Discover What Agents
Can Build
Explore a curated collection of open-source Python agent examples—from lightweight
command-line utilities to full-stack, deployable applications—to kickstart your agent
development journey!
Alias
A versatile agent supporting ReAct, Planner-Worker, and built-in specialized agents.
Data Juicer Agent
A multi-agent system leveraging Data-Juicer to generate data processing pipelines from natural language.
Browse-use Agent
A command-line browser automation agent built with AgentScope.
Browse-use Agent Pro
An enhanced pure-Python browser agent for complex web tasks.
Browse-use Agent Full-stack
A full-stack browser automation app with secure sandbox and web interface.
Deep Research Agent
A pure-Python multi-agent system for conducting in-depth research.
Deep Research Agent Full-stack
A full-stack deep research application powered by Qwen and LangGraph.
Werewolves Game
A multi-agent simulation of the social deduction game "Werewolves".
Chatbot
A basic single-agent conversational chatbot.
Chatbot Full-stack
A deployable chatbot with frontend UI and backend API.
Multi-agent Conversation
Structured conversation among multiple agents on a shared topic.
Multi-agent Debate
Agents engage in structured debates with opposing viewpoints.
ACE Bench
Systematic evaluation of agent capabilities using ACE Bench.
Evotraders
A self-evolving multi-agent trading system with real-time and backtesting support.
Tune: Math Agent
Quick-start example for tuning a math-solving agent with AgentScope Tuner.
Tune: Frozen Lake
Teaches an agent to navigate the Frozen Lake environment via multi-step planning.
Tune: Learn to Ask
Uses LLM as a judge to improve the agent's question-asking strategy during tuning.
Tune: Email Search
Enhances an agent's ability to use email search tools through tuning.
Tune: Werewolves Game
Optimizes strategies and behaviors in the Werewolves multi-agent game.
Tune: Data Augmentation
Generates high-quality synthetic data to support agent tuning.