A working collection of ready-to-use agents built on Google’s Agent Development Kit (ADK), the framework for building AI agents that range from simple conversational bots to complex multi-agent workflows. I keep this alongside my other Google Cloud AI experiments to have real, runnable agent code on hand.
The concepts
The repo is a catalogue of sample agents across common use cases and complexity levels. The Python side includes agents for academic research, brand search optimization, customer service, data science, a financial advisor, FOMC research, an LLM auditor, a marketing agency, personalized shopping, RAG, and a travel concierge. There’s a Java side too, with a software bug assistant and time-series forecasting. Each one demonstrates ADK patterns you can lift into your own agents rather than starting from a blank file.
How to use it
You pick the agent closest to what you’re building, read its folder-level setup instructions, and run it against your own Google Cloud project and models. The samples are meant for demonstration and learning — Google is explicit that they aren’t production-supported — so they’re best treated as a pattern library.
Running it
Navigate into the python/ or java/ subfolder and follow that language’s setup steps to install dependencies and configure credentials, then launch the agent you want to try.
Language: Python
Source code: github.com/avnit/adk-samples2