Our take
Maintaineda few commits in the last 90 days- Activity
- 28
- Community
- 14
- Issues
- 21
- Pull requests
- 43
+0 stars in the last 30 days
- Fewer contributors than 86% of the projects we track
Agentarium offers a focused starting point for Python-based multi-agent simulations, but its recent activity is light and development is concentrated in one contributor.
Good fit if
- You need a Python framework for coordinating agents that communicate and take context-dependent actions.
- Your project benefits from custom actions, interaction memory, or checkpoint-based state restoration.
- You are comfortable with Python 3.10 or later and the Apache-2.0 license.
Look elsewhere if
- You need a project with frequent recent releases, since there has been no release in over a year.
- Your adoption process requires development activity to be spread across multiple contributors, as most of the code comes from one person.
- You need a simulation engine with features beyond the agent management and interaction capabilities described here.
All health signals
| Last commit | 2026-10-09 (5 days ago) |
|---|---|
| Commits, last 90 days | 1 |
| Releases, last 12 months | 0 (latest v0.3.1, 2025-01-30) |
| Contributors | 2 (top contributor: 96% of commits) |
| Issues closed, last 90 days | 0 |
| Pull requests merged, last 90 days | 1 |
| Project age | 22 months |
Checked on 2026-10-15 with the GitHub API.
Project history
Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.
Stars
935Forks
75Change since Nov 16, 2026: 0 stars·0 forks
Overview
Agentarium is a Python framework for creating and managing AI-powered agents that interact, make decisions, and evolve in simulations. It addresses the work of coordinating multiple agents and their interactions, with AI provider integration through aisuite.
The project is aimed at developers building interactive environments or multi-agent workflows. Agents can communicate directly, choose actions based on context, and retain memories of past interactions. Checkpoints support saving and restoring state for reproducible runs.
Key Features
- Create and coordinate multiple agents with distinct roles and characteristics.
- Let agents choose actions based on their context.
- Define custom actions and add them to agents.
- Maintain memory of past interactions for contextual responses.
- Save and restore agent states and interactions with checkpoints.
- Track interactions through an agent interaction manager.
- Configure an AI provider and model through YAML, using providers supported by aisuite.
Use Cases
- Prototype a multi-agent conversation or collaboration workflow and inspect how agents respond to one another.
- Build a simulation with agents that have distinct roles, context, and custom actions.
- Run repeatable experiments by saving and restoring agent state between steps.
- Extend agent behavior with domain-specific actions when the default talk-and-think capabilities are insufficient.
What you need
Detected in the repository
- Python >=3.10 (from pyproject.toml)
- A test suite and automated checks on GitHub Actions
License in plain words
Apache-2.0permissive
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the license, the NOTICE file and a note of your changes
- Share your changes: no
- Includes an explicit patent grant from the contributors:
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
Install from PyPI:
pip install agentarium
See the README for provider configuration and examples.
Alternatives
- autogen: AutoGen focuses on general-purpose agent orchestration and human-agent collaboration, while Agentarium emphasizes interactive environments, memory, actions, and checkpoints.
- ROMA: ROMA centers on recursive planning for complex tasks, while Agentarium focuses on coordinating agents in interactive environments and simulations.
- OxyGent: OxyGent assembles modular tools, models, and agents with observable execution, while Agentarium emphasizes simulation environments, memory, and checkpoints.
- lagent: Lagent provides synchronous and asynchronous agent workflows with tools and memory, while Agentarium focuses on interactive simulations and checkpointed runs.
| Project | Language | License | Stars | Status |
|---|---|---|---|---|
| Agentarium | Python | Apache-2.0 | 935 | Maintained |
| autogen | Python | CC-BY-4.0 | 61.3k | Quiet |
| ROMA | Python | – | 5.2k | Quiet |
| OxyGent | Python | – | 2.1k | Maintained |
| lagent | Python | Apache-2.0 | 2.3k | Quiet |
Considerations
The project is marked maintained and has had a recent commit, but the measured pace is only a few commits in the last 90 days, with one pull request merged and no issues closed during that period. There has been no release in over a year, and contributor activity is highly concentrated, so teams should assess whether the current maintenance pattern suits their needs. Python 3.10 or later is required. The README does not specify a required GPU or database. The Apache-2.0 license permits broad use, subject to its license terms.
Found this useful?
Share it with someone who would like Agentarium.