AgentScope Runtime
A runtime for agent applications with sandboxed tool execution, streaming Agent-as-a-Service APIs, and deployment locally, on Kubernetes or serverless, now folded into AgentScope 2.0.
AgentScope Runtime is a runtime framework for agent applications, built around three areas: tool sandboxing that runs tool calls inside a hardened sandbox, Agent-as-a-Service APIs that expose an agent as a streaming, production-ready service, and scalable deployment locally, on Kubernetes, or in a serverless environment. It also provides full-stack observability through logs and traces, and compatibility with mainstream agent frameworks.
Its quick start walks through installing from PyPI or source, building a streaming Agent-as-a-Service API with an init / query / shutdown pattern, running Python, shell, GUI, browser, filesystem and mobile tools inside an isolated sandbox, and deploying with a DeployManager reachable over A2A, a Response API, or the OpenAI SDK. The project's own README states that all of its capabilities have been merged into AgentScope 2.0, and that this repository will be archived.