Cursor Automations: Always-On AI Agents for Your Codebase

Cursor launched Automations on March 5, 2026 — a system that lets AI coding agents run independently based on triggers: a GitHub pull request opens, a PagerDuty alert fires, a Slack message arrives, a timer hits. You do not need to be watching.

What Are Cursor Automations?

Automations is a trigger-based orchestration layer for AI coding agents built into Cursor. When a configured event fires, Cursor spins up an isolated environment, executes the configured agent instructions using your connected MCP tools, verifies its own work, and reports results. The agent can also draw on memory from prior runs.

Three categories of triggers are available at launch:

• Code events: new PR opened, push to branch, test failure
• External signals: Slack message, Linear issue created, PagerDuty incident alert
• Scheduled timers: run every hour, every night, every Monday morning

Each automation runs sandboxed. It cannot affect your production environment unless you explicitly configure write access.

How This Differs From What Cursor Could Do Before

Until now, Cursor operated in a prompt-and-monitor dynamic: you write a prompt, an agent runs, you watch it, you approve or redirect. Cloud Agents already removed the monitoring step for discrete tasks. Automations takes that further: you do not even initiate the task.

Automations inverts that. The agent monitors. You review outputs.

Bugbot, an early Automation built on this system, activates on every pull request opening. It scans for stylistic inconsistencies, security vulnerabilities, and performance problems, then posts findings as code review comments. The developer reads a summary rather than writing one. That is a meaningful shift in how time gets spent.

The Revenue Signal Behind This

Cursor crossed $2 billion in annual revenue, doubling in the past three months according to Bloomberg. That growth rate is worth understanding in context. GitHub Copilot took roughly two years to reach $100 million ARR. Cursor hit $2 billion in annual revenue in about 18 months.

The market is not just paying for autocomplete. It is paying for a system that reduces total cognitive load across the development workflow, from writing code to reviewing it to responding to incidents.

What to Actually Try First

Automations is available now for Cursor team plans. Practical starting points:

Set up Bugbot on your main branch PR trigger first. It's the lowest-risk automation because it only adds comments, it does not change code. See how accurate the findings are for your codebase before trusting more write-heavy agents.

Try a PagerDuty automation that queries server logs through an MCP connection and posts an incident summary to Slack when an alert fires. This one saves real time at 2am.

Schedule a nightly dependency audit that opens a PR when package versions fall behind. Boring, but it tends to be the thing that never gets done manually.

Start read-only before you give agents write access to production branches. The risk with agentic tools is scope — agents that do too much without human checkpoints are harder to debug when something goes wrong.

Key Takeaways

• $2 billion annual revenue — Cursor doubled revenue in the 3 months before this launch
• March 5, 2026 — Automations launch date
• Supported triggers: GitHub PRs, Slack, Linear issues, PagerDuty alerts, custom timers
• Each automation runs in an isolated sandboxed environment with configurable MCP tool access
• For developers: Start with Bugbot on PR triggers; read-only automations first, write access second — build trust incrementally
• What to watch: Whether OpenAI Codex or GitHub Copilot Workspace ships comparable trigger-based orchestration by mid-2026, and how Cursor responds