lobehub/lobehub

Open-source agent workspace that organizes AI agents into teams with scheduling, shared memory, plugins and MCP skills; self-hostable via Docker or Vercel.

  • 83.1k GitHub stars
  • TypeScript
  • ⚖️ NOASSERTION
  • 🎯 Intermediate
docker compose up -d
lobehub/lobehub preview image

What it is

LobeHub is a work-and-lifestyle space for finding, building, and collaborating with AI agent teammates, treating agents as the unit of work. It offers an Agent Builder, Agent Groups, scheduled runs, Pages, Projects, Workspaces, and a white-box Personal Memory. It can be self-hosted via Vercel, Zeabur, Sealos, Alibaba Cloud, or Docker, and supports plugins to extend function calling.

Who it's for

  • Developers and users who want a playground for managing multiple AI agents in one place
  • Teams wanting shared agent workspaces with scheduling and project organization
  • Self-hosters who want to deploy their own chatbot instance
  • Plugin developers extending function calling capabilities

Requirements

Requirements

  • An OpenAI API Key (OPENAI_API_KEY is required for the one-click deployment)
  • Docker and Docker Compose for the Docker deployment path
  • pnpm for local development installation (bun is used for the dev:spa script)
  • A GitHub account for the Vercel one-click deploy

Setup

  1. Create a storage folder

    Create a folder for storage files and enter it.

    fish
    $ mkdir lobehub-db && cd lobehub-db
  2. Initialize the LobeHub infrastructure

    Run the setup script to init the infrastructure.

    fish
    bash <(curl -fsSL https://lobe.li/setup.sh)
  3. Start the service

    Start the LobeHub service with Docker Compose.

    fish
    docker compose up -d
  4. Local development

    Clone the repo, install dependencies, and run the dev servers (full-stack, or SPA frontend only on port 9876).

    fish
    $ git clone https://github.com/lobehub/lobehub.git
    $ cd lobehub
    $ pnpm install
    $ pnpm dev          # Full-stack (Next.js + Vite SPA)
    $ bun run dev:spa   # SPA frontend only (port 9876)

Examples

Docker self-hosting

fish
fish
$ mkdir lobehub-db && cd lobehub-db
bash <(curl -fsSL https://lobe.li/setup.sh)
docker compose up -d

What it does: The three documented steps to deploy LobeHub on a private device with Docker.

Customize the model list

text
text
OPENAI_MODEL_LIST=qwen-7b-chat,+glm-6b,-gpt-3.5-turbo

What it does: Uses + to add a model, - to hide a model, and model_name=display_name to rename; the value format follows the README's example.

Override the OpenAI base URL

text
text
OPENAI_PROXY_URL=https://api.chatanywhere.cn

What it does: Overrides the default OpenAI API base URL (https://api.openai.com/v1) when using a proxy; example value taken from the README.

Pros & cons

Pros

  • Pro:Multiple deployment options: Vercel, Zeabur, Sealos, Alibaba Cloud, RepoCloud, and Docker
  • Pro:Agent Groups, scheduling, and shared Workspaces support team-style multi-agent work
  • Pro:White-box, editable Personal Memory gives users control over what agents remember
  • Pro:Extensible via a plugin system with SDK, template, and gateway, plus a library of 10,000+ skills and MCP-compatible plugins

Cons

  • Con:README states the project is under active development, so issues may be encountered
  • Con:One-click deployment requires an OpenAI API key; obtaining one may involve third-party proxies that the project disclaims responsibility for
  • Con:Vercel one-click deploys create a new project rather than a fork, causing constant 'updates available' prompts unless redeployed per the sync guide

Images