TL;DR
AgentScope 2.0 is an agent framework with production-ready features including event system, permission management, multi-tenancy, and sandbox, designed to fully utilize LLM reasoning and tool-use capabilities.
Key features
Event System: Unified event bus for frontend and human-in-the-loop support
Permission System: Fine-grained control over tools and resources
Multi-tenancy and Multi-session Service: Production-grade serving with tenant and session isolation
Workspace/Sandbox: Tool and code execution in isolated environments via local, Docker, or E2B backends
Extensible Middleware System: Configurable hooks to customize the agent's reasoning-action loop
When to use it
When deploying multi-agent collaboration systems to production
When fine-grained control over agent tool usage and permissions is needed
When building scalable agent services integrated with various LLM backends