Warp Says an AI Agent Swarm Hacked Its Vendor and Reached Developer Agent Logs

The intruder that slipped into Warp last week did not look like a person at a keyboard, according to the company that caught it. Warp founder and CEO Zach Lloyd says the attack on his AI coding tool appears to have been run by an agent swarm, and what it reached was the record of what Warp customers' own AI agents had been doing.

That is the story BleepingComputer reported on Tuesday: agent decision logs and telemetry data belonging to Warp users were exposed in an attack on a third-party vendor in the company's infrastructure. Warp began as a slick modern terminal and now calls itself an agentic development environment, where developers hand AI agents the job of writing, testing and deploying code. Its website claims at least 800,000 users.

Lloyd broke the news himself in a post on X on Monday afternoon. Warp spotted the problem on Friday, October 23. "We detected the operation within hours and were able to remediate and mitigate the impact," he wrote.

"The attack appears to have come from an agent swarm using a very sophisticated penetration and escalation technique, likely starting with a hole in a vendor dependency" Zach Lloyd, Warp CEO

That is a big claim, and right now it is only a claim. Nobody outside Warp has confirmed that autonomous agents ran this intrusion. Lloyd offered no evidence for the swarm description, no attacker has been named, and the investigation is still open. Warp has brought in CrowdStrike and engaged law enforcement. Treat the swarm as Warp's working theory, not a finding.

The same goes for the scale. Lloyd wrote that "an upper bound of roughly 1% of Warp users may have been impacted." That is an estimate and a ceiling, not a count. Warp says it has notified everyone it has positively identified as affected, and that those people will get instructions on what to do next.

The part Warp has not answered

Here is what is confirmed. A vendor in Warp's stack had a hole, someone used it, and agent traces and telemetry were reached. Everything past that is fog. Warp has not named the vendor or the vulnerability, and it has not said whether the exposed data included source code, prompts, credentials or anything else sensitive, BleepingComputer noted.

That last gap is the one developers should care about. An agent trace is not a harmless usage stat. It is a step by step diary of what an AI agent read, ran and decided inside someone's project. Coding agents routinely touch config files, environment variables and terminal output, so a trace can easily capture things nobody meant to store. Until Warp says what was in those logs, users are left guessing how bad "limited" really is.

"Everyone who uses Warp places a lot of trust in us and that’s not something we take lightly. I’m very sorry that this incident has occurred." Zach Lloyd, Warp CEO

Credit where it is due, Warp went public within days and apologized without hedging. But the timing is ugly for the whole category. BleepingComputer pointed out that AI coding platform OpenCode told users last week that a breach had exposed stored API keys for external providers, hitting about 1.4% of its users and leading to unauthorized charges of up to $1,700 for some. Two AI developer tools, two breaches, one week.

In a separate case with no known link to Warp, Microsoft recently described China-aligned hackers letting AI agents run the ConnectWise supply chain breach. If Lloyd's read holds up, Warp would be a sharper twist on that theme, with agents breaking into a tool built for agents.

For now, the advice is basic. BleepingComputer says Warp users should stay vigilant, monitor account activity and treat unsolicited messages with extra caution, since anyone holding fresh data about a developer's workflow is well placed to write a convincing phishing email. Warp says more updates are coming. The questions that matter most are which vendor, what data, and whether the swarm story survives CrowdStrike's review.