Repolode
Submit
Thytu

Thytu/Agentarium

Python framework for creating simulations populated with AI agents that talk, act, remember and take custom actions in shared environments.

View on GitHub
Stars
934
Forks
75
Open issues
1
Last commit
29 days ago

What is it?

Agentarium is a Python framework for building and managing multi-agent simulations. Each agent is created with a name and optional traits such as occupation, age, location and bio, and agents can send messages to one another, decide their next action from their current context, and keep memory of past interactions.

Beyond the built-in talk and think behaviors, developers can register custom actions with a name, description, parameters and a Python function, then have agents execute them. A CheckpointManager saves and restores agent states and interactions, which supports repeatable runs of a simulation. LLM access goes through aisuite, so the provider and model are set in a YAML configuration file.

The package is published on PyPI and requires Python 3.10 or later. Example projects are included in the examples directory.

Key features

  • 01Create agents with names, occupations and other personal attributes.
  • 02Agents talk to each other and autonomously choose their next action.
  • 03Register custom actions with parameters and Python functions.
  • 04Save and restore agent states and interactions with CheckpointManager.
  • 05Connect to multiple LLM providers through aisuite configured via YAML.

Use cases

  • Prototyping simulations where multiple LLM-driven characters interact.
  • Studying agent behavior that depends on memory and past interactions.
  • Running repeatable multi-agent experiments with saved checkpoints.
  • Adding domain-specific actions to LLM agents in a Python codebase.

Quick start

  1. 1

    Install the package from PyPI

    pip install agentarium
  2. 2

    Configure an LLM provider using the example config file

  3. 3

    Create agents and make them talk and act

Things to know

  • Requires configuring an LLM provider with credentials, such as an API key, before agents can run.
  • Relies on aisuite for provider integration, adding an external dependency.

In collections

Open-source AI agent frameworks & appsRobotics & embodied AI projects

Topics

#multi-agent#llm#simulation#python#agent-framework#orchestration

FAQ

What is Agentarium?
Python framework for creating simulations populated with AI agents that talk, act, remember and take custom actions in shared environments. Agentarium is a Python framework for building and managing multi-agent simulations. Each agent is created with a name and optional traits such as occupation, age, location and bio, and agents can send messages to one another, decide their next action from their current context, and keep memory of past interactions.
Is Agentarium free to use? What license does it have?
Agentarium is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is Agentarium still maintained?
Actively maintained. The last commit to Agentarium was on Oct 9, 2026.
What can I use Agentarium for?
Prototyping simulations where multiple LLM-driven characters interact. Studying agent behavior that depends on memory and past interactions. Running repeatable multi-agent experiments with saved checkpoints. Adding domain-specific actions to LLM agents in a Python codebase.
How do I install Agentarium?
Install the package from PyPI: pip install agentarium Configure an LLM provider using the example config file Create agents and make them talk and act
Where can I get Agentarium?
The source code and installation instructions are on GitHub: https://github.com/Thytu/Agentarium

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects