AgentVerse
StaleDescription
AgentVerse is a multi-agent deployment framework by Tsinghua OpenBMB, offering task-solving and simulation paradigms for collaborative multi-LLM-agent systems.
Key Features
- Dual framework support for task-solving (multi-agent collaboration) and simulation (environment observation)
- Configurable agent definitions via YAML with role, tool, and environment customization
- Built-in recruiter component for dynamically assembling agent teams based on task requirements
- Pre-built simulation environments including NLP Classroom, Prisoner's Dilemma, and Software Design
- Support for multiple LLM backends including OpenAI API and local models like LLaMA and Vicuna
- ICLR 2024 accepted research with published paper demonstrating emergent multi-agent behaviors
Use Cases
💡 Multi-agent software development with code writers, testers, and reviewers collaborating iteratively
💡 Social behavior research observing emergent interactions among LLM-based agents
💡 Database diagnosis with specialized expert agents analyzing root causes and suggesting optimizations
💡 Educational simulations like NLP classrooms with professor and student agent interactions
💡 Text evaluation using multi-agent referee teams for automated quality assessment
Strengths & Limitations
✅ Strengths
- • High community interest (5.1k stars)
- • Permissive open-source license (Apache-2.0)
- • Established track record (3 years in production)
- • Responsive to issues, low backlog
⚠️ Limitations
- • No updates in over 23 months
Categories
Quick Start
Clone the repository, install dependencies with pip install -r requirements.txt. Define agents and environment in a YAML config file under agentverse/. Run a task-solving example: python agentverse_command/main.py --task tasksolving/<task_name>. For simulation: python agentverse_command/main_simulation_gui.py --task simulation/<task_name>.