agentscope: A production-ready, easy-to-use agent framework for building and running agents you can see, understand, and trust.
What problem it solves
Developers need abstractions that can accommodate increasingly agentic LLMs without limiting their reasoning and tool-use abilities through rigid prompts or opinionated orchestration. They also need a way to turn agent components into multi-tenant, multi-session applications with supporting infrastructure.
Key features
Composable agent building blocks — Compose agents from abstractions for reasoning and acting, tools, models, context, events, permissions, middleware, memory, and isolated workspaces.
Source: README, src
Production-oriented agent service — The included FastAPI backend and pre-built Web UI provide multi-tenancy, session isolation, persistence, scheduling, resource sharing, and integration with communication channels.
Source: README
Agent teams and task planning — Leaders can delegate work to member agents through team tools, while agents can break complex work into tracked plans and update them as execution proceeds.
Source: README
Realtime and multimodal support — The framework includes realtime voice agents, speech-to-speech capabilities, and event streaming for text, images, audio, reasoning, and tool calls.
Source: README
Controlled execution and context management — Fine-grained permissions, confirmation controls, automatic compaction, tool-result offload, and context injection help govern agent behavior and context handling.
Source: README
Quick start
Copied verbatim from the project's README.
uv pip install agentscope# 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 .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())Who it may suit
Inference: this may suit Python developers building agent applications with composable SDK components. It may also suit teams that need a ready-made backend and Web UI for multi-tenant, multi-session agent deployments.
Topics to discuss
Which agent-service capabilities are most useful for a production deployment?
How should teams balance permission confirmations against unattended execution?
Which model, tool, workspace, and storage integrations should receive more community documentation and examples?
Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.
Read the original README https://github.com/agentscope-ai/agentscope#readme
ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.
Comments
…