Overview
LobeHub is a full-featured AI agent orchestration platform with nearly 80,000 GitHub stars that positions itself as a Chief Agent Operator. It handles hiring, scheduling, and reporting on AI agents with white-box transparent memory, multi-channel IM integration, and a 10,000+ skill plugin ecosystem. Its Agents-as-the-Unit-of-Work philosophy treats agents as persistent, schedulable team members.
Features
- ✓Central operator hub for all agents
- ✓IM Gateway for Telegram, Discord, Slack, Feishu, WeChat
- ✓Agent Builder with one-click setup
- ✓Unified Intelligence across any LLM model
- ✓10,000+ skills and MCP-compatible plugins
- ✓White-box editable and transparent memory
Installation
docker compose up -dPros
- +Most comprehensive agent platform on GitHub
- +White-box memory for transparency and trust
- +Rich scheduling and automation capabilities
- +Extensive plugin and skills ecosystem
- +Self-hostable with flexible deployment options
Cons
- −Heavy infrastructure for simple use cases
- −Steeper setup complexity
- −Requires significant compute and API budget
- −English documentation still growing
Alternatives
Documentation
LobeHub
Overview
LobeHub is an ambitious AI agent orchestration platform positioning itself as a "Chief Agent Operator" for the 2026 agent era. With nearly 80,000 GitHub stars, it organizes AI agents into 7×24 continuous operations — handling hiring, scheduling, and reporting on an entire AI team while keeping the human operator in control.
LobeHub's philosophy — "Agents as the Unit of Work" — treats agents not as chat interfaces but as persistent, schedulable team members. The platform integrates agent creation, deployment, collaboration, and evolution into a single workspace, making it a compelling alternative to point-solution agent frameworks.
Features
- Operator Hub: Central control plane to bring all agents "under one roof"
- IM Gateway: Places agents where you already chat — Telegram, Discord, Slack, Feishu, WeChat
- Agent Builder: Describe what you need once; LobeHub instantiates and configures the agent
- Unified Intelligence: Access any LLM model through a single abstraction layer
- 10,000+ Skills: Extensive plugin ecosystem including MCP-compatible integrations
- Agent Groups: Structured teamwork between multiple agents with shared context
- Schedule: Time-based agent execution — agents work while you sleep
- Personal Memory: Continual learning across sessions with user-specific knowledge
- White-Box Memory: Structured, editable, transparent memory for full visibility
- Workspace & Projects: Organizational structure for team-based agent operations
- Pages: Content creation and knowledge sharing within agent teams
Installation
One-Click Deployment: Available on Vercel, Zeabur, Sealos, RepoCloud, and Alibaba Cloud.
Docker Deployment:
git clone https://github.com/lobehub/lobehub
cd lobehub
docker compose up -d
Local Development:
pnpm install
pnpm dev
Required configuration: OPENAI_API_KEY environment variable. Optional: OPENAI_PROXY_URL.
Quick Start
- Deploy LobeHub via Docker or one-click provider
- Configure API keys for your preferred model providers
- Use the Agent Builder to create your first agent by describing its role
- Assign skills and connect chat channels
- Schedule autonomous tasks
Core Concepts
Agents as the Unit of Work
Unlike chat-centric interfaces where each conversation is isolated, LobeHub treats agents as persistent entities with memory, skills, and schedules. An agent can continue working across sessions, building on prior context.
Agent Groups
Groups enable structured collaboration between agents — similar to a team project. Members share context, delegate tasks, and produce coordinated outputs.
White-Box Memory
Memory is not a black box. LobeHub exposes structured, human-readable memory that operators can inspect, edit, and curate — addressing the trust and transparency concerns common in AI agent systems.
Advanced Features
Custom Agent Configuration
agent:
name: "Research Assistant"
role: "Deep research and synthesis"
memory: true
channels:
- telegram
- discord
skills:
- web-search
- pdf-reader
- report-writer
Scheduled Autonomous Work
schedule:
task: "summarize-news"
cron: "0 8 * * *"
channels:
- telegram
Examples
- 24/7 Content Pipeline: Research agent gathers sources → Writer drafts → Editor reviews, all scheduled autonomously
- Personal Assistant Team: Calendar manager, email triager, and research agent collaborating in a group
- Customer Support Squad: Multiple specialized agents handling tickets with escalation routing
- Data Analysis Workflow: Scheduled agent runs daily reports and notifies on anomalies
Pros
- ✅ Most comprehensive agent platform on GitHub
- ✅ White-box memory for transparency and trust
- ✅ Rich scheduling and automation capabilities
- ✅ Extensive plugin and skills ecosystem
- ✅ Multi-channel IM integration out of the box
- ✅ Self-hostable with flexible deployment options
Cons
- ❌ Heavy infrastructure compared to lightweight frameworks
- ❌ Steeper setup complexity than single-purpose tools
- ❌ Requires significant compute and API budget for full functionality
- ❌ Chinese-language origin may have limited English documentation
- ❌ Overkill for simple single-agent use cases
When to Use
Use LobeHub when you need a full-featured agent orchestration platform with scheduling, memory, multi-agent collaboration, and multi-channel deployment. Best suited for teams building persistent AI operations rather than one-off agent prototypes.