Learn how total novice engineers -- even some that can’t program at all -- implement "vibe coding" with AI to ship projects at warp speed and build amazing things with ease.
1. where are memory and context7 MCP? i couldnt find it
2. on your cursor rules, is it general rules you shared? do you have project rules u can share?
3. do you have method to refactor or review your code? I heard coderabbit is good for review
4. do you have tips when trying to replicate UI by uploading UI screenshot generated by Loveable to be implemented by Cursor? The problem sometimes, when I ask to blend the UI to my code, it still ignoring it and end up keeping the screenshot UI and implement it on my code. Then I have to work around to remove unnecessary UI I dont want.
2. Those cursor rules I shared were messy BUT serve as a decent enough outline in how to prompt with specificity in rules. I don’t have a fixed rules document and I make them for projects as I go. Project specific ones include the outline of that one plus context about the stack, database, integrations, etc.
3. i’m lazy, so i make a remix of my original lovable scaffold, make a new github repository out of it, push my cursor changes there, enter chat mode with Lovable asking it to review my codebase and refactor (sometimes it has issues with new dependencies but not often — be aware though).
4. Upload your UI picture to ChatGPT first, ask for an addition to your prompt about design, then include the UI in lovable with the prompt.
Interesting article. What kind of software security work does vide coding do? Doest it check code against OWASP Top Ten or anything like that? How do you verify the generated code is free from vulnerabilities?
Hey! So, this entirely depends on the rules and prompts. Generally speaking, it attempts to use “best practices” for your predefined stack or the stack it builds. You can check your code for vulnerabilities with a few tools I’ll be talking about in some up and coming articles, and I could also make an MCP server for checking code against OWASP (if I can’t simply prompt it) for the community!
So far though, I’ve built a few SaaS model applications using the Lovable stack with my methodology above and haven’t heard any bad from my pentesters yet.
Sure! I’d be happy to write an article on it. Any MCP server or implementation I make for Cursor will be open source and get an article on it. Thanks for the idea for OWASP addition — I’ll credit you in the article and on the GitHub repository!
Hah! You’re right. Unfortunately, when you get good at programming, you become one of those crotchety MFs yourself. But I always try to treat people with kindness on SO!
damn this is awesome!
questions:
1. where are memory and context7 MCP? i couldnt find it
2. on your cursor rules, is it general rules you shared? do you have project rules u can share?
3. do you have method to refactor or review your code? I heard coderabbit is good for review
4. do you have tips when trying to replicate UI by uploading UI screenshot generated by Loveable to be implemented by Cursor? The problem sometimes, when I ask to blend the UI to my code, it still ignoring it and end up keeping the screenshot UI and implement it on my code. Then I have to work around to remove unnecessary UI I dont want.
Thanks!
Love to see another series of this!
Thank you brother!
1. check MCP.so
2. Those cursor rules I shared were messy BUT serve as a decent enough outline in how to prompt with specificity in rules. I don’t have a fixed rules document and I make them for projects as I go. Project specific ones include the outline of that one plus context about the stack, database, integrations, etc.
3. i’m lazy, so i make a remix of my original lovable scaffold, make a new github repository out of it, push my cursor changes there, enter chat mode with Lovable asking it to review my codebase and refactor (sometimes it has issues with new dependencies but not often — be aware though).
4. Upload your UI picture to ChatGPT first, ask for an addition to your prompt about design, then include the UI in lovable with the prompt.
No worries! More on the way for you!
Interesting article. What kind of software security work does vide coding do? Doest it check code against OWASP Top Ten or anything like that? How do you verify the generated code is free from vulnerabilities?
Hey! So, this entirely depends on the rules and prompts. Generally speaking, it attempts to use “best practices” for your predefined stack or the stack it builds. You can check your code for vulnerabilities with a few tools I’ll be talking about in some up and coming articles, and I could also make an MCP server for checking code against OWASP (if I can’t simply prompt it) for the community!
So far though, I’ve built a few SaaS model applications using the Lovable stack with my methodology above and haven’t heard any bad from my pentesters yet.
Great question!
I would love to see how you integrate OWASP Top Ten and similar frameworks into your pipeline!
Sure! I’d be happy to write an article on it. Any MCP server or implementation I make for Cursor will be open source and get an article on it. Thanks for the idea for OWASP addition — I’ll credit you in the article and on the GitHub repository!
Thank you!
Thanks for sharing. Gotta be careful asking questions on StackOverflow. That’s one tough crowd.
Hah! You’re right. Unfortunately, when you get good at programming, you become one of those crotchety MFs yourself. But I always try to treat people with kindness on SO!
Love you all! Hope you enjoyed this article.
can’t wait to build at warp speed