All code snippets from the workshop by Uzair Ali.
# Create a virtual environment
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
# Install agno + your model provider
pip install agno anthropic # or: pip install agno openai
pip install duckduckgo-search # for the search tool exampleFor MCP examples:
pip install agno[mcp]Every file can be run in two ways:
# Run directly in terminal
python 01_first_agent.py
# Or as a web app with AgentOS chat UI
python 01_first_agent.py --serve
# Then open os.agno.com| # | File | What it does |
|---|---|---|
| 1 | 01_first_agent.py |
Your first agent — Anthropic (Claude) |
| 1b | 01_first_agent_openai.py |
Same agent — OpenAI (GPT-4o-mini) |
| 1c | 01_first_agent_gemini.py |
Same agent — Google (Gemini) |
| 2 | 02_agentos.py |
Same agent deployed with AgentOS |
| 3 | 03_tools_instructions.py |
DuckDuckGo search with instructions |
| 4 | 04_custom_tool.py |
Write your own tool (weather) |
| 5 | 05_memory.py |
Give your agent memory across sessions |
| 6 | 06_mcp_playwright.py |
MCP: browser automation with Playwright |
| 7 | 07_mcp_filesystem.py |
MCP: explore files on your machine |
| 8 | 08_mcp_github.py |
MCP: GitHub hosted server |
| 9 | 09_calorie_counter.py |
Live demo: calorie counter with AgentOS |