AgentScope — la plateforme multi-Agent flexible d’Alibaba

Quick Start — Two-Agent Debate

sequentialpipeline exécute les Agents dans l’ordre, en passant la sortie de l’un en entrée du suivant. Pour des topologies plus complexes, utilisez msghub (broker de messages partagé) ou IfElsePipeline pour le routage conditionnel. init() lit une liste de configs de modèles — changer de modèle est un changement de config, pas de code.

# pip install agentscope
import agentscope
from agentscope.agents import DialogAgent, UserAgent
from agentscope.message import Msg
from agentscope.pipelines import sequentialpipeline

# Initialize with a model config
agentscope.init(
    model_configs=[{
        "config_name": "gpt-4o-mini",
        "model_type": "openai_chat",
        "model_name": "gpt-4o-mini",
        "api_key": "sk-...",
    }],
)

bull = DialogAgent(
    name="Bull",
    sys_prompt="You argue that AI agents are overhyped. Be concise and cite one concrete data point per turn.",
    model_config_name="gpt-4o-mini",
)
bear = DialogAgent(
    name="Bear",
    sys_prompt="You argue AI agents will transform every industry within 3 years. Concise, data-backed.",
    model_config_name="gpt-4o-mini",
)

topic = Msg(name="moderator", content="Resolved: multi-agent systems will dominate enterprise SW by 2028.", role="user")
# 3 rounds of Bull → Bear using a simple pipeline
for _ in range(3):
    topic = sequentialpipeline([bull, bear], x=topic)

print("Final:", topic.content)

Fonctionnalités clés

Comparaison

Cas d'usage

Tarification et licence

Assets associés sur TokRepo

AgentScope Studio — Agent Tracing UI

AgentScope Studio is a local visualization toolkit: Projects/Runs UI, OpenTelemetry-based tracing, and eval dashboards for AgentScope agents.

AgentScope — Distributed Multi-Agent Platform

AgentScope is a multi-agent framework supporting distributed agent communication, built-in fault tolerance, and an actor-based runtime for building complex multi-agent applications at scale.

Questions fréquentes