agentscope and agent-runtimes
AgentScope provides a complete framework for building and executing multi-agent systems with transparency and control, while Agent Runtimes offers a lightweight protocol abstraction layer for exposing pre-built agents—making them complements that could be used together, with AgentScope handling agent development and Agent Runtimes handling their deployment interfaces.
About agentscope
agentscope-ai/agentscope
Build and run agents you can see, understand and trust.
Provides built-in support for ReAct agents, multi-agent orchestration via message hub, and ecosystem integrations including MCP (Model Context Protocol), A2A (Agent-to-Agent) protocol, and voice capabilities. Designed for model-centric reasoning rather than prompt-constrained workflows, with production deployment options including serverless, Kubernetes, and local execution with integrated OpenTelemetry observability.
About agent-runtimes
datalayer/agent-runtimes
🤖 🚀 Agent Runtimes - Expose AI Agents through multiple protocols.
Supports multiple agent frameworks (Pydantic AI, LangChain, Jupyter AI) through unified adapters and exposes them via protocol abstraction (ACP, Vercel AI SDK, MCP-UI, A2A) without code changes. Built on FastAPI with a tool registry for MCP and custom tools, plus React components (ChatBase, ChatSidebar, ChatFloating) for frontend integration. Includes cloud runtime management via Zustand for launching compute resources and orchestrating notebook/document editor AI assistants.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work