Cloud-hosted AI workflow platform Trigger.dev disclosed Sunday that attackers had gained unauthorized access to environment variables belonging to a subset of its customers. The company, which provides tools for building AI agents and workflows in TypeScript, offered limited details in its security incident report but confirmed that the exploit used to gain entry has been blocked. Affected organizations have been notified and instructed on steps to prevent additional security issues.

The company's investigation identified exposure across a portion of organizations using the platform, though Trigger hasn't publicly named which customers were impacted. Among the platform's clients are high-profile names including fashion brand Emporio Armani, Pentagon-supporting AI acquisitions platform GovSignals, and SaaS calendar platform Cal.com. The compromised environment variables could contain sensitive credentials that effectively hand attackers "the keys to the kingdom," potentially granting access to connected services and digital infrastructure.

According to the security disclosure, Trigger markets itself as a durable execution platform that manages retries, scaling, and sustains long-running tasks through crashes and redeployments while offering monitoring tools. The platform enables companies to "build durable AI agents that survive refreshes, redeploys, and crashes" and handles "long-running tasks with retries, queues, AI observability and elastic scaling." When reached for comment, a company spokesperson declined to share additional information and directed inquiries to wait for a complete incident report.

The breach carries particularly serious implications because of what Trigger's platform does. Companies running AI operations on Trigger likely need continuous uptime, making stolen credentials especially hazardous. The platform's elastic scaling, automated retries, and capacity to run numerous concurrent operations without timeouts could magnify the damage from compromised API keys, depending on what those credentials let an attacker reach. An intruder with stolen environment variables might interfere with a customer's connected services or workflows, and Trigger recently added webhook support for services like Stripe, Supabase, and Slack, potentially widening the range of integrations that could be compromised if their credentials were exposed.

For Trigger customers and those worried about SaaS security as AI routinely creates new vulnerabilities in widely used platforms, one alternative exists: Trigger is open source and can be self-hosted, which may prove a safer option than depending on a hosted service in the current environment. Organizations relying on cloud platforms to run critical AI workflows face a persistent trade-off between operational convenience and control over their security perimeter, a calculation this incident will likely prompt many to revisit.