DevOps-Security-Agent-Skills/infrastructure/local-ai/ollama-stack/SKILL.md

name, description, license, metadata

name	description	license	metadata
ollama-stack	Run local LLM workloads with Ollama, Open WebUI, and GPU-aware tuning for private development environments.	MIT	author version devops-skills 1.0

Ollama Stack

Deploy a local LLM stack for offline and privacy-first workflows.

Minimal Setup

curl -fsSL https://ollama.com/install.sh | sh
ollama serve
ollama pull llama3.1:8b
ollama run llama3.1:8b

Docker Compose Pattern

• Ollama container with persistent model volume
• Open WebUI for chat interface
• Optional LiteLLM proxy for unified API routing

Best Practices

• Pin model versions for reproducibility.
• Monitor VRAM, RAM, and swap utilization.
• Restrict network exposure to trusted subnets.

Related Skills

• mac-mini-llm-lab - Apple Silicon optimization
• docker-compose - Service orchestration