates, network volumes, and container registry authentications through the RunPod REST API.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@RunPod MCP Server list all my current pods"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
This Model Context Protocol (MCP) server enables you to interact with the RunPod REST API through Claude or other MCP-compatible clients.
The server provides tools for managing:
Pods: Create, list, get details, update, start, stop, and delete pods
Endpoints: Create, list, get details, update, and delete serverless endpoints
Templates: Create, list, get details, update, and delete templates
Network Volumes: Create, list, get details, update, and delete network volumes
Container Registry Authentications: Create, list, get details, and delete authentications
Related MCP server: Hyperbolic GPU MCP Server
Node.js 18 or higher
A RunPod account and API key
Claude for Desktop or another MCP-compatible client
To install runpod-mcp-ts for Claude Desktop automatically via Smithery:
npx -y @smithery/cli install @runpod/runpod-mcp-ts --client claudeClone the repository
Install dependencies:
npm installBuild the server:
npm run buildSet your RunPod API key as an environment variable:
# Linux/macOS
export RUNPOD_API_KEY=your_api_key_here
# Windows (Command Prompt)
set RUNPOD_API_KEY=your_api_key_here
# Windows (PowerShell)
$env:RUNPOD_API_KEY="your_api_key_here"You can get your API key from the RunPod console.
Start the server:
npm startOpen Claude for Desktop
Edit the config file: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows)
Add the server configuration:
{
"mcpServers": {
"runpod": {
"command": "node",
"args": ["/path/to/runpod-mcp-server/build/index.js"],
"env": {
"RUNPOD_API_KEY": "your_api_key_here"
}
}
}
}Make sure to replace the "args": ["/path/to/runpod-mcp-server/build/index.js"] with the path to the build folder in the repository.
Restart Claude for Desktop
Here are some examples of how to use the server with Claude:
Can you list all my RunPod pods?Create a new RunPod pod with the following specifications:
- Name: test-pod
- Image: runpod/pytorch:2.1.0-py3.10-cuda11.8.0-devel-ubuntu22.04
- GPU Type: NVIDIA GeForce RTX 4090
- GPU Count: 1Create a RunPod serverless endpoint with the following configuration:
- Name: my-endpoint
- Template ID: 30zmvf89kd
- Minimum workers: 0
- Maximum workers: 3This server requires your RunPod API key, which grants full access to your RunPod account. For security:
Never share your API key
Be cautious about what operations you perform
Consider setting up a separate API key with limited permissions
Don't use this in a production environment without proper security measures
MIT
Inputs: name (required), username (required), password (required).
Inputs: templateId (required), name, computeType, gpuTypeIds, +4 more.
Inputs: name (required), size (required), dataCenterId (required).
Inputs: imageName (required), name, cloudType, gpuTypeIds, +7 more.
Inputs: name (required), imageName (required), isServerless, ports, +7 more.
Inputs: containerRegistryAuthId (required).
Inputs: endpointId (required).
Inputs: networkVolumeId (required).
Inputs: podId (required).
Inputs: templateId (required).
Inputs: containerRegistryAuthId (required).
Inputs: endpointId (required), includeTemplate, includeWorkers.
Inputs: networkVolumeId (required).
Inputs: podId (required), includeMachine, includeNetworkVolume.
Inputs: templateId (required).
Inputs: no named parameters.
Inputs: includeTemplate, includeWorkers.
Inputs: no named parameters.
Inputs: computeType, gpuTypeId, dataCenterId, name, +2 more.
Inputs: no named parameters.
Inputs: podId (required).
Inputs: podId (required).
Inputs: endpointId (required), name, workersMin, workersMax, +3 more.
Inputs: networkVolumeId (required), name, size.
Inputs: podId (required), name, imageName, containerDiskInGb, +4 more.
Inputs: templateId (required), name, imageName, ports, +2 more.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 26 tools
Each tool targets a distinct resource and action (e.g., create-pod vs delete-pod), with no overlap. Even without descriptions, the naming makes each tool's purpose clear.
All tools follow a consistent verb_noun pattern using underscores (e.g., create-endpoint, list-pods). Verbs are limited to create, delete, get, list, update, start, stop, and nouns are the resource types.
With 26 tools covering multiple resource types (pods, endpoints, templates, network volumes, container registry auth), the count is appropriate but slightly above the typical well-scoped range.
For each resource, the standard CRUD operations (create, get, list, delete, update) are present, plus start/stop for pods. No obvious gaps in the lifecycle coverage inferred from the tool names.
On-demand GPU nodes for agents: create nodes, run commands, and submit jobs, billed by the minute.
Manage Rackspace Spot Kubernetes Cloudspaces, node pools, and VMs from your AI assistant.
Provides capabilities that let LLM agents perform a range of infrastructure management tasks.
Create, watch, pay for and connect hosted AI agent pods on AgentsPodium. Needs an API key.
