Skill Detail

Build enterprise-ready JVM agents with JetBrains Koog

Use JetBrains Koog to define typed, fault-tolerant AI agents that run inside JVM, Kotlin, backend, mobile, or browser contexts.

Developer ToolsCustom Agents
Developer Tools Custom Agents Security Reviewed
โญ 4.3k GitHub stars
INSTALL WITH ANY AGENT
npx skills add agentskillexchange/skills --skill build-enterprise-ready-jvm-agents-with-jetbrains-koog Copy
Works best when you want a reusable capability, not another fragile one-off prompt.
At a glance
Tools required
JetBrains Koog, Kotlin or Java runtime, model provider credentials
Install & setup
Follow the Koog repository setup instructions, add the framework to the JVM project, configure model access, define tools and agent behavior, then run the workflow in the target service.
Author
JetBrains
Publisher
Open Source Project
Last updated
Jun 8, 2026
Quick brief

JetBrains Koog is a JVM framework for building predictable, fault-tolerant AI agents. This skill is for teams that need to define a typed agent, connect tools and model calls, add fault-tolerance policies, and run the workflow inside JVM-backed applications or services. Invoke it when the agent must live in a Kotlin or Java production environment rather than as a standalone chat workflow. The boundary is custom JVM agent implementation and operation, not a generic Kotlin library listing.

How it works

What this skill actually does

Inputs and prerequisites: JetBrains Koog, Kotlin or Java runtime, model provider credentials.

Setup notes: Follow the Koog repository setup instructions, add the framework to the JVM project, configure model access, define tools and agent behavior, then run the workflow in the target service.

Source and verification boundary: use https://github.com/JetBrains/koog as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.

Framework fit: publish this as a Custom Agents workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.