Cursor AI Streamlines Legacy Code Refactoring

Cursor Cloud agents are designed to operate autonomously on remote machines, allowing developers to initiate refactoring tasks and close their laptops while the agents continue working. This capability is particularly beneficial for long-term projects that require extended processing times without direct user supervision.

These agents ensure the completion of tasks by providing various verification methods, including video recordings, screenshots, and automated test results. This comprehensive verification process allows for reliable and efficient project management, reducing the need for constant monitoring.

By offloading intensive tasks to remote environments, Cursor Cloud agents free up local machine resources and enable developers to manage their time more effectively, knowing that complex refactoring processes are progressing independently.

The Grok-46 model, which was released recently, has been highlighted as a powerful new option for long-running and complex tasks within the Cursor platform. This model is engineered to deliver strong performance, comparable to or exceeding other leading frontier models available in the market.

Its introduction is particularly significant due to its high cost-efficiency, making it an attractive choice for organizations seeking to optimize their refactoring and development budgets. The model’s capabilities are expected to streamline operations and enhance the overall efficiency of AI-assisted coding.

The Cursor Engineer noted, "Grok-46 was actually released, I want to say yesterday. Um, and it was or it is an amazing model." This emphasizes the model's newness and the high regard for its capabilities right from its launch.

A structured 4-step process has been introduced for managing large-scale legacy code migrations and refactoring projects within Cursor. This workflow aims to provide a clear and efficient pathway for developers tackling complex codebases, based on both internal company practices and customer feedback.

The first step involves auditing the codebase using Cursor Canvas and Plan Mode, which helps in diagnosing the current state and strategizing the migration. This initial phase is crucial for understanding the scope and potential challenges of the refactoring effort.

The second step focuses on generating Jira tickets through Cursor's plugin integrations, which automates the creation and organization of tasks based on the defined plan. This streamlines the project management aspect and integrates with existing developer tools.

The third step delegates these tasks to autonomous Cloud agents, allowing them to perform code modifications and testing without continuous human intervention. The final step involves maintaining stability through Cursor automations, ensuring that the refactored code remains robust and functional.

Cursor Canvas provides a built-in skill for auditing and visualizing test coverage across a codebase. By simply using the '/canvas' command, developers can analyze their projects and generate visual representations of testing efforts.

The platform converts raw data into interactive charts or tables, which are highly beneficial for team presentations and understanding complex codebase structures. This feature simplifies the process of assessing where testing is robust and where it might be lacking.

The generated canvas acts as a dynamic, living document that can be shared via a simple link, allowing for real-time collaboration and updates. This ensures that all team members are working with the most current information regarding test coverage and code health.

Cursor's Canvas feature promotes collaborative workflows by allowing real-time updates to shared documents. Once a canvas is created and shared via a link, anyone with access can view and see progress, ensuring that all team members are always on the same page.

The Cursor Engineer described it as, "it's a living, breathing document that is really easy to spin up. Just use /canvas." This ease of use encourages frequent creation and sharing of canvases for various project aspects.

This capability is particularly useful for tracking progress, reviewing architectural changes, or discussing test coverage, as updates are reflected immediately. It reduces communication overhead and ensures that everyone involved has the most current information without needing manual updates or repeated meetings.

When working with Cursor, developers can choose between the traditional Integrated Development Environment (IDE) or the dedicated Agents window, a decision that often comes down to personal preference and workflow. The IDE is favored by those who rely heavily on the VS Code debugger and their existing set of VS Code extensions.

The speaker, however, expressed a preference for the Agents window, explaining, "I like the agents window because I spend most of my time in my agent and then when I do need to review code, I do most of that actually in the cloud." This highlights a workflow optimized for cloud-based task management and code review.

Cursor ensures ongoing support for both environments, allowing users to select the interface that best suits their individual development practices and tool dependencies.

Cursor has expanded its capabilities by integrating a wide range of third-party plugins, making it a more versatile tool for developers. The platform now supports popular tools like Atlassian (Jira), DataDog, Figma, and the entire Google Workspace suite, including Google Drive, Google Calendar, and Gmail.

The speaker confirmed this, stating, "If you guys use Jira, if you use linear, um use monday.com, those all live actually in here now... We now actually have Google Drive, Google Calendar, um and your Gmail all as plugins within Cursor." This broad integration aims to create a seamless development environment.

These plugins enable more efficient workflows, such as publishing development plans directly to Google Docs, managing calendar schedules, or tracking project progress within Jira, all from the Cursor interface. This reduces the need to switch between multiple applications, streamlining the overall development process.

Cursor's Multitask mode significantly enhances productivity by allowing agents to work on multiple, unrelated tasks in parallel. This feature enables developers to expedite workflows that traditionally involve sequential operations.

For instance, an agent can simultaneously shorten an executive summary for a report while dividing a comprehensive plan into individual Jira tickets. This demonstrates the system's ability to manage diverse tasks concurrently without requiring manual intervention.

