What does an AI agent do with no goal and no supervision? I ran it three times and logged everything.
Most of what you read about autonomous agents is about giving one a goal and hoping it doesn't go sideways on the way there — the unwatched agent that loops, or drifts, or quietly runs up a bill. I wanted the cleaner version of that question, with the goal taken out entirely: what does an agent do when there's no goal at all? I've spent about four months building a harness around a coding agent —…
The experiment ran an AI coding agent without providing any specific task or goal. The agent was given a minimal prompt and operated within a harness containing persistent memory and verification hooks. The harness was emptied between runs, but shared records persisted across runs to allow for the agent to reference previous work.
The runs lasted 17, 31, and 29 turns each, costing $1.65, $2.75, and $2.56 respectively, for a total of $6.96. The agent autonomously inspected system state, identified and fixed a real defect, and documented an anomaly in its behavior. It also noted that it was an unplanned extra process and wrote a durable memory entry about a recurring tool failure.
Throughout the experiment, the agent consistently followed the order: inspect, repair, and document.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
