AgentScope — Agent Frameworks

What is AgentScope?

AgentScope starts where most frameworks start — an agent that reasons, calls a tool, looks at the result and goes round again — and then supplies the parts that usually have to be built afterwards. Tools run inside an isolated environment rather than in your process; context is compressed and offloaded as it grows instead of simply overflowing; a person can be asked to approve a step; and the finished agent can be served as a multi-tenant backend with its own interface. The price of that reach is that version 2.0 was a deliberate break from 1.0, so material written for the older releases does not carry over.

What can you do with AgentScope?

Run tools somewhere they cannot hurt you — Tool execution is isolated in a container or a remote sandbox, and the isolation can be scoped per user, per agent or per session.
Keep one toolkit for functions, MCP servers and skills — Your own functions, tools reached over MCP and packaged skills are registered in the same place, so the agent sees one list.
Compress context instead of overflowing it — Long histories are summarised or moved out of the prompt and retrieved again when needed, rather than being truncated at the limit.
Put a person in the path — A run can stop and ask for review before a tool is allowed to proceed, which is the check that matters once an agent can change things.
Change behaviour without editing the agent — Middleware wraps calls at runtime, so logging, limits or overrides can be added around an agent you would rather not modify.
Serve it as a backend, not a script — The same agent can be exposed as a service with tenant separation, distributed execution and a browser interface for watching runs.

Before you choose AgentScope

Version 2.0 is described by the project as a breaking change from 1.0, so tutorials, blog posts and sample code written for the earlier releases will not run as written.
Most of what distinguishes it — sandboxed execution, the service layer, tenant separation — assumes Docker or Kubernetes, so a single script exercises only a small part of the framework.

Frequently asked questions

Is AgentScope free for commercial use?

AgentScope is released under the Apache-2.0 licence — OSI-approved open source, which permits commercial use.

How can AgentScope be deployed?

AgentScope is available as Self-hosted / Runs locally.

Documentation

Reproduced from the agentscope-ai/agentscope README, published under Apache-2.0. Read the original ↗

中文主页 | Documentation | Roadmap

What is AgentScope 2.0?

AgentScope 2.0 is a production-ready, easy-to-use agent framework with essential abstractions that keep up with rising model capability.

We design for increasingly agentic LLMs. Our approach leverages the models’ reasoning and tool use abilities rather than constraining them with strict prompts and opinionated orchestrations.

News

[2026-08] FEAT: Console supported — test and debug agents in the terminal. Example | Docs
[2026-08] INTE: Feishu (Lark) and Discord channels supported. Feishu | Discord
[2026-08] FEAT: Channels supported — connect agents to IM platforms in agent service. Example | Docs
[2026-08] INTE: GitHub MCP Registry and ClawHub supported as built-in hubs. Example | Docs
[2026-08] FEAT: MCP & Skill Hub supported — browse a hub, install into your library, add to a workspace. Example | Docs
[2026-07] INTE: Daytona-based workspace/sandbox supported. Docs
[2026-07] INTE: K8s, OpenSandbox-based workspace/sandbox supported. Docs
[2026-07] INTE: ReMe long-term memory supported. Example | Docs
[2026-06] FEAT: Agentic Memory supported. Example | Docs
[2026-06] FEAT: Distributed & Multi-Tenancy & Multi-Session RAG service supported. Docs

More news →

Community

Welcome to join our community on

Discord	DingTalk

Quickstart

Installation

AgentScope requires Python 3.11 or higher.

From PyPI

uv pip install agentscope

From source

# Pull the source code from GitHub
git clone -b main https://github.com/agentscope-ai/agentscope.git

# Install the package in editable mode
cd agentscope

uv pip install -e .

Agent

The SDK layer — compose an agent from a rich set of building blocks:

Building block	What’s inside
ReAct	Reasoning-acting loop with structured output, realtime interruption & resume, and batched (sequential / concurrent) tool acting
Toolkit	Agentic tool management over Python tools, MCP servers, and skills; ships with built-in coding tools (shell, file edit, search) and task/plan tools
Model	LLM, embedding, and TTS across major providers (OpenAI, Anthropic, Gemini, DashScope, DeepSeek, Moonshot, xAI, Ollama)
Context	Automatic compaction, tool-result offload, and context injection (system prompt, RAG, memory) via built-in middleware
Event System	Unified event bus streaming reasoning, tool calls, and multimodal content (text, image, audio) to the frontend
Permission & HITL	Fine-grained control over tools and resources, confirmation, bypass mode
Middleware	Composable hooks across the loop — reply, reasoning, acting, model calling, permission checking, context compression, system prompt
Memory	Agentic memory with switchable backends (ReMe, Mem0)
Workspace / Sandbox	Isolated tool & code execution — local, Docker, Apple Container, Bubblewrap, E2B, OpenSandbox, Daytona, K8s

Start your first agent with AgentScope 2.0 in console:

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())

Agent Service — All You Need to Build Your App

AgentScope ships a batteries-included agent service — a FastAPI backend with a pre-built Web UI (examples/web_ui) that turns your agents into a multi-tenant, multi-session application, with rich capabilities out of the box:

Capability	What you get
Serving	Multi-tenancy, multi-session isolation, FastAPI backend, pre-built Web UI
Agent Team	Leader–worker orchestration, built-in team tools, task planning
Channels	Connect agents to IM platforms — Feishu (Lark), Discord, custom channels, message routing
RAG Service	Blob storage, index worker, multi-tenant retrieval
MCP & Skill Hub	Browse hubs (GitHub MCP Registry, ClawHub), install into your library, add to a workspace
Resource Sharing	Group- and org-level management for sharing models, MCP servers, skills, and workspaces
Persistence	SQL & NoSQL persistence of agent state and sessions
Scheduling	Scheduled tasks, agent wakeup, background task offloading

Everything above is composable, so you can assemble your own application on top of the service with minimal glue code.

Run the following commands to start the agent service backend and the web UI:

git clone -b main https://github.com/agentscope-ai/agentscope.git
cd agentscope/examples/agent_service

# start the agent service backend
python main.py

Then open another terminal to start the web UI:

cd agentscope/examples/web_ui

# start the webui
pnpm install
pnpm dev

Publications

If you find our work helpful for your research or application, please cite our papers.

@article{agentscope_v1,
    author  = {Dawei Gao, Zitao Li, Yuexiang Xie, Weirui Kuang, Liuyi Yao, Bingchen Qian, Zhijian Ma, Yue Cui, Haohao Luo, Shen Li, Lu Yi, Yi Yu, Shiqi He, Zhiling Luo, Wenmeng Zhou, Zhicheng Zhang, Xuguang He, Ziqian Chen, Weikai Liao, Farruh Isakulovich Kushnazarov, Yaliang Li, Bolin Ding, Jingren Zhou}
    title   = {AgentScope 1.0: A Developer-Centric Framework for Building Agentic Applications},
    journal = {CoRR},
    volume  = {abs/2508.16279},
    year    = {2025},
}

@article{agentscope,
    author  = {Dawei Gao, Zitao Li, Xuchen Pan, Weirui Kuang, Zhijian Ma, Bingchen Qian, Fei Wei, Wenhao Zhang, Yuexiang Xie, Daoyuan Chen, Liuyi Yao, Hongyi Peng, Zeyu Zhang, Lin Zhu, Chen Cheng, Hongzhu Shi, Yaliang Li, Bolin Ding, Jingren Zhou}
    title   = {AgentScope: A Flexible yet Robust Multi-Agent Platform},
    journal = {CoRR},
    volume  = {abs/2402.14034},
    year    = {2024},
}