101 Cursor AI

Learning to use for Java Enterprise projects

v0.3.0-SNAPSHOT (60min talk)

Version control

• V3: W-JAX (2025/11/06)
• V2: Devoxx BE (2025/10/07)
• V1: Madrid JUG (2025/05/06)

https://cursor.com/blog/java

"There is nothing permanent except change."
- Heraclitus

"If you can't describe what you are doing as a process, you don't know what you're doing."
- W. Edwards Deming

Who I am

	Juan Antonio Breña Moral Software Engineering Manager @ Capital One, C1 Twitter | Github | LinkedIn
https://github.com/jabrena/pml https://github.com/jabrena/cursor-rules-java https://github.com/jabrena/churrera

Acknowledgements

Many thanks for the feedback and suggestions:

• David Gómez García
• François Poirier Troyano
• José Ignacio Dominguez Montoya

Poll

Can you raise your hand ✋, if you use Intellij Idea Ultimate?

Poll

Can you raise your hand ✋, if you use VSCode?

Poll

Can you raise your hand ✋, if you use Eclipse?

Poll

Can you raise your hand ✋, if you have tried Cursor AI?

Poll

Can you raise your hand ✋, if you use Cursor AI/CLI in combination with another IDE?

Poll

Can you raise your hand ✋, if you have tried Claude Code?

Poll

Can you raise your hand ✋, if you have tried Github Copilot?

Agenda (60 min)

• Who I am
• Context
• SLDC
• Cursor platform
• Data privacy support
• Demos
• Takeaways
• Q&A

Agenda (60 min)

• Who I am
• Context (5min)
• SLDC (5 min)
• Cursor platform (10min)
• Data privacy support (5min)
• Demos (25min)
• Takeaways (10min)
• Q&A

Context

https://www.cursor.com/

Context

Cursor version used for the talk:

Context

Why Cursor is changing the way of working?

ThoughtWorks Radar #32: 2025Q1 / Tools

ThoughtWorks Radar #33: 2025Q4 / Techniques

SLDC

SLDC (Software Lifecycle Development Cycle)

SLDC

How Cursor can help in each stage of the SLDC?

SLDC

Source: https://cursor.com/docs

SLDC

Design > Coding / Testing > Operations

Let`s put focus in the Coding/Testing stage where Cursor can help in the most cases.

🔍

Cursor platform

Cursor little by little is increasing the opportunities to use in Organizations with features for:

• For Software Engineers
• For DevOps pipelines

Cursor platform

For Software Engineers:

• Cursor AI Desktop + Tab model + Mode Plan
• Intellij Idea Ultimate + Cursor Agent CLI
• VSCode + Cursor Agent CLI
• Github Codespaces + Cursor Agent CLI
• Cursor Web
• Cursor BugBot
• Cursor Rules
• Cursor Composer 1

Cursor platform

For DevOps pipelines:

• Cursor Cloud Agents API
• Cursor Rules
• Cursor Composer 1
• Cursor BugBot

Cursor platform

For Software Engineers

Cursor platform

For Software Engineers: AS IS

Cursor platform

For Software Engineers: TO BE

Cursor platform

For Software Engineers: TO BE

Type in the terminal:

							
cursor-agent
							
						

Cursor platform

For Software Engineers: TO BE

Type in the terminal:

                    							
                    curl https://cursor.com/install -fsS | bash

                    cursor-agent \
                    -p "analyze the latest threaddump and the flamegraph \
                    to provide some insights to fix this memory leak" \
                    --print
                    							
                    						

Cursor platform

For Software Engineers: TO BE

It is common to use Wider screen to handle your favourite Java IDE + Cursor Agent CLI.

Cursor platform

For Software Engineers: TO BE

You can submit tasks from your Smartphone using Cursor Web associated to one Github repository.

Cursor platform

For Software Engineers

Cursor Composer 1, is the new agent model designed for software engineering intelligence and speed.

Cursor platform

For DevOps pipelines

Cursor platform

For DevOps pipelines: AS IS

Cursor platform

For DevOps pipelines: TO BE

Data privacy support

• Security features
  • Privacy mode
  • Trust

Source: https://cursor.com/docs/agent/security

Privacy mode

Note: with a Pro account, the developer needs to disable. But with a Business account, this feature is handle at the organization level.

Privacy mode

Trust

Cursor operates a continuously monitored and 3rd-party audited security program.

SOC 2 is a compliance framework used to evaluate and validate an organizations information security practices.

Trust

Subprocessors are third-party services that Cursor uses to provide its services.

Demos

• Software Engineer features
  • Cursor AI Desktop
    • Tab model
    • Modes (Ask, Agent & Plan)
    • Cursor rules
• DevOps pipelines features
  • Cursor Cloud Agents API

Demos

Software Engineer features

Tab model

Tab completion in Cursor is powered by advanced AI models that understand your code context. As you type, you'll receive intelligent suggestions that:

• Complete your current line of code
• Suggest entire function implementations
• Help with common patterns and boilerplate
• Adapt to your coding style over time

Tab model

Demo:

Implement the interface Cache in the class CacheImpl.java

                            
                                public interface Cache {
                                    void put(String key, Object value);
                                    Optional< Object> get(String key);
                                }
                            
                        

Modes

Natural language interface for exploring, editing, and managing code with contextual AI assistance in Chat mode.

