AgentScope
AgentScope 2.0 is a production-ready, easy-to-use agent framework for building agentic applications. It provides essential abstractions and building blocks to compose agents, and includes a batteries-included agent service for deploying multi-tenant applications. It is designed for developers who want to leverage LLMs' reasoning and tool use abilities without strict prompts or opinionated orchestrations.
✨ Key features
- ReAct agent loop with structured output and interruption/resume
- Toolkit for Python tools, MCP servers, and skills
- Model support for major providers (OpenAI, Anthropic, Gemini, etc.)
- Event system streaming reasoning, tool calls, and multimodal content
- Agent service with FastAPI backend and pre-built Web UI
- Channels to connect agents to IM platforms (Feishu, Discord, DingTalk)
🎯 Use cases
- Build a conversational agent with tool use and permission control
- Deploy a multi-tenant, multi-session agent application with web UI
- Create agent teams with leader-worker orchestration and task planning
- Integrate agents with IM platforms like Feishu or Discord
- Implement RAG services with blob storage and multi-tenant retrieval
📦 Installation
🧰 Requirements: Python 3.11 or higher. API keys for model providers (e.g., DASHSCOPE_API_KEY) are needed for using cloud models.
Installation
AgentScope requires Python 3.11 or higher.
From PyPI
uv pip install agentscope
From source
git clone -b main https://github.com/agentscope-ai/agentscope.gitcd agentscopeuv pip install -e .🚀 Usage
from agentscope.agent import Agent
from agentscope.console import launch_console
from agentscope.tool import Toolkit, Bash, Grep, Glob, Read, Write, Edit
from agentscope.credential import DashScopeCredential
from agentscope.model import DashScopeChatModel
import os, asyncio
async def main() -> None:
agent = Agent(
name="Friday",
system_prompt="You're a helpful assistant named Friday.",
model=DashScopeChatModel(
credential=DashScopeCredential(
api_key=os.environ["DASHSCOPE_API_KEY"]
),
model="qwen3.6-plus",
),
toolkit=Toolkit(
tools=[
Bash(),
Grep(),
Glob(),
Read(),
Write(),
Edit(),
]
),
)
# Chat with the agent in the terminal — streamed output, tool-call
# confirmation and Ctrl+C interruption are all handled for you
await launch_console(agent)
asyncio.run(main())
❓ FAQ
What Python version is required?
AgentScope requires Python 3.11 or higher.
How can I install AgentScope?
You can install it from PyPI with uv pip install agentscope or from source by cloning the repository and running uv pip install -e ..
What model providers are supported?
The README mentions support for OpenAI, Anthropic, Gemini, DashScope, DeepSeek, Moonshot, xAI, and Ollama.
Can I deploy agents to IM platforms?
Yes, AgentScope supports channels to connect agents to Feishu (Lark), Discord, DingTalk, and custom channels.
📊 Repository
🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.