AgentVerse
Open-source framework for deploying collaborative LLM agent teams and simulating emergent behavior.
About
AgentVerse (OpenBML/AgentVerse) is an open-source framework for deploying multiple LLM-based agents, offering two primary modes: task-solving, where agents collaborate as an automatic multi-agent system on shared goals (e.g., software development teams, consulting), and simulation, where agents interact in custom environments to study emergent behaviors (e.g., NLP classroom, prisoner's dilemma, software design teams). Inspired by human group dynamics, it supports dynamic expert recruitment and collaborative decision-making through horizontal (democratic) or vertical (solver-plus-reviewers) structures.
It has around 5,000 GitHub stars and is Apache 2.0 licensed, with examples spanning HumanEval coding, math reasoning, brainstorming, and tool use. Honest caveat: it is a published-and-stable academic artifact rather than active commercial infrastructure — the repo's last significant updates date to 2024, its "coming soon" docs items remain unchecked, and it is closer to a research notebook (in the lineage of CAMEL-AI and early AutoGen) than to production frameworks like LangGraph. Best suited for researchers and experimenters, not production deployments.
What it can do
- Task-solving framework: collaborative agent teams for coding, math, and reasoning tasks
- Simulation framework: custom environments for studying emergent agent behaviors
- Dynamic expert recruitment with horizontal or vertical decision structures
- Example tasks including HumanEval, logic grids, brainstorming, and tool use
Setup
Clone OpenBML/AgentVerse from GitHub; Python-based with task configs under agentverse/tasks/.
Prerequisites
Python environment and an LLM API key; familiarity with multi-agent concepts helps.
Use cases
Works with
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 71; Trending score 60/100 from researcher assessment.
★ 5,041 GitHub stars
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.