dstackai/dstackSKILLClaude Skill

dstack

Vendor-agnostic orchestration for training, inference and agentic workloads across NVIDIA, AMD, TPU, and Tenstorrent on clouds, Kubernetes, and bare metal.

2.2kstars

232forks

Python

agent-skillsagentic-orchestrationamdcloudcontainersdockerfine-tuninggpuinferencek8skubernetesllms

Quality signals

ActiveNo testsNo CIThin docsNo licensePopular

Open Repository Download ZIP

AI summary

Auto-generated overview of what this skill does and when to reach for it.

Unified orchestration tool for running GPU workloads across clouds, Kubernetes, and bare metal with any hardware or framework.

Use this when

Need to run ML training across multiple cloud providers
Want to deploy inference services with auto-scaling
Running agentic workloads on heterogeneous GPU hardware
Managing on-prem GPU clusters without vendor lock-in
Provisioning dev environments with GPU acceleration

Don't use when

Running non-GPU workloads
Need a simple local-only scheduler without networking
Prefer cloud-specific managed services

Capabilities

Supports NVIDIA, AMD, TPU, and Tenstorrent accelerators
Orchestrates across clouds, Kubernetes, and bare metal
Auto-scales GPU fleets based on workload demand
Manages persisted volumes for training data
Handles port-forwarding and networking automatically
Provides SSH access to provisioned GPU instances

InstallSKILLInstall as a Claude Code skill

Universal installANY TOOL

$ npx skillcity@latest dstackai/dstack

Auto-detects the skill or MCP server in this repo and installs it for Claude Code, Codex, Cursor or Cline.

Requires Node.js 18+ (ships with npx).

Or install manually

# Install into ~/.claude/skills/
mkdir -p ~/.claude/skills/dstack
curl -fsSL https://raw.githubusercontent.com/dstackai/dstack/main/SKILL.md -o ~/.claude/skills/dstack/SKILL.md

Loading README…

Discover the toolingbehind modern intelligence

dstack