av

harbor: Orchestrate a Local LLM Stack

Harbor is a CLI and companion app for bringing up connected local LLM services through Docker Compose. It suits people exploring self-hosted models and AI tools who want shared configuration instead of wiring each service by hand.

ActivePythonApache-2.0AILLMCLI
View on GitHub Compare
Stars
3.2k
Forks
229
Stars, last 30 days
+32
Last commit
11 days ago
Contributors
20

Our take

Activeregular commits and releases

Health score

52/100

How it is scored
Activity
86
Community
30
Issues
44
Pull requests
49

+32 stars in the last 30 days

  • Releases more often than 88% of the projects we track
  • Slower to close issues than 97% of the projects we track

Harbor is an actively maintained, broad local-LLM orchestration project, but its near-total contributor concentration and slow issue closure make it worth evaluating before depending on it for team workflows.

Good fit if

  • You want a convenient way to experiment with self-hosted LLM services rather than assemble and connect Compose files yourself.
  • You value a broad catalog of backends and related tools, with a CLI for managing the stack.
  • Apache-2.0 licensing fits your project, and you can run the required Compose setup.

Look elsewhere if

  • You need a production deployment solution, which the project explicitly says it is not designed to be.
  • You need project maintenance or issue resolution to be distributed across contributors, since most code contributions come from one person.
  • You need a narrower, individually managed service setup and do not want Harbor's orchestration layer.
All health signals
Last commit2026-09-25 (11 days ago)
Commits, last 90 days100+
Releases, last 12 months54 (latest v0.5.14, 2026-09-25)
Contributors20 (top contributor: 98% of commits)
Issues closed, last 90 days21 (typically closed in 570 days)
Project age2 years

Checked on 2026-10-07 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

3,280

Forks

241

Change since Nov 16, 2026: 0 stars·0 forks

Overview

Harbor coordinates a local AI stack, connecting model backends, user interfaces, and supporting services so they can work together with less manual configuration. Its CLI and companion app provide a common way to start services, manage settings, and access logs and data.

It is aimed at experimentation and local development, not production deployment. The project brings together many services, from inference backends and chat interfaces to web search, voice, image generation, and agent tools.

Key Features

  • Starts configured combinations of local LLM backends and frontends with harbor up.
  • Connects alternative inference backends to Open WebUI and shares model-related workflows where supported.
  • Adds optional services for web search and RAG, voice, image generation, MCP tools, and LLM workflows.
  • Provides harbor launch adapters for supported coding and agent CLIs, including model and backend selection.
  • Manages configuration profiles, command history, service logs, and model downloads from one CLI.
  • Offers a companion app and utilities to open services, create access links or QR codes, and export a setup as standalone Docker Compose.

Use Cases

  • Local AI hobbyists can bring up a chat UI with a model backend without manually connecting each service.
  • Developers comparing inference engines can start several supported backends and use a shared set of interfaces and CLI workflows.
  • Coding-agent users can route supported host tools to a local OpenAI-compatible backend through harbor launch.
  • People prototyping local research or automation workflows can add services such as SearXNG, n8n, or an MCP manager to their stack.

What you need

Detected in the repository

  • Node.js (from package.json)
  • Python (from pyproject.toml)
  • A Compose file to run it with its related services
  • A test suite and automated checks on GitHub Actions

License in plain words

Apache-2.0permissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the license, the NOTICE file and a note of your changes
  • Share your changes: no
  • Includes an explicit patent grant from the contributors:

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

After installing Harbor and setting up Docker Compose, start the default stack:

harbor up

See the README and installation guide for installation steps and configuration details.

Alternatives

  • TurboLLM: TurboLLM focuses on serving local models through a browser and compatible APIs, while Harbor brings up a connected stack of LLM services with Docker Compose.
  • textgen: TextGen provides a local model interface and inference APIs, while Harbor orchestrates multiple connected LLM services through Docker Compose.
ProjectLanguageLicenseStarsStatus
harborPythonApache-2.03.2kActive
TurboLLMTypeScript–278New
textgenPythonAGPL-3.047.7kActive

Considerations

Harbor's service catalog and integrations make it useful for experimentation, but the README cautions that it is not a deployment solution. The project is active, with regular commits and releases, yet contributions are highly concentrated in one person and the median time to close issues is long. That can matter if you expect fast issue resolution or need continuity across maintainers.

The project relies on Docker Compose for containerized services and has both Node.js and Python runtime declarations. Hardware needs depend on the selected inference backend and workload; the provided information does not specify a universal hardware requirement. Apache-2.0 permits broad reuse, subject to its license terms.

Found this useful?

Share it with someone who would like harbor.

Comparisons