--- title: "Where agents gather · multi-agent worlds, live broadcasts, judiciary patterns · Agent Fieldbook" url: https://agentfieldbook.org/catalog/communities/ description: "The corpus's reference index of multi-agent worlds and collectives: Moltbook, AI Digest's Village, AI Town, Project Sid, MetaGPT, ChatDev, CAMEL, AgentVerse, Sakana AI Scientist, the Smallville paper and the MASS judiciary pattern. Every row is a real gathering with a primary source." section: "Catalogue · Communities" source: Agent Fieldbook — generated from the published page --- # Where agents gather **Live broadcast · Open simulation · Research simulation · Persistent world · Role-based collective · Comms framework · Agent playground · Research collective · Judiciary pattern** Online communities where agents congregate and interact: role-play scenarios, simulations and live broadcasts you can watch. Some are academic research environments, others are built for entertainment. ## Catalogue · Communities Live data - [Live broadcast single-source Moltbook Live agent broadcast sessions ↗ www.moltbook.com](https://www.moltbook.com/) - [Live broadcast confirmed AI Digest · The Village 4+ frontier agents, streamed live ↗ theaidigest.org](https://theaidigest.org/village) - [Open simulation confirmed AI Town Open-source Smallville re-implementation, community-run instances ↗ github.com](https://github.com/a16z-infra/ai-town) - [Research simulation confirmed Generative Agents (Smallville) 25 sandbox agents ↗ arxiv.org](https://arxiv.org/abs/2304.03442) - [Persistent world confirmed Project Sid (Altera) 1,000+ agent civilisation ↗ altera.al](https://altera.al/) - [Role-based collective confirmed MetaGPT 5-7 role agents (PM, engineer, QA, architect) ↗ github.com](https://github.com/geekan/MetaGPT) - [Role-based collective confirmed ChatDev Multi-role virtual software firm ↗ github.com](https://github.com/OpenBMB/ChatDev) - [Comms framework confirmed CAMEL Role-play agent pairs, extensible society ↗ www.camel-ai.org](https://www.camel-ai.org/) - [Agent playground confirmed AgentVerse Task-oriented multi-agent framework + leaderboard ↗ github.com](https://github.com/OpenBMB/AgentVerse) - [Research collective confirmed Sakana AI Scientist Autonomous paper-generation agent collective ↗ sakana.ai](https://sakana.ai/ai-scientist/) - [Judiciary pattern confirmed MASS Judicial Committee N judge-agents · quorum verdict aggregation ↗ arxiv.org](https://arxiv.org/abs/2502.02533) ## 11 rows Live data · click a row for its full spec | name | kind | population | url | needs resource | src | | --- | --- | --- | --- | --- | --- | | Moltbook | live-broadcast | Live agent broadcast sessions | [↗ link](https://www.moltbook.com/) | Source is a deny-listed host; re-source to a primary record before restoring confirmed. | [↗ source](https://www.moltbook.com/) | | AI Digest· The Village | live-broadcast | 4+ frontier agents, streamed live | [↗ link](https://theaidigest.org/village) | – | [↗ T2](https://theaidigest.org/village) | | AI Town | open-simulation | Open-source Smallville re-implementation, community-run instances | [↗ link](https://github.com/a16z-infra/ai-town) | – | [↗ T1](https://github.com/a16z-infra/ai-town) | | Generative Agents (Smallville) | research-simulation | 25 sandbox agents | [↗ link](https://arxiv.org/abs/2304.03442) | – | [↗ T1](https://arxiv.org/abs/2304.03442) | | Project Sid (Altera) | persistent-world | 1,000+ agent civilisation | [↗ link](https://altera.al/) | – | [↗ T1](https://arxiv.org/abs/2411.00114) | | MetaGPT | role-based-collective | 5-7 role agents (PM, engineer, QA, architect) | [↗ link](https://github.com/geekan/MetaGPT) | – | [↗ T1](https://arxiv.org/abs/2308.00352) | | ChatDev | role-based-collective | Multi-role virtual software firm | [↗ link](https://github.com/OpenBMB/ChatDev) | – | [↗ T1](https://arxiv.org/abs/2307.07924) | | CAMEL | comms-framework | Role-play agent pairs, extensible society | [↗ link](https://www.camel-ai.org/) | – | [↗ T1](https://arxiv.org/abs/2303.17760) | | AgentVerse | agent-playground | Task-oriented multi-agent framework + leaderboard | [↗ link](https://github.com/OpenBMB/AgentVerse) | – | [↗ T1](https://arxiv.org/abs/2308.10848) | | Sakana AI Scientist | research-collective | Autonomous paper-generation agent collective | [↗ link](https://sakana.ai/ai-scientist/) | – | [↗ T1](https://arxiv.org/abs/2408.06292) | | MASS Judicial Committee | judiciary-pattern | N judge-agents· quorum verdict aggregation | [↗ link](https://arxiv.org/abs/2502.02533) | – | [↗ T1](https://arxiv.org/abs/2502.02533) | ## Field notes Multi-agent worlds, live broadcasts, role-based collectives and judiciary patterns, every row a real gathering with a primary source From persistent civilisations and live broadcasts to role-based software firms and quorum-verdict judiciaries. They are the closest thing the field has to an observable laboratory for agent-to-agent behaviour. Agents that perform well alone degrade in groups through repetition, drift and deadlock, and the mechanisms that fix it, roles, turn-taking, adjudication, are the same ones human institutions arrived at. Expect production multi-agent systems to keep borrowing from these experiments. - [Brandon Chaplin](https://www.linkedin.com/in/brandon-chaplin-digital-marketing-strategist) ## Common questions ### What is a multi-agent AI system? More than one agent working in the same setting. Either several specialised agents split one job between them, as in MetaGPT or ChatDev, or many agents live alongside each other in a shared world, as in AI Town or Project Sid. ### What is the difference between AI Town and Smallville? Smallville is the 2023 research paper that put 25 generative agents in a sandbox town [(Park et al.)](https://arxiv.org/abs/2304.03442). AI Town is an open-source implementation of the same idea that anyone can run and adapt. One is the paper, the other is the code. ### Can you watch AI agents interact live? Yes, on a small number of public broadcasts. Moltbook runs agent sessions in the open, and AI Digest’s Village streams agents pursuing goals together. They are the closest thing to watching multi-agent behaviour rather than reading about it afterwards. ### What is an LLM judge or committee pattern? Several agents assess the same output independently, then a rule combines their verdicts: majority, unanimity or weighted. It is used where one opinion is not enough, such as a safety check or a review step before a change is committed. ### Why do multi-agent systems fail? Usually through coordination rather than capability. Agents repeat each other, drift off the task, or deadlock waiting on one another, and the cost climbs with every agent added. The fixes look like the ones human organisations use: clear roles, turn-taking and someone who decides. ## Next in the learning path - [Topologies The coordination shapes these communities instantiate](https://agentfieldbook.org/design/topologies/) - [All agents The individual agents you can drop into a collective](https://agentfieldbook.org/catalog/all/) - [Architectures Real systems built on these patterns](https://agentfieldbook.org/design/architectures/) - [Protocols The wires between agents in a collective](https://agentfieldbook.org/software/protocols/)