lobehub/lobehub

LobeHub is a TypeScript-based platform designed to organize and manage AI agents as operational units within a 7×24 work environment. The repository serves as the Chief Agent Operator, enabling users to hire, schedule, and report on entire AI teams while maintaining control without requiring constant online presence. The platform is built to address fundamental limitations in current agent systems, which typically operate as isolated, one-off task-driven tools lacking persistent context and requiring manual coordination between different interfaces and models.

The core architecture treats agents as the fundamental unit of work rather than individual conversations or tasks. The Operator component consolidates multiple agents under a single interface, reducing tool fragmentation and providing an instant messaging gateway where agents can be accessed through existing chat platforms. This consolidation approach aims to increase productivity by eliminating the need to toggle between separate applications and windows.

The Create feature includes an Agent Builder that allows users to describe their needs once, triggering automatic configuration and immediate usability. The platform provides access to over 10,000 skills through a library of tools and MCP-compatible plugins, enabling agents to connect with external services and capabilities. The system supports unified intelligence across any model and modality, giving users centralized control over their AI infrastructure.

Collaboration capabilities are introduced through Agent Groups, which function as real team members working in parallel. The platform includes Pages for multi-agent content refinement with shared context, scheduling functionality for running agents at specified times, Projects for organizing work by initiative, and Workspaces for team-based collaboration with clear ownership and visibility. These features enable structured productivity and iterative improvement across agent teams.

The Evolve component focuses on human-agent co-evolution through Personal Memory systems that build understanding of user needs over time. The memory system is structured and editable, providing white-box transparency rather than opaque black-box learning. Agents adapt their behavior based on user work patterns, enabling them to act at appropriate moments without explicit instruction.

LobeHub supports multiple deployment options including Vercel, Zeabur, Sealos, Alibaba Cloud, and Docker, making self-hosting accessible to users with varying technical capabilities. The platform is actively developed and welcomes community feedback through its GitHub issues channel.