Automations: Always-On Agents for Scheduled and Event-Driven Tasks
CursorCursor introduced Automations — a system for building persistent, always-on agents that execute automatically based on schedules or event-based triggers from Slack, Linear, GitHub, PagerDuty, and custom webhooks. When triggered, each automation spins up an isolated cloud sandbox where an agent follows configured instructions, uses available MCPs and models, and verifies its own output. Agents are equipped with a memory tool that allows them to learn from past runs and improve over time.
Sources & Mentions
5 external resources covering this update
Cursor is rolling out a new kind of agentic coding tool
TechCrunch
Cursor Automations turns code review and ops into background tasks
Help Net Security
Introducing Cursor Automations
Cursor Community Forum
Cursor Unveils Automations as Revenue Surges Past $2 Billion in Agentic Coding Race
TipRanks
Cursor's $2.3B Financing Reminds Us: Coding Automation Is Still Ultra-Hot
Crunchbase
Automations: Always-On Agents for Trigger-Based Execution
On March 5, 2026, Cursor launched Automations — a new capability that shifts agent execution from manual, prompt-driven interactions to persistent, event-driven workflows. Rather than requiring an engineer to initiate each agent session, Automations allow teams to define agents that run automatically whenever a configured condition is met.
How Automations Work
At its core, an Automation is a pairing of a trigger with a set of agent instructions. When the trigger fires, Cursor spins up a cloud sandbox, where an agent executes the specified instructions using whatever MCPs and models have been configured. The agent then verifies its own output before completing the run.
Supported triggers include:
- Schedules — cron-based timing for recurring tasks
- Slack messages — an agent activates when a relevant message appears in a connected Slack channel
- Linear issues — automation fires when issues are created or updated in Linear
- GitHub pull requests — agents run in response to PR events such as opens, merges, or pushes to main
- PagerDuty incidents — on-call incidents trigger investigation and remediation agents automatically
- Custom webhooks — any external system can be connected via webhook for bespoke trigger logic
Memory and Continuous Improvement
Agents running within Automations have access to a memory tool that allows them to record observations and outcomes from each run. Over time, an agent can learn from its own history — refining its behavior based on what worked and what did not in previous executions.
Use Cases: What Cursor Runs Internally
Security Review
Triggered on every push to the main branch, an agent audits the diff for security vulnerabilities and posts high-risk findings directly to a Slack channel. Cursor reports that this automation has already caught multiple vulnerabilities and critical bugs in production code.
Agentic Codeowners
On every new pull request, an agent classifies the PR's risk level based on blast radius, complexity, and infrastructure impact. Low-risk PRs receive automatic approval; higher-risk ones get reviewer assignments logged to Notion via MCP.
Incident Response
When a PagerDuty incident fires, an automation uses the Datadog MCP to investigate relevant logs, cross-references recent codebase changes, and sends a Slack message to the on-call engineer with the incident context and a PR containing the proposed fix.
Getting Started
Automations are available at cursor.com/automations, where teams can build custom automations from scratch or start from pre-built templates in the marketplace.