Cursor’s latest changelog pushes its agent product further into recurring engineering work. Automations are now available inside the Agents Window, not only at cursor.com/automations, so teams can create and manage scheduled or event-style agent work in the same place they handle normal agent tasks.
The larger change is context. Cursor says automations can now attach multiple repositories, allowing an agent to reason across more than one codebase and work across repos to deliver, test and verify tasks. For agencies and product teams, that is the shape many real jobs take. A small feature may touch a Laravel API, a Next.js frontend, a shared package and a deployment config. A single-repo assistant can help, but it often breaks down at the boundary between systems.
Cursor is also adding no-repo automations. These are aimed at operational tasks where the useful context sits in other tools rather than source code. The company lists new marketplace templates including a Slack digest agent, product analytics agent, product FAQ agent, product finance agent and customer health agent. The product FAQ example watches a Slack channel and drafts a first response using docs, codebase context and past threads; the finance example pulls billing data from a provider such as Stripe for recurring revenue reports.
That points to an important broadening of “coding agent”. Cursor is not only competing to autocomplete code. It is trying to own a daily work surface where agents can perform repeatable software-adjacent jobs: triage, reporting, customer signals, documentation and cross-repo maintenance.
For Alex, the useful test is whether these automations can be made boring. A multi-repo agent is valuable if it can update a shared package and consuming app, run the relevant tests, open a readable PR and stop when permissions or uncertainty require a human. It is much less valuable if it creates a hidden branch of half-working changes across client systems.
The immediate caveat is governance. Multi-repo and no-repo agents need scoped access, logs, approval paths and cost limits. Cursor says newly created automation agent runs are 50% off for the next seven days, which may encourage experimentation, but the real question is what teams are comfortable letting recur every morning or every week.
The direction is clear: agentic IDEs are becoming workflow orchestrators. Cursor’s Automations update is another step from “ask the editor” toward “delegate the loop”.