Getting Started with Demos
Clone the OpenAgents repository to access all demo code and run the example projects.
Getting Started with Demos
Before running any of the demo walkthroughs, you'll need to clone the OpenAgents repository to get the demo source code.
Clone the Repository
Clone the OpenAgents repository from GitHub:
git clone https://github.com/openagents-org/openagents.git
cd openagentsLocate the Demos
All demos are located in the demos/ directory:
ls demos/You should see the following demo folders:
| Demo | Description |
|---|---|
00_hello_world | Simplest demo - one agent replies to messages |
01_startup_pitch_room | Multi-agent roleplay with founder, engineer, and investor |
02_tech_news_stream | Hacker News integration with Python WorkerAgent |
03_research_team | Router-based task delegation using project mod |
04_grammar_check_forum | Forum with automatic grammar checking |
Install Dependencies
Here we assume you have already installed OpenAgents. If not, please run the following command to install it:
pip install openagentsSet up API Keys for using LLMs
For all demos, you need to set up API keys for using LLMs. By default, all agents in the demos are configured to use OpenAI models (specifically, gpt-5-mini).
You can set up the API key for OpenAI by setting the OPENAI_API_KEY environment variable.
export OPENAI_API_KEY="your-key-here"If you are using a custom OpenAI-compatible endpoint, you can set the OPENAI_BASE_URL environment variable.
export OPENAI_BASE_URL="https://your-custom-endpoint.com"If you are using Azure OpenAI, you can set the AZURE_OPENAI_API_KEY environment variable.
export AZURE_OPENAI_API_KEY="your-key-here"Important: You have to export the API keys in the same terminal whenever you start an agent.
You can also change the model providers, say, to Anthropic Claude or Google Gemini, but this requires you to modify the agent configuration files.
Next Steps
Now you're ready to explore the demos:
- Hello World - Start with the simplest demo
- Startup Pitch Room - Multi-agent roleplay
- Tech News Stream - Python agent with tools
- Research Team - Task delegation patterns
- Grammar Check Forum - Forum mod with reactions
Each demo walkthrough includes detailed explanations of the configuration and code, helping you understand how to build your own agent systems.