python-ai-agent-frameworks-demos

Report Abuse

Basic Information

This repository is a collection of hands-on example projects that demonstrate how to build and run Python AI agents with multiple popular frameworks using GitHub Models and optionally Azure OpenAI. It is targeted at developers who want runnable samples and patterns for single agents, multi-agent workflows, orchestrators and retrieval-augmented generation. The README documents how to run examples in GitHub Codespaces, VS Code Dev Containers, or a local Python 3.10+ environment and describes required environment variables such as GITHUB_TOKEN and optional GITHUB_MODEL. The repo includes an examples directory with scripts that show distinct agent patterns and an infra directory with infrastructure-as-code to provision Azure OpenAI deployments. The content focuses on practical, reproducible demos rather than production services and explains provisioning, authentication, and resource cleanup steps.

Links

App Details

Features
A broad set of example scripts that each demonstrate a different Python agent framework or pattern. Included frameworks and patterns shown in examples are AutoGen (single agent, tools, and orchestrators including MagnetitOne and Swarm), LangGraph state graphs, LlamaIndex for RAG, OpenAI Agents and function calling, PydanticAI multi-agent workflows, Semantic Kernel workflows, and a SmolAgents code agent that can search the web and run code. Cross-environment support is provided via GitHub Codespaces and VS Code Dev Containers plus local setup instructions and a requirements file. The repo provides configuration guidance for GitHub Models using a GITHUB_TOKEN, shows how to set GITHUB_MODEL, and includes Azure IaC to provision gpt-4o and embedding deployments with azd.
Use Cases
It gives developers ready-to-run, side-by-side examples to learn and compare agent frameworks and common patterns such as tool-enabled agents, orchestrators, swarm coordination, sequential two-agent workflows, function calling, and RAG across multiple frameworks. The Codespaces and Dev Container options let users run demos quickly without local setup while the local instructions enable reproducible development. The Azure infrastructure-as-code lets teams test the same examples against Azure OpenAI deployments when needed and shows provisioning and teardown steps. Configuration guidance for GITHUB_TOKEN and optional model selection makes it straightforward to try the examples with GitHub Models subject to rate limits.

Please fill the required fields*