"Guys, I’m under attack"
I came across this post where a founder shared how his SaaS got exploited right after he started sharing how he built his SaaS using Cursor.
I came across this post where a founder shared how his SaaS got exploited right after he started sharing how he built his SaaS using Cursor.
As he shared, some weird people: ❌ Maxed out his API usage ❌ Bypassed subscriptions ❌ Even messed with his database.
The scariest part? He admitted he wasn’t technical, so fixing it took much longer than expected.
If you are vibe coding, I’m sure you don’t want to end up like him.
A couple of basic practices can save you from a lot of pain later:
—
- Do not hardcode your API keys in code
If your code gets exposed, so do your keys. Always use environment variables instead.
For example, if you see something like this in your code: const apiKey = “sk-…”;
You should prompt the AI to: “Move the API key to a .env file and load it as an environment variable.”
Your new code should look like this after that: const apiKey = process.env.API_KEY;
And your .env file will have: API_KEY=“sk-…”
If you have been working on your code for a while, ask the AI to: “Scan this project and move any hardcoded passwords, API keys or secrets in the code to .env and load them from environment variables.”
If the AI detects any such code, it likely means you have already committed that code to your Git repo.
In this case, you should invalidate all existing keys and replace them with new ones.
As a future-proofing step, you should instruct the AI to add a rule to your .cursor/rules (or CLAUDE.md or GEMINI.md): “Do not hardcode passwords, API, and secret keys in code. Do not provide default passwords, API, and secret keys in code. Always load from environment variables.”
A caveat: AI doesn’t guarantee it will follow your rules. They treat them as suggestions. This only lowers the risk. You should still review the code in the future.
—
- Do not commit your .env file to your Git repo
Your .env file contains sensitive secrets. If it’s pushed to GitHub, attackers can gain access if your code leaks . Use .gitignore to keep it out of version control.
Your .gitignore file should have an entry like: .env*
This instructs Git not to commit any files start with .env.
Important: You should back up your .env file elsewhere.
—
Disclaimer:
These are basic but important security tips, intended to be beginner-friendly. They won’t help you achieve top-notch security, but they will prevent you from making the most basic mistakes.
More advanced techniques exist, like storing secret keys in a third-party vault or secret manager.
Security isn’t just for “big companies.” Even small SaaS projects or side hustles can be targets once they’re public.
Build fast, but build securely.
What’re other security mistakes you’ve seen or made that others can learn from?
—
I share practical tips about business, marketing and AI 🔔 Follow me to learn more! ♻️ Re-post this to help others! 🔖 Save this for future reference! 💬 DM me for collaboration!
#VibeCoding #security #Cursor #ClaudeCode #GeminiCli
Enjoyed this? Subscribe for more.
Practical insights on AI, growth, and independent learning. No spam.
More in AI Security
**GenAI Pitfalls**
ChatGPT has recently encountered various outages. These outages took down our AI services and disrupted business operations, both for us and our clients.
5 Warning Signs Your Marketing Hire Is Just an Operator, Not a Strategist
Many founders make the mistake of hiring marketers who can execute… but not strategize.
Want to monitor your context window, session, and weekly limit in real time?
One of the lesser-known Claude Code features: you can customize your status line.
Claude Code is for software developers, and OpenClaw is more for business users.
A learner said this to another learner during a recent workshop. I think this is the most common and most dangerous misconception about these two tools.
Are AI labs trying too hard to anthropomorphise AI to keep the illusion of AGI going?
--
My two cents...
I think we are overestimating the short-term impact of AI but underestimating the long-term impact again like what we did with other transformative technolog...
**GenAI Pitfalls**
ChatGPT has recently encountered various outages. These outages took down our AI services and disrupted business operations, both for us and our clients.
Want to monitor your context window, session, and weekly limit in real time?
One of the lesser-known Claude Code features: you can customize your status line.
Are AI labs trying too hard to anthropomorphise AI to keep the illusion of AGI going?
--
5 Warning Signs Your Marketing Hire Is Just an Operator, Not a Strategist
Many founders make the mistake of hiring marketers who can execute… but not strategize.
Claude Code is for software developers, and OpenClaw is more for business users.
A learner said this to another learner during a recent workshop. I think this is the most common and most dangerous misconception about these two tools.
My two cents...
I think we are overestimating the short-term impact of AI but underestimating the long-term impact again like what we did with other transformative technolog...