The speaker elaborated, "I'm actually going to show you a mode called multitask mode where it can actually work on both of these things in parallel. As you can see, these two agents aren't really that related. One of them is again shortening the executive summary. The other one is dividing the planet of Jira tickets. So it can work on both of those things at once." This capability is designed to dramatically cut down on project completion times by optimizing agent utilization.

Cursor's long-running agents are designed to handle persistent background tasks, such as dependency or version updates, that may require extended periods to complete. These agents function similarly to a 'slashgoal' command, continuing their work until the task is fully accomplished.

Users have the flexibility to set time limits for these agents, allowing them to define when a task should be considered done or how long an agent should continue working. This provides control over resource allocation and project timelines.

The Cursor Engineer explained, "It's basically similar to slashgoal where it will keep working on a task until it's finished. Um if you ever want to turn it on and just use it, you can actually set like time limits as to when it should be, you know, done or when it should work until." A critical aspect of these agents is their commitment to full testing before finalizing any task, ensuring stability and reliability of the code.

Cursor Cloud agents leverage remote Linux Virtual Machines (VMs) to perform comprehensive self-verification of code changes. These VMs provide full desktop and mouse access, enabling the agents to interact with applications just like a human user would.

This capability allows Cursor to conduct autonomous UI testing and screen recording, making it possible to thoroughly test full-stack and front-end applications. The agents can simulate user interactions, observe application behavior, and record any issues or successful operations.

The Cursor Engineer stated, "Cursor cloud agents run on remote VMs, Linux VMs. And so, because of that, they have access to a computer and mouse. So, if you have any full stack front-end applications that need testing, cursor can actually do that testing for you." This self-verification process significantly reduces regression risks by identifying potential problems before changes are reviewed by human developers.

The Cursor app, available on app stores, extends the functionality of the platform to mobile devices, allowing developers to manage their workflows on the go. This app enables users to launch and monitor their agents remotely, providing flexibility and convenience.

From the mobile interface, users can access their workspaces, review the history of agent activities, and monitor the real-time execution status of ongoing tasks. This ensures that developers can stay informed and in control of their projects even when away from their primary workstations.

The Cursor Engineer demonstrated, "The cursor app again is available on the app store, but this is basically what it looks like. So you can see here I have one agent working. That's the Jira tickets agent that we just launched." This mobile accessibility is critical for modern development teams requiring constant connectivity and oversight.

Cursor's automation capabilities extend to incident management, significantly reducing the burden on on-call personnel. Instead of immediately escalating every issue, agents can perform an initial investigation to triage incidents.

The speaker explained, "Instead of waking up someone at 3:00 a.m., um, it will do an initial kind of investigation and then send the person on call like a brief summary of what it's found already." This means that critical alerts often accompanied by disruptive late-night wake-up calls can be preceded by a concise summary of the agent's findings, allowing on-call staff to respond more efficiently and with greater context.

Despite their autonomous capabilities, Cursor Cloud agents are designed with a crucial 'human-in-the-loop' requirement, ensuring that user input and oversight remain an integral part of the development process. This design allows for a balance between automation and human control.

The speaker clarified, "cloud agents are meant to be autonomous, but they are also meant to have user input. So, if you ever want to add a follow-up, if you ever want to interfere with a cloud agent or if you don't agree with what it's doing, you can always add that here." This means users can provide follow-up feedback, correct agent actions, or manually interrupt processes if they disagree with an agent's approach, even from mobile apps.

This interactive capability ensures that while agents can handle routine and complex tasks independently, developers retain the ultimate authority to guide and refine their work, preventing unintended outcomes and maintaining quality control.

Security is a critical aspect of the Cursor Cloud environment, with robust mechanisms for managing sensitive information. Environment keys, API keys, and other secrets are securely stored and managed within the Cursor Cloud environment setup.

The Cursor Engineer confirmed, "So environment keys, API keys, secrets, all of those things live in your cursor cloud environment setup." This centralized management ensures that sensitive credentials are not exposed in code or insecure locations.

For enhanced control and access management, IP allow-listing is available, restricting access to authorized IP addresses. Additionally, Cursor provides comprehensive enterprise documentation detailing options for private connectivity and advanced security configurations, catering to organizations with stringent security requirements.

Cursor offers multiple private connectivity options for Cloud Agents to ensure secure and isolated communication with customer infrastructures. These solutions are tailored to various network setups and preferences.

For users not on AWS, Cloudflare tunnels are available to establish secure connections. If an organization operates within AWS, AWS PrivateLink provides a dedicated and private network connection. The Cursor Engineer mentioned, "Options that we offer are Cloudflare tunnels if you're not on AWS AWS private link if you are on AWS. And then sometimes like VPN work so if you guys are on tail scale sometimes we offer that as well."

Additionally, VPN solutions like Tailscale are supported for certain configurations. While self-hosted Cloud Agents are technically an option, Cursor generally does not recommend them as the primary solution, preferring its managed cloud-based services for optimal performance and security.

An upcoming workshop is scheduled for next Thursday, focusing on the integration of cloud and local environments within Cursor's ecosystem. The workshop aims to provide practical solutions for developers looking to combine the strengths of both setups to enhance their productivity.