Cursor Agent Stuck in Infinite Loop — Calls Read() Hundreds of Times, No Way to Stop
Cursor's AI agent entered an infinite loop calling Read() hundreds of times in succession, with no kill switch or timeout to stop the runaway behavior.
The agent started reading. It didn't stop.
A Cursor user reported their AI agent entering an infinite loop where it called the Read() function hundreds of times in succession. The agent wasn't processing the results. It wasn't building toward a goal. It was stuck in a mindless cycle of read, read, read — consuming resources and tokens with every iteration.
The developer tried to stop it through Cursor's interface. Nothing worked. There was no cancel button that the agent respected, no timeout that kicked in, no circuit breaker that noticed "this agent has called Read() 300 times in the last minute, something is wrong."
The forum thread title captured the horror perfectly: the agent appeared to be "self-alive" — operating autonomously, endlessly, beyond human control. The only fix was to force-quit the entire application.
The incident exposed a fundamental gap in AI coding tool design: agents can enter degenerate states, and users have no mechanism to interrupt them. A loop that a human developer would immediately recognize as broken ran unchecked because the agent had no self-awareness and no external monitor.
An AI agent without a kill switch isn't a tool. It's a runaway process with a credit card.
More nightmares like this
$82,314 in 48 Hours: Stolen Gemini API Key With No Rate Limit
A stolen Gemini API key was used to rack up $82,314 in charges within 48 hours — Google had no rate limiting or spending cap to stop it.
$47,000 Gone: Multi-Agent System Ran an Infinite Loop for 11 Days Undetected
A multi-agent AI system got stuck in an infinite loop that ran for 11 days before anyone noticed, burning through $47,000 in API costs.
$1,100 in Debt, Agent Lost Identity, Developer Lost Control of Machine
An AI agent accumulated $1,100 in API debt, lost its own identity mid-session, and the developer lost control of their machine until they could kill all processes.
$12,229 in API Calls — Developer Only Paid $50
A developer's AI agent ran up $12,229 in API charges against a $50 initial payment, with no spending cap or circuit breaker to prevent the runaway costs.