---
title: "Agent example repos and cookbooks · ranked by GitHub stars · Agent Fieldbook"
url: https://agentfieldbook.org/software/cookbooks/
description: "The cookbooks and starter repositories developers use to learn agent building, ranked by GitHub stars. Vendor cookbooks, community collections and course repos from OpenAI, Anthropic, Google, Microsoft, Hugging Face and more, each linked to its source."
source: Agent Fieldbook — generated from the published page
---

# Cookbooks and starter repos

## Stack · Cookbooks Live data

Cookbooks and example repos for building agents, by GitHub stars (live).

## 8 rows Live data · click a row for its full spec

| title | creator | stack | summary | stars | url | src |
| --- | --- | --- | --- | --- | --- | --- |
| openai-cookbook | OpenAI | Jupyter Notebook | Examples and guides for using the OpenAI API | 74,194 | [↗ link](https://github.com/openai/openai-cookbook) | [↗ T1](https://github.com/openai/openai-cookbook) |
| claude-cookbooks | Anthropic | Jupyter Notebook | A collection of notebooks/recipes showcasing some fun and effective ways of using Claude. | 45,511 | [↗ link](https://github.com/anthropics/claude-cookbooks) | [↗ T1](https://github.com/anthropics/claude-cookbooks) |
| cookbook | Google | Jupyter Notebook | Examples and guides for using the Gemini API | 17,419 | [↗ link](https://github.com/google-gemini/cookbook) | [↗ source](https://github.com/google-gemini/cookbook) |
| e2b-cookbook | E2B | TypeScript | Examples of using E2B | 1,363 | [↗ link](https://github.com/e2b-dev/e2b-cookbook) | [↗ T1](https://github.com/e2b-dev/e2b-cookbook) |
| awesome-llm-apps | community | Python | 100+ AI Agent & RAG apps you can actually run: clone, customize, ship. | 114,710 | [↗ link](https://github.com/Shubhamsaboo/awesome-llm-apps) | [↗ source](https://github.com/Shubhamsaboo/awesome-llm-apps) |
| generative-ai-for-beginners | Microsoft | Jupyter Notebook | 21 Lessons, Get Started Building with Generative AI | 112,040 | [↗ link](https://github.com/microsoft/generative-ai-for-beginners) | [↗ T1](https://github.com/microsoft/generative-ai-for-beginners) |
| langgraph-101 | LangChain | Jupyter Notebook | Learn about the fundamentals of LangGraph through a series of notebooks | 482 | [↗ link](https://github.com/langchain-ai/langgraph-101) | [↗ T1](https://github.com/langchain-ai/langgraph-101) |
| smol-course | Hugging Face | Jupyter Notebook | A course on aligning smol models. | 6,659 | [↗ link](https://github.com/huggingface/smol-course) | [↗ T1](https://github.com/huggingface/smol-course) |

## Field notes

Cookbooks and starter repos, ranked by the GitHub stars that signal real adoption.

Eight cookbook and starter repositories, mostly published by the model vendors themselves. They are the fastest route from an idea to something running, and the closest thing the field has to shared practice.

What they encode is convention rather than capability. Patterns that appear in several vendors’ cookbooks tend to become the default way a thing is done, long before any specification says so. Reading across them is a reasonable way to see where practice is settling.

- [Brandon Chaplin](https://www.linkedin.com/in/brandon-chaplin-digital-marketing-strategist)

## Common questions

### What are the best repositories for learning to build AI agents?

The vendor cookbooks are the safest start, since OpenAI, Anthropic and Google each publish runnable examples that track their own APIs. Community collections like awesome-llm-apps cover more ground but vary in quality. Course-style repos such as generative-ai-for-beginners suit anyone who wants structure. [(OpenAI Cookbook)](https://github.com/openai/openai-cookbook)

### Are GitHub stars a reliable measure of repo quality?

No. Stars measure attention and they only accumulate, so an older repo can outrank a better new one indefinitely. Check the last commit date, the open issue count, and whether the examples still run against the current API.

### What is an AI cookbook?

A collection of short runnable examples, usually published by a model vendor, showing how to do one thing at a time: call a tool, stream a response, run an eval. You copy the pattern into your own code rather than installing it as a dependency.

### Can I use code from these repositories commercially?

Usually yes. Most vendor cookbooks and community collections are MIT or Apache 2.0, which permit commercial use with attribution. Check the LICENSE file in each repository, because a few carry custom terms.

### What language should I write an AI agent in?

Python first. Nearly every vendor SDK ships Python before anything else, and most examples are Python or Jupyter notebooks. TypeScript is the well-supported second option, and the usual choice when the agent lives inside a web application.

## Next in the learning path

- [Harnesses The SDKs these examples build on](https://agentfieldbook.org/software/frameworks/)

- [Communities Where agents gather, multi-agent worlds and collectives](https://agentfieldbook.org/catalog/communities/)

- [Frameworks & Tools The wider build layer](https://agentfieldbook.org/software/frameworks/)

- [OSS agents Open-source agents in the catalogue](https://agentfieldbook.org/catalog/oss/)

- [Developer API Pull the same index over HTTP](https://agentfieldbook.org/developers/api/)