agentscope Documentation

1. Project Overview & Quickstart (agentscope-ai/agentscope) Copy Chapter

File: README.md Copy Chapter

中文主页 | Documentation | Roadmap

What is AgentScope 2.0? Copy Chapter

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 Copy Chapter

• [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 Copy Chapter

Welcome to join our community on

Quickstart Copy Chapter

Installation Copy Section

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 Copy Chapter

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

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 Copy Chapter

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:

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

Contributing Copy Chapter

We welcome contributions from the community! Please refer to our CONTRIBUTING.md for guidelines
on how to contribute.

License Copy Chapter

AgentScope is released under Apache License 2.0.

Publications Copy Chapter

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

• AgentScope 1.0: A Developer-Centric Framework for Building Agentic Applications

• AgentScope: A Flexible yet Robust Multi-Agent Platform

@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},
}

Contributors Copy Chapter

All thanks to our contributors:

File: docs/NEWS.md Copy Chapter

• [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
• [2026-06] FEAT: RAG supported. Example | Docs
• [2026-06] INTE: Mem0 supported. Example | Docs
• [2026-06] FEAT: Agent Team supported. Example | Docs
• [2026-05] RELS: AgentScope 2.0 released! Docs

File: docs/NEWS_zh.md Copy Chapter

• [2026-08] 功能: 支持 Console —— 在终端中测试与调试智能体。样例 | 文档
• [2026-08] 集成: 支持飞书（Lark）与 Discord 频道。飞书 | Discord
• [2026-08] 功能: 支持消息频道 —— 将智能体接入即时通讯平台。样例 | 文档
• [2026-08] 集成: 内置集成 GitHub MCP Registry 与 ClawHub。样例 | 文档
• [2026-08] 功能: 支持 MCP & Skill Hub —— 浏览 hub、安装到个人库、再添加到工作区。样例 | 文档
• [2026-07] 集成: 集成 Daytona 工作区/沙箱实现。 文档
• [2026-07] 集成: 集成 K8s，OpenSandbox 工作区/沙箱实现。 文档
• [2026-07] 集成: 集成 ReMe 长期记忆。 样例 | 文档
• [2026-06] 功能: 支持 Agentic Memory。 样例 | 文档
• [2026-06] 功能: 支持分布式 & 多租户 & 多会话 RAG 服务。 文档
• [2026-06] 功能: 支持多模态 RAG。 样例 | 文档
• [2026-06] 集成: 集成 Mem0 长期记忆。 Example | Docs
• [2026-06] 功能: 支持 Agent Team。样例 | 文档
• [2026-05] 发布: AgentScope 2.0 已发布！文档

File: docs/roadmap.md Copy Chapter

Roadmap

Long-term Goals Copy Chapter

Offering agent-oriented programming (AOP) as a new programming paradigm to organize the design and implementation of next-generation LLM-empowered applications.

Current Focus (January 2026 - ) Copy Chapter

🎙️ Voice Agent Copy Section

Voice agents are a domain we are highly focused on, and AgentScope will continue to invest in this direction.

AgentScope aims to build production-ready voice agents rather than demonstration prototypes. This means our voice agents will:

• Support production-grade deployment, including seamless frontend integration
• Support tool invocation, not just voice conversations
• Support multi-agent voice interactions

Development Roadmap

Our development strategy for voice agents consists of three progressive milestones:

1. TTS Models → 2. Multimodal Models → 3. Real-time Multimodal Models

Phase 1: TTS (Text-to-Speech) Models

• Build TTS model base class infrastructure

  • Design and implement a unified TTS model base class
  • Establish standardized interfaces for TTS model integration

• Horizontal API expansion

  • Support mainstream TTS APIs (e.g., OpenAI TTS, Google TTS, Azure TTS, ElevenLabs, etc.)
  • Ensure consistent behavior across different TTS providers

Phase 2: Multimodal Models (Non-Realtime)

• Enable ReAct agents with multimodal support

  • Integrate multimodal models (e.g., qwen3-omni, gpt-audio) into existing ReAct agent framework
  • Support audio input/output in non-realtime mode

• Advanced multimodal agent capabilities

  • Enable tool invocation within multimodal conversations
  • Support multi-agent workflows with multimodal communication

Phase 3: Real-time Multimodal Models

• Beyond request-response: Explore streaming, interrupt handling, and concurrent multimodal processing
• New programming paradigms: Design agent programming models specifically tailored for real-time interactions
• Production readiness: Ensure low-latency performance, stability, and scalability for production deployment

🛠️ Agent Skill Copy Section

Provide production-ready agent skill integration solutions.

🌐 Ecosystem Expansion Copy Section

• A2UI (Agent-to-UI): Enable seamless agent-to-user interface interactions
• A2A (Agent-to-Agent): Enhance agent-to-agent communication capabilities

🚀 Agentic RL Copy Section

• Support using Tinker backend to tune agent applications on devices without GPU.
• Support tuning agent applications based on their run history.
• Integrate with AgentScope Runtime to provide better environment abstraction.
• Add more tutorials and examples on how to build complex judge functions with the help of evaluation module.
• Add more tutorials and examples on data selection and augmentation.

📈 Code Quality Copy Section

Continuous refinement and improvement of code quality and maintainability.

Completed Milestones

AgentScope V1.0.0 Roadmap Copy Section

We are deeply grateful for the continuous support from the open-source community that has witnessed AgentScope's
growth. Throughout our journey, we have maintained developer-centric transparency as our core principle,
which will continue to guide our future development.

As the AI agent ecosystem rapidly evolves, we recognize the need to adapt AgentScope to meet emerging trends and
requirements. We are excited to announce the upcoming release of AgentScope v1.0.0, which marks a significant shift
towards deployment-focused and secondary development direction. This new version will provide comprehensive support for agent developers
with enhanced deployment capabilities and practical features. Specifically, the update will include:

• ✨New Features

  • 🛠️ Tool/MCP

    • Support both sync/async tool functions
    • Support streaming tool function
    • Support parallel execution of tool functions
    • Provide more flexible support for the MCP server

  • 💾 Memory

    • Enhance the existing short-term memory
    • Support long-term memory

  • 🤖 Agent

    • Provide powerful ReAct-based out-of-the-box agents

• 👨‍💻 Development

  • Provide enhanced AgentScope Studio with visual components for developing, tracing and debugging
  • Provide a built-in copilot for developing/drafting AgentScope applications

• 🔍 Evaluation

  • Provide built-in benchmarking and evaluation toolkit for agents
  • Support result visualization

• 🏗️ Deployment

  • Support asynchronous agent execution
  • Support session/state management
  • Provide sandbox for tool execution

Stay tuned for our detailed release notes and beta version, which will be available soon. Follow our GitHub
repository and official channels for the latest updates. We look forward to your valuable feedback and continued
support in shaping the future of AgentScope.

2. Official Technical Reference & Guides (agentscope-ai/docs) Copy Chapter

File: README.md Copy Chapter

AgentScope Documentation

This repository is the source for the unified
AgentScope documentation site. It publishes English and
Chinese documentation for AgentScope and
ReMe with
Mintlify.

Documentation Sets Copy Chapter

The two projects share one site but keep separate content and release models:

ReMe Documentation Copy Chapter

The ReMe repository README is the concise
project entry point. This repository owns the complete public documentation: concepts,
verified workflows, integrations, reference material, troubleshooting, and contribution
guidance.

ReMe documentation should preserve the product principles expressed by its README:

• memory files are user-owned source data;
• indexes, graphs, and caches are rebuildable derived state;
• conversations and resources become correctable, traceable Markdown memory;
• Python, AgentScope, MCP, CLI, HTTP, plugins, and Skills are integration boundaries over
  the same file-native memory model.

The current ReMe content is organized by user goal:

Chinese pages mirror the same structure under reme/latest/zh/. Update both languages
together and keep their intent, headings, examples, and links aligned.

Local Development Copy Chapter

Use an active Node.js LTS release (Node 20 or 22).

Install the Mintlify CLI:

npm i mint

From the repository root, validate the site or start a preview:

mint validate
mint dev

The preview is available at http://localhost:3000 by default.

Repository Structure Copy Chapter

.
├── versions/
│   └── <version>/{en,zh}/   # Versioned AgentScope documentation
├── reme/
│   └── latest/{en,zh}/      # Current ReMe documentation
├── images/                  # Shared static assets
├── scripts/                 # Documentation maintenance scripts
├── docs.json                # Mintlify navigation and redirects
└── CLAUDE.md                # Writing and review guidelines

Content Boundaries Copy Chapter

Keep public documentation focused on user goals and stable behavior:

• Verify ReMe behavior against current code, public schemas, tests, CLI help, and
  reme/config/default.yaml.
• Explain the stable contract and link to authoritative source instead of duplicating
  implementation details.
• Keep concepts focused on product intent; place detailed development mechanics in source
  code, tests, and repository-local guidance.
• Lead with the outcome, use direct language, and introduce every table, image, diagram,
  and code block.
• Start each MDX page with title and description frontmatter.

See CLAUDE.md for the complete writing and component guidelines.

Version Management Copy Chapter

AgentScope documentation is immutable within each published version directory. For a new
AgentScope release:

1. Copy the latest relevant version into a new project version directory.
2. Update every version-specific internal link in the copied pages.
3. Add the version under the matching project tab for both languages in docs.json.
4. Point the relevant latest or stable redirect at the new version.
5. Run mint validate before submitting the change.

AgentScope versions can be created with scripts/create-version.sh.

ReMe does not keep historical release directories. Update reme/latest/{en,zh}/
directly, keep canonical internal links under /reme/latest/..., and leave its
docs.json version label as latest. The /reme/stable/... alias and obsolete
numeric-version URLs redirect to the current pages for compatibility.

Contribution Workflow Copy Chapter

When adding or updating documentation:

1. Verify the behavior in the corresponding project source.
2. Update the correct project, version, and language pages.
3. For ReMe, make the equivalent English and Chinese changes in reme/latest/.
4. Update docs.json when navigation, versions, or compatibility redirects change.
5. Check page links and run mint validate before submitting a pull request.

Run these checks from the repository root before submitting:

git diff --check
mint validate

Resources Copy Chapter

• AgentScope GitHub
• ReMe GitHub
• ReMe English documentation
• ReMe 中文文档
• Mintlify Documentation

File: reme/README.md Copy Chapter

ReMe documentation

ReMe documentation is part of the unified Mintlify site in this repository.

• latest/ contains the current en/ and zh/ MDX pages and is updated in place.
• ReMe does not maintain numbered or historical documentation directories.
• Shared ReMe media lives in images/reme/.
• Navigation is defined by the ReMe tab for each language in docs.json.
• Canonical internal links use /reme/latest/...; compatibility redirects preserve old URLs.

Keep the documentation focused on product philosophy, verified user workflows, and stable contracts. Implementation details belong in the ReMe source, schemas, tests, and runtime help.