Guide local LLM setup and operation with the Harbor toolkit.
npx skills add https://github.com/av/harbor --skill run-llms
Please help me install this Agent Skill. Skill: run-llms Source: https://github.com/av/harbor/tree/main/skills/run-llms Command: npx skills add https://github.com/av/harbor --skill run-llms
This Skill provides a comprehensive guide for setting up and running local Large Language Models (LLMs) using the Harbor toolkit. It helps users troubleshoot common issues and integrate various local AI services.
Run the 'run-llms' skill to start the setup process for local LLMs using Harbor.
High-intent search queries and answers about installing and using this skill.
To set up local LLMs using Harbor, you need to install Docker and Docker Compose first. The run-llms skill then guides you through configuring Harbor and pulling models from Ollama, HuggingFace, and llama.cpp registries to build your local AI service.
Running local AI services with Harbor requires Docker, Docker Compose, and basic knowledge of local AI environments. You will use these dependencies to deploy integrated services like Open WebUI, SearXNG, and Open Terminal for a complete local AI toolkit.
Yes, you can manage models from HuggingFace and Ollama in one local setup. The Harbor toolkit integrates model management across multiple registries, allowing you to pull and configure models from Ollama, HuggingFace, and llama.cpp within a single environment.
To troubleshoot GPU optimization and VRAM usage for local LLMs, the Harbor setup guide provides specific tips for resolving common GPU-related issues. It helps you configure Docker and local AI services to maximize VRAM efficiency during model execution.
Yes, Harbor works with Open WebUI and SearXNG for local chat services. The toolkit supports service integration with Open WebUI, SearXNG, and Open Terminal, allowing you to handle chat and code execution through a comprehensive local AI interface.
diego-tobalinaGenerate Dockerfiles and Docker Compose configurations for Spring Boot, Node.js, and React applications.
crycriMInstall and update community plugins on a local Open WebUI instance via its API.
inho-teamConfigure local LLM servers and integrate them with Claude Code.
danielkhigayDeploys and validates multi-container Docker environments for local development.
