RunPod Model Context Protocol Gateway

This implementation of the Model Context Protocol (MCP) server bridges communication between AI agents (like Claude) and the underlying RunPod REST API services.

Core Capabilities

The gateway furnishes operational tools for the following RunPod assets:

• Compute Pods: Procedures for instantiation, enumeration, retrieval of specifics, modification, activation, suspension, and termination of compute environments.
• Serverless Endpoints: Functionality to deploy, list, inspect, adjust, and retire scalable, on-demand inference endpoints.
• Deployment Templates: Management routines for defining, viewing, revising, and eliminating reusable configuration blueprints.
• Persistent Storage (Volumes): Controls for provisioning, querying, updating, and deallocating network-attached storage volumes.
• Registry Credentials: Tools for establishing, listing, modifying, and revoking authentication tokens for private container image registries.

Deployment Guide

Prerequisites

• Node.js runtime (version 18 or newer is required)
• Valid RunPod credentials (API Key)
• An MCP-aware application client (e.g., Claude for Desktop)

Automated Installation (Smithery)

Integration into Claude Desktop is simplified using the Smithery utility:

npx -y @smithery/cli install @runpod/runpod-mcp-ts --client claude

Manual Installation Steps

1. Obtain the source repository via cloning.
2. Install necessary JavaScript packages: npm install
3. Compile the TypeScript source into executable JavaScript: npm run build

Credential Configuration

The server requires authentication via an environment variable holding your secret key:

# For Unix-like shells
export RUNPOD_API_KEY="your_secret_key"

# For Windows Command Prompt
set RUNPOD_API_KEY="your_secret_key"

# For Windows PowerShell
$env:RUNPOD_API_KEY="your_secret_key"

Your confidential key is accessible within the RunPod user settings interface.

Initiating the Service

To launch the local MCP server:

npm start

Integrating with Claude Desktop

1. Launch the Claude Desktop application.
2. Navigate to and edit the configuration file located typically at ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows).
3. Inject the server definition into the mcpServers map:

{
  "mcpServers": {
    "runpod-gateway": {
      "command": "node",
      "args": ["/absolute/path/to/runpod-mcp-server/build/index.js"],
      "env": {
        "RUNPOD_API_KEY": "your_configured_api_key"
      }
    }
  }
}

Crucially, update the args array to point to the absolute location of your compiled build artifact.

1. Save the configuration and relaunch the desktop client for changes to take effect.

Usage Scenarios

Examples demonstrating invocation through the AI agent interface:

Enumerating Deployed Pods

Please retrieve and display a list of all active RunPod compute instances associated with my account.

Provisioning a New Compute Instance

Instantiate a fresh RunPod compute instance adhering to these specifications:
- Instance Label: development-test-01
- Base Image URI: runpod/pytorch:2.1.0-py3.10-cuda11.8.0-devel-ubuntu22.04
- Required GPU Model: NVIDIA GeForce RTX 4090
- GPU Count Allocation: One unit

Setting Up Serverless Inference

Configure a new serverless inference endpoint using the following parameters:
- Endpoint Identifier: high-throughput-model
- Base Template ID: 30zmvf89kd
- Minimum Worker Capacity: 0 instances
- Maximum Worker Capacity: 3 instances

Security Posture

Access to this service necessitates the RunPod API key, which confers broad administrative rights over your cloud tenancy. Adhere strictly to these security protocols:

• Confidentiality: Never disseminate your API key to untrusted parties.
• Operational Prudence: Exercise restraint and verification before executing destructive or costly actions.
• Least Privilege: For enhanced safety, utilize a dedicated API key scoped only to necessary permissions, rather than your primary credential.
• Production Deployment: Implement robust security layering if this tool is integrated into a production workflow.

Licensing

This software is distributed under the MIT License.

WIKIPEDIA: Cloud computing, per ISO definitions, describes "a paradigm for enabling network access to a scalable and elastic pool of shareable physical or virtual resources with self-service provisioning and administration on-demand," often simply termed "the cloud."

== Essential Attributes == In 2011, the U.S. National Institute of Standards and Technology (NIST) formalized five core traits necessary for a system to qualify as cloud-based. These criteria are precisely defined as follows:

On-demand self-service: "A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service provider." Broad network access: "Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, tablets, laptops, and workstations)." Resource pooling: " The provider's computing resources are pooled to serve multiple consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand." Rapid elasticity: "Capabilities can be elastically provisioned and released, in some cases automatically, to scale rapidly outward and inward commensurate with demand. To the consumer, the capabilities available for provisioning often appear unlimited and can be appropriated in any quantity at any time." Measured service: "Cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported, providing transparency for both the provider and consumer of the utilized service. Subsequent updates by the International Organization for Standardization (ISO) have further elaborated upon this foundational list as of 2023.

== Historical Precursors ==

The theoretical foundation for modern cloud services traces back to the 1960s through the adoption of time-sharing concepts, which facilitated remote job execution (RJE). During this era, computing access was centralized around the 'data center' model, where users submitted batch jobs to dedicated system operators managing powerful mainframe computers. This period was characterized by intensive efforts to maximize the utility of expensive computational assets, optimizing infrastructure layers and improving end-user workflow efficiency. The metaphorical depiction of computing services as a "cloud" originated in 1994, utilized by General Magic to visualize the distributed network space accessible by its Telescript mobile agents. This visual shorthand is credited to David Hoffman, a General Magic communications specialist, building upon pre-existing conventions in telecommunications diagrams. The term 'cloud computing' gained wider industry recognition in 1996 following internal documentation from Compaq Computer Corporation outlining strategic future directions for internet-based computation. The company aimed to revolutionize access to computing power...