Definition
Plain language
A debugging system that lets an AI agent watch what a program actually does, instead of just reading the source code.
As stated in the literature
An agentic debugging framework using Frame Lifetime Traces, on-demand instrumented re-execution, and reformatted hierarchical reports to give LLMs runtime visibility into program behavior.
Why it matters: It closes the gap between how a human debugger works — by observing what code actually does — and how LLMs typically debug from text alone.
For example, instead of guessing why a test failed from the stack trace, DAIRA reruns the program with extra instrumentation around the suspect function and feeds the resulting trace to the LLM.
Heard on the show
“… Now — the system this paper introduces is called DAIRA, and the reason that gap between agent and human engineer matters is that DAIRA closes it in a very …”Episode 012 — Why AI Coding Agents Keep Trying to Debug Without a Debugger