Hot take: AI coding assistants are Python developers cosplaying as polyglots.
They’re fluent in the languages with the most training data and the loosest rules and it shows the second you hand them Go, Rust, or anything that enforces correctness at compile time.
Agree or fight me
What’s the worst thing a ‘passing’ CI has ever let through for you?
Be honest does your current tooling check whether findings can actually be *chained*, or just list them?
Founders building developer tools:
What's something your users taught you that completely changed how you thought about your product?
Building ZenVeil taught me something I wasn't expecting.
I assumed the hardest part would be finding security issues.
It wasn't.
The hard part was helping developers understand what the issue actually means and what they should do next.
I kept making the scanner smarter.
Users kept asking for simpler explanations.
That was a humbling lesson.
Curious what assumption your users completely broke.
Not the polished founder story.
The painful one.
Founders building developer tools:
What's something your users taught you that completely changed how you thought about your product?
Building ZenVeil taught me something I wasn't expecting.
I assumed the hardest part would be finding security issues.
It wasn't.
The hard part was helping developers understand what the issue actually means and what they should do next.
I kept making the scanner smarter.
Users kept asking for simpler explanations.
That was a humbling lesson.
Curious what assumption your users completely broke.
Not the polished founder story.
The painful one.
I'm a software engineer currently building a developer security tool called ZenVeil (https://zenveil.dev).
One thing I've noticed over the last year is that AI coding assistants are changing the shape of security issues that make it into repositories.
The issues themselves aren't always new, but the volume and speed of code generation seem to make things like:
• hardcoded secrets
• insecure authentication patterns
• missing security headers
• dependency risks
much easier to miss during normal reviews.
For those working in DevSecOps:
Have AI coding tools changed your security review process at all?
Are there specific classes of findings you're seeing more frequently now than before?
Genuinely interested in hearing how other teams are adapting.
Question for developers using AI coding tools:
What's the most common security mistake you've seen generated by AI?
I've seen everything from exposed secrets to weak authentication patterns while working on a developer security product.
Curious whether others are seeing similar patterns or completely different ones.
Building ZenