AgentScope

Health v2 · maintenance only

Adoption not part of the score

32785 stars · 3611 forks observed · 2026-10-06

What it is AI-extracted, prompt v1, taxonomy v1, 2026-10-05, confidence 0.97

AgentScope is a Python library for building, orchestrating, and deploying multi‑agent LLM applications. It provides a full‑stack framework with tool middleware, sandboxed workspaces, and a web UI for agent services.

Use cases

• Create a multi‑agent chatbot that can call external tools and APIs.
• Develop a production‑ready LLM agent with human‑in‑the‑loop oversight.
• Deploy sandboxed agents in Docker or Kubernetes for secure execution.
• Integrate retrieval‑augmented generation (RAG) and long‑term memory into agents.
• Build a custom AI assistant that coordinates several specialized agents.

When to choose

• You need a robust, production‑grade framework for multi‑agent LLM workflows.
• Your project requires secure sandboxed tool execution and permission controls.
• You want built‑in support for RAG, vector databases, and long‑term memory.
• You prefer a Python‑first SDK with optional TypeScript bindings for front‑end integration.

When to avoid

• You only need a simple single‑LLM call without agent orchestration.
• Your stack is not Python‑centric or you require a language other than Python.
• You need a lightweight library with minimal dependencies and no sandboxing features.

Facets

library · maturity active

api-framework middleware developer-tools state-management large-language-models python cross-platform multi-agent llm-agent sandboxed-execution mcp react-agent agent-service frontend-ui vector-database-integration ai-agents retrieval-augmented-generation devops docker kubernetes

Member repositories

For agents

markdown · JSON · MCP: product_card(name="agentscope-ai/agentscope")