Cursor 3.5 Lets one AI Agent Work Across Multiple Codebases at Once
Cursor 3.5 lets agents work across multiple repos simultaneously and introduces repo-free automations for monitoring Slack, Stripe, and more
Takeaways
- Multi-repo automations: Cursor 3.5 lets a single agent reason and make changes across multiple repositories in one run.
- No-repo automations: New automation type that monitors external tools (Slack, Stripe, Databricks) without any attached codebase.
- Automations in the IDE: Automations are now manageable directly inside the Agents Window, not just at cursor.com/automations.
- 5 new marketplace templates: Slack digest, product analytics, FAQ agent, finance agent, and customer health agent are ready to use.
- 50% launch discount: All agent runs for newly created automations are 50% off for 7 days after creation.
- Pricing: Individual Pro starts at $20/mo; cloud agents and automations with team context require the Teams plan at $40/user/mo.
Cursor just shipped version 3.5, and the headline feature is deceptively simple: you can now attach more than one repository to an automation. That might sound like a minor config option, but for anyone running microservices or a monorepo-adjacent architecture, it fundamentally changes what an AI agent can actually do for you.
One agent, many codebases
A lot of engineering work spans more than one codebase. You can now attach multiple repos to an automation so agents reason across all required context and work across repos to deliver, test, and verify tasks. Before this, the workflow was painful: automations operated on one repo, so a change that needed coordination across a shared library and its consumers required multiple separate automations or manual orchestration.
Now it's one automation, one agent, multiple repos. In practice, that means an agent can find a type mismatch in a shared types library, update the export, propagate the fix to the frontend interface, and update the corresponding backend model -- all in a single run. With multi-repo, it can find the mismatch in the types library, update the export, update the consuming interface in the frontend, and update the corresponding Pydantic model in the backend -- one session, three coordinated PRs.
Automations without a repo
Many useful automations exist apart from code, where agents monitor your tools and act on key signals. You can now create automations without an attached repository. This is a genuinely new category of agent task -- think scheduled jobs that watch your business tooling rather than your codebase.
Automations are now in the Agents Window, and you can configure them with multiple attached repos or no repos at all. Cursor shipped five ready-to-use templates for the no-repo case in the Cursor Marketplace:
- Slack digest agent -- summarizes unread DMs and key channels every morning, ranked by importance
- Product analytics agent -- delivers a weekly digest of key metrics from a data warehouse like Databricks
- Product FAQ agent -- watches a Slack channel for questions and drafts a first response using docs, codebase context, and past threads
- Product finance agent -- pulls recurring revenue data from billing providers like Stripe
- Customer health agent -- monitors signals across Granola, Slack, and Databricks and flags accounts where health is shifting
Automations move into the IDE
Cursor Automations previously required going to cursor.com/automations to manage them. They are now also accessible inside the Agents Window in the IDE directly. Both access paths remain; this is about friction reduction. The practical effect: you no longer context-switch to a browser tab to set up or tweak a scheduled agent run.
This is part of a broader pattern. This is where "IDE as control plane" stops being a metaphor and becomes literal. Cursor is bidding to be the interface through which a development team manages not just code, but the operational signals around it.
What it's good at -- and where it gets expensive
Multi-repo automations shine in a few specific situations:
- Microservice architectures where a single feature touches multiple services
- Monorepo splits where a shared library has many consumers
- Cross-team PRs that need coordinated changes in frontend, backend, and infra repos
- Scheduled security scans or dependency audits that span multiple services
The tradeoff is cost. Credits go fast on heavy agent tasks. A complex agent run on a large codebase can consume a meaningful chunk of a $20 Pro credit pool. If you're planning to run multiple agents daily, Pro+ at $60 is the practical minimum. Multi-repo runs, by definition, are heavier than single-repo ones -- you're feeding the agent more context and asking it to produce more coordinated output.
Pricing and availability
Cursor 3.5 is available now via cursor.com/download. For the next 7 days, all agent runs for newly created automations are 50% off. Cloud agents and automations are included on the Teams plan at $40/user/month and above. Individual Pro starts at $20/month but automations with shared team context are a Teams-tier feature. Enterprise plans add pooled usage, audit logs, and repository-level access controls.
The no-repo automation templates are free to browse in the Marketplace, and the official docs cover how to wire up triggers, attach repos, and configure agent behavior. If you're running a microservices stack and have been manually coordinating cross-repo changes, this is the update worth testing first.