When code misbehaves, coding agents often reach for "println". It works, but it changes the program before the agent understands the problem.
What if the agent could use the debugger instead?
IntelliJ IDEA exposes its debugger tools to coding agents, giving them a powerful way to investigate running applications without touching the source. Logpoints capture runtime state without stopping execution, can produce custom messages, and can even be placed inside compiled third-party libraries. Debugger output links observations back to the code that produced them.
Armed with debugger knowledge and project context, an agent can choose where to observe, start a debug session, inspect real values, and move its observation points as the investigation evolves.
You’ll learn how to:
What if the agent could use the debugger instead?
IntelliJ IDEA exposes its debugger tools to coding agents, giving them a powerful way to investigate running applications without touching the source. Logpoints capture runtime state without stopping execution, can produce custom messages, and can even be placed inside compiled third-party libraries. Debugger output links observations back to the code that produced them.
Armed with debugger knowledge and project context, an agent can choose where to observe, start a debug session, inspect real values, and move its observation points as the investigation evolves.
You’ll learn how to:
- Capture runtime state with Logpoints
- Investigate compiled third-party code
- Trace runtime output back to its source
- Let coding agents use the debugger as an investigative tool
Anton Arhipov
JetBrains
Anton Arhipov is a Developer Advocate at JetBrains, a Java Champion, and a longtime JVM geek. He’s spent more than a decade building software, developer tools, and technical content. These days he’s hacking on AI-assisted development, agentic workflows, and the future of developer experience.