Harbor
Stop configuring your AI stack. Start using it. One command brings a complete pre-wired LLM stack with hundreds of services to explore.
What is Harbor?
Harbor is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to stop configuring your ai stack. start using it. one command brings a complete pre-wired llm stack with hundreds of services to explore.
Stop configuring your AI stack. Start using it. One command brings a complete pre-wired LLM stack with hundreds of services to explore.
This server falls under the Cloud Services category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- Stop configuring your AI stack. Start using it. One command
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx harborConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use Harbor
Harbor is a CLI tool that deploys a complete, pre-wired local LLM stack using Docker Compose with a single command. It bundles hundreds of services — inference engines (Ollama, llama.cpp, vLLM), Open WebUI, SearXNG web search, ComfyUI image generation, speech tools, and an MCP ecosystem manager called MetaMCP — all pre-connected out of the box. Developers use it to skip the tedious manual wiring of local AI infrastructure and get a production-like stack running in minutes on their own hardware.
Prerequisites
- Docker Desktop (or Docker Engine + Docker Compose) installed and running
- Bash-compatible shell (macOS, Linux, or WSL2 on Windows)
- Node.js 18+ for the npx launcher (or install Harbor via the shell script)
- Sufficient disk space for model weights (varies by model, typically 4–20 GB)
- An MCP client such as Claude Desktop or Claude Code to connect to Harbor's MetaMCP service
Install Harbor CLI
Run the one-line installer to download and configure Harbor. The installer adds the `harbor` binary to your PATH and sets up the default config directory at ~/.harbor.
curl -fsSL https://raw.githubusercontent.com/av/harbor/main/install.sh | bashStart the core stack
Bring up Open WebUI and Ollama — the minimal stack needed to start chatting with local models. Harbor pulls the required Docker images on first run.
harbor upAdd optional services
Extend the stack by naming additional services on the same command. For example, add SearXNG for web search and Speaches for voice (OpenAI-compatible TTS/STT), both pre-wired to Open WebUI automatically.
harbor up searxng speachesPull and run a model
Use Harbor's shortcut commands to pull a model via Ollama and open it in the browser. The `harbor open webui` command launches Open WebUI in your default browser.
harbor ollama pull llama3.2
harbor open webuiConnect an MCP client via MetaMCP
Harbor includes MetaMCP, a meta-server that bridges the MCP ecosystem. Start Harbor with MetaMCP enabled, then point your Claude Desktop config to it so Claude can access all MCP servers managed by Harbor.
harbor up metamcpSave and restore configuration profiles
Once you have your preferred set of services running, save the profile so you can recreate the exact same stack on any machine or after a reset.
harbor profile save my-stack
harbor profile load my-stackExport a portable Docker Compose file
Eject your current Harbor setup into a standalone docker-compose.yml so you can run the same stack without Harbor on any Docker host.
harbor eject searxng llamacpp > docker-compose.harbor.ymlHarbor Examples
Client configuration
Connect Claude Desktop to Harbor's MetaMCP service running on localhost. Replace the port if you changed the default MetaMCP port.
{
"mcpServers": {
"harbor-metamcp": {
"url": "http://localhost:8811"
}
}
}Prompts to try
After connecting Claude to Harbor's MetaMCP, try these prompts to exercise the local AI stack.
- "Search the web for the latest news on open-source LLMs using SearXNG"
- "Generate an image of a futuristic city using ComfyUI"
- "List all services currently running in my Harbor stack"
- "Pull the Mistral 7B model and summarize the project README"Troubleshooting Harbor
Docker containers fail to start with a port conflict error
Another service is already using the default port. Stop the conflicting service or override Harbor's port by editing ~/.harbor/.env and changing the relevant PORT variable before running `harbor up` again.
`harbor` command not found after installation
The installer adds Harbor to ~/.local/bin. Run `source ~/.bashrc` (or `~/.zshrc`) to reload your PATH, or add `export PATH="$HOME/.local/bin:$PATH"` to your shell profile manually.
Open WebUI is slow or models take very long to respond
Ensure your Docker Desktop resource limits (CPU/RAM/GPU) are set high enough. For macOS Metal acceleration, use `harbor up dmr mlx` to switch to the MLX backend instead of Ollama.
Frequently Asked Questions about Harbor
What is Harbor?
Harbor is a Model Context Protocol (MCP) server that stop configuring your ai stack. start using it. one command brings a complete pre-wired llm stack with hundreds of services to explore. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install Harbor?
Follow the installation instructions on the Harbor GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with Harbor?
Harbor works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is Harbor free to use?
Yes, Harbor is open source and available under the Apache-2.0 license. You can use it freely in both personal and commercial projects.
Harbor Alternatives — Similar Cloud Services Servers
Looking for alternatives to Harbor? Here are other popular cloud services servers you can use with Claude, Cursor, and VS Code.
Open WebUI
★ 138.2kUser-friendly AI Interface (Supports Ollama, OpenAI API, ...)
Anything LLM
★ 60.4kThe all-in-one AI productivity accelerator. On device and privacy first with no annoying setup or configuration.
LocalAI
★ 46.4kLocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.
Nacos
★ 33.0kan easy-to-use dynamic service discovery, configuration and service management platform for building AI cloud native applications.
Xiaozhi ESP32
★ 26.7k本项目为xiaozhi-esp32提供后端服务，帮助您快速搭建ESP32设备控制服务器。Backend service for xiaozhi-esp32, helps you quickly build an ESP32 device control server.
Gateway
★ 11.8kA blazing fast AI Gateway with integrated guardrails. Route to 1,600+ LLMs, 50+ AI Guardrails with 1 fast & friendly API.
Browse More Cloud Services MCP Servers
Explore all cloud services servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up Harbor in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use Harbor?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.