salad_mcp
Manage GPU workloads on SaladCloud, including container groups and inference endpoints. Operate queues, jobs, logs, and quotas to run and monitor deployments. Check CPU/GPU availability to plan capacity and scale efficiently.
Ask AI about salad_mcp
Powered by Claude · Grounded in docs
I know everything about salad_mcp. Ask me about installation, configuration, usage, or troubleshooting.
0/500
Reviews
Documentation
Salad MCP Server
A Model Context Protocol (MCP) server for SaladCloud - the GPU provider with super nice features!
This MCP server provides seamless integration with SaladCloud's API, enabling Claude and other MCP clients to manage GPU workloads, container groups, inference endpoints, and more.
Features
- Container Groups Management: Create, list, update, start, stop, and delete container groups
- Inference Endpoints: List and manage inference endpoints
- Queue Operations: Manage message queues for async workload processing
- Logs & Monitoring: Access system logs and workload errors
- Quotas: Check organization resource quotas
- LLM-as-Judge Auto-Eval: Comprehensive integration testing using Claude API
Installation
Quick Start (Recommended)
Use npx to run without installation:
npx salad-mcp-server
Via Smithery
To install with Smithery:
npx -y @smithery/cli install salad-mcp-server --client claude
From npm
npm install -g salad-mcp-server
salad-mcp-server
From Source
npm install
npm run build
Configuration
Set your SaladCloud API key as an environment variable:
export SALAD_API_KEY=your_api_key_here
export SALAD_ORG_NAME=your_organization_name
For auto-eval testing, also set:
export CLAUDE_API_KEY=your_claude_api_key_here
Usage
As MCP Server
Add to your MCP client configuration (e.g., Claude Desktop):
Using npx (recommended):
{
"mcpServers": {
"salad": {
"command": "npx",
"args": ["salad-mcp-server"],
"env": {
"SALAD_API_KEY": "your_api_key_here",
"SALAD_ORG_NAME": "your_organization_name"
}
}
}
}
Using local installation:
{
"mcpServers": {
"salad": {
"command": "node",
"args": ["/path/to/salad_mcp/dist/index.js"],
"env": {
"SALAD_API_KEY": "your_api_key_here",
"SALAD_ORG_NAME": "your_organization_name"
}
}
}
}
Run Auto-Eval CLI
Test the full integration with LLM-as-judge evaluation:
npm run eval
Run Integration Tests
npm test
Available Tools
list_container_groups: List all container groups in an organization/projectget_container_group: Get details of a specific container groupcreate_container_group: Create a new container groupupdate_container_group: Update an existing container groupstart_container_group: Start a container groupstop_container_group: Stop a container groupdelete_container_group: Delete a container grouplist_inference_endpoints: List available inference endpointsget_quotas: Get organization resource quotaslist_queues: List message queuescreate_queue: Create a new message queueget_queue: Get queue detailsdelete_queue: Delete a queue
API Reference
Based on SaladCloud API Documentation
Sources
License
MIT