Ask Mode

• Ask questions about specific code sections
• Get explanations of complex functions
• Find code patterns and examples
• Discover and understand your codebase

Source: https://docs.cursor.com/get-started/introduction

Ask Mode

2000's	2010's	2024+
Books	Google, StackOverflow, Blogs, etc...	ChatGPT, Claude, Gemini...

Agent Mode

• Make codebase-wide changes and refactoring
• Implement new features from requirements
• Debug complex issues across multiple files
• Generate tests and documentation
• Maintain consistency across your entire project

Source: https://docs.cursor.com/get-started/introduction

Agent Mode

Example to fix a test:

Plan Mode

When you need to implement a new feature with a higher level of complexity, you can use the Plan Mode to break down the task into smaller steps.

Cursor rules

Using rules in Cursor you can control the behavior of the underlying model. You can think of it as instructions and/or a system prompt for LLMs.

Cursor rules

AS IS:

User Prompt (Request) > Result from Model (Response)

TO BE:

User Prompt (Request) + System prompts > Result from Model (Response) <- Better results

Cursor rules

Where I can find specific cursor rules for Java? Example:

• https://github.com/jabrena/cursor-rules-java

How to apply them in my repository? $(pwd).cursor/rules

                    		
                    sdk install jbang

                    jbang cache clear
                    jbang catalog list jabrena

                    jbang setup@jabrena init \
                    --cursor https://github.com/jabrena/cursor-rules-java
                            
                        

Demos

DevOps pipelines features

Cursor Cloud Agents API

With Cursor Cloud Agents API, you can delegate tasks in the cloud, not in your local machine in order to automate some tasks in your repository and submit a PR at the end of the process.

Cursor Cloud Agents API

Cursor Cloud Agents API, provides a RESTful API organized in 3 areas: Agent Management, Agent Information & General Endpoints.

Cursor Cloud Agents API

Cursor Cloud Agents API

Demo:

                                                                    
                                                                In the Maven module `sandbox`
                                                                in the package info.jab.examples
                                                                Develop a classic Java class HelloWorld.java program
                                                                that print "Hello World" in the console only
                                                                and not create any test class for this example
                                                                verify the changes with `./mvnw clean compile -pl sandbox`
                                                                    
                                                                

Cursor Cloud Agents API

Cursor Cloud Agents API

Let`s automate the workflow execution a bit...

Cursor Cloud Agents API

Tasks like Churros!

Cursor Cloud Agents API

PML Workflow examples:


Source: https://github.com/jabrena/pml

Cursor Cloud Agents API

PML examples:

Source: https://github.com/jabrena/pml

Cursor Cloud Agents API

• https://github.com/jabrena/churrera

                                                
                                                    sdk install jbang

                                                    jbang cache clear
                                                    jbang catalog list jabrena

                                                    jbang churrera@jabrena
                                                
                                            

Cursor Cloud Agents API

Takeaways

• Practice with Tab Model, Ask Mode, Agent Mode & Plan Mode in this order to get familiar with the Cursor features.
• Educate the team to use 'Auto' model for tasks with average complexity to save budget. It is the same mindset when you share the Azure/AWS/GCP cost dashboards.
• Model Composer 1, is a Game changer in the Cursor ecosystem.

Takeaways

• AI Features as the new evolution in the SDLC. It is another element Shifting left.
  Jenkins > Sonar > Dependabot > AI Coding Agents...
• Explore Cursor Cloud Agents API to evolve traditional DevOps pipelines with AI features for repetitive tasks.
• Explore Cursor Rules to develop your Corporate guidelines for Java. One starting point could be https://github.com/jabrena/cursor-rules-java.

Takeaways

• A key aspect to be considered to adopt Cursor AI/others is the Dev Quality culture of the company.
• Using this kind of solutions in an unattended way could increase Technical Debt.
• Models solve problems in many cases in a Suboptimal way and it doesn't understand your team coding standards.

Takeaways

• Consider to review the code generated by models to mitigate issues related to Cyclomatic Complexity, OOP Design issues like (Extra constructors, Package design, Simplicity), Test overlapping, ...
• AI Agents doesn't develop libraries, use them to develop features.
• AI Agents doesn't have creative thinking, use them to develop features.

Takeaways

• Models are an amazing integrators of different technologies, soon one key aspect to be considered is what is cheaper to integrate with "AI Glue" or implemement with a traditional coding approach.
• Remember that you could go faster with your AI Agents, they are awesome, but size in the right way the amount of resources to operate them. (Cognitive Load issue) Not repeat the same issue with Microservices or Platform engineering teams.

Slides

Slides

Feedback

Slides

                                
                                    jbang trust list
                                    jbang cache clear
                                    jbang catalog list jabrena

                                    jbang qr-code@jabrena \
                                        --url https://jabrena.github.io/101-cursor/
                                    jbang qr-code@jabrena \
                                        --url https://forms.gle/11cXEpWWg5ZLf3DZA