Skip to content

Latest commit

 

History

History

kotlin-agent-with-koog

README.md

Building AI Agents in Kotlin with Koog

Build your first AI agent in Kotlin using JetBrains' Koog framework, progressing from a basic prompt-response agent through tool calling and structured output.

What You'll Learn

  • Koog Fundamentals: Set up a Kotlin project with Koog and create a minimal AI agent that sends prompts to an LLM
  • Tool Calling: Define custom tools that the agent can invoke autonomously using the ReAct pattern (Reason, Act, Observe)
  • Structured Output: Get typed Kotlin data classes back from the LLM instead of free-form text
  • Kotlin-Native AI: Leverage Kotlin coroutines, serialization, and type safety for AI agent development

Quick Start

git clone https://github.com/NirDiamant/agents-towards-production.git
cd agents-towards-production/tutorials/kotlin-agent-with-koog

export OPENAI_API_KEY="sk-..."

./run.sh   # Interactive guided runner (recommended for first time)

Or run each step directly:

./gradlew step1   # Basic agent
./gradlew step2   # Agent with tools
./gradlew step3   # Structured output

The Gradle wrapper is included -- no need to install Gradle separately. The first run downloads dependencies automatically (takes about a minute).

Tutorial

Full Tutorial Walkthrough

Requirements

  • JDK 17+ installed (Adoptium or brew install openjdk@17)
  • OpenAI API Key (with billing enabled)

What You'll Build

A smart assistant that answers questions using custom tools (weather lookup, calculations, fact retrieval) and returns typed, structured responses as Kotlin data classes.

Total Tutorial Time: ~25-30 minutes

Difficulty: Beginner-Intermediate (Kotlin basics, Gradle)