harbor
One command brings a complete pre-wired LLM stack with hundreds of services to explore.
open-sourcetool-integration
2.5k
Stars
+45
Stars/month
10
Releases (6m)
Star Growth
+10 (0.4%)
Overview
Harbor是一个一键部署LLM技术栈的工具，通过单个命令即可启动包含数百个服务的完整预配置环境。该工具同时提供NPM和PyPI包，支持多种编程语言，使用Docker Compose文件来管理服务编排。Harbor旨在简化LLM开发和实验环境的搭建过程，让开发者能够快速获得一个功能齐全的AI开发环境，而无需手动配置各种依赖服务。项目在GitHub上获得了2500+星标，拥有活跃的Discord社区支持，表明其在AI开发者中具有一定的认可度和用户基础。
Deep Analysis
Key Differentiator
The all-in-one local LLM stack orchestrator — spin up 30+ pre-wired services (backends, frontends, RAG, voice, images) with a single harbor up command
⚡ Capabilities
- • One-command local LLM stack deployment
- • 30+ service orchestration (backends, frontends, satellites)
- • MCP ecosystem management via MetaMCP
- • Web RAG and deep research integration
- • Image generation with ComfyUI + Flux
- • Voice chat with Speaches
- • Docker Compose auto-orchestration
- • Cross-service pre-wired connectivity
🔗 Integrations
Ollamallama.cppvLLMOpen WebUISearXNGComfyUIDifyLangFlowPerplexicaTabbyAPIAphroditeSGLangDocker
✓ Best For
- ✓ Developers wanting full local LLM stack without manual setup
- ✓ Teams evaluating multiple inference backends side-by-side
- ✓ Privacy-conscious users running AI completely locally
✗ Not Ideal For
- ✗ Cloud-first deployments
- ✗ Users wanting a single simple chatbot
Languages
Shell/BashTypeScriptPython
Deployment
CLI (npm/pip)Docker ComposeLocal machine
Pricing Detail
Free: Fully open-source
Paid: N/A
⚠ Known Limitations
- ⚠ Requires Docker and significant disk space
- ⚠ GPU needed for local model inference
- ⚠ Complex service dependencies when running many services
- ⚠ Linux/macOS focused (Windows support via WSL)
Pros
- + 一键部署完整LLM技术栈，极大简化环境搭建
- + 提供数百个预配置服务，覆盖AI开发全流程
- + 支持多语言环境（NPM和PyPI），适配不同开发栈
Cons
- - 文档信息有限，具体功能和配置选项不够清晰
- - 可能存在资源占用较大的问题（数百个服务）
- - 对Docker环境有依赖，需要一定的容器化基础
Use Cases
- • AI研究人员快速搭建实验环境进行模型测试
- • 开发团队建立统一的LLM开发和测试环境
- • 教育场景中为学生提供完整的AI开发实践平台
Getting Started
1. 安装Harbor包：npm install @avcodes/harbor 或 pip install llm-harbor；2. 确保Docker和Docker Compose已安装并运行；3. 执行Harbor启动命令来部署完整的LLM服务栈