Cursor Launches Automations for Event-Driven AI Coding Agents

Cursor Launches Automations for Event-Driven AI Coding Agents

Cursor, the AI-powered code editor that has become a dominant force in developer tools, has launched "Automations," a new system that enables event-driven AI coding agents to run autonomously without constant human initiation. The feature debuted on March 5, 2026, representing a significant shift from the traditional "prompt-and-monitor" dynamic that has characterized AI coding assistants.

Unlike conventional AI coding tools that require developers to manually trigger each task, Automations allows engineers to set up triggers based on specific events: code commits, Slack messages, scheduled timers, or incident alerts from systems like PagerDuty. When triggered, AI agents automatically execute tasks such as code reviews, bug detection, security audits, and incident response.

"The system keeps humans in the loop at the right points in this conveyor belt rather than removing them entirely," said Jonas Nelle, Cursor's engineering chief for asynchronous agents, in comments cited by TechCrunch. "We're not trying to replace engineers—we're trying to create a software factory where agents handle the repetitive stuff."

The expanded Bugbot feature now conducts more thorough security audits by allocating additional computational resources to find complex issues. Cursor also demonstrated an incident response automation that integrates with PagerDuty to immediately query server logs through an MCP connection when issues arise.

According to Cursor, the company runs hundreds of automations per hour across various use cases. Early adopters report faster development velocity and improved code quality, with engineering teams able to focus on complex problem-solving rather than routine monitoring tasks.

The launch positions Cursor further ahead in the competitive AI coding market, where the company reportedly generates over $2 billion in annual revenue. Competition continues to intensify from OpenAI and Anthropic, both of which have been expanding their agentic coding capabilities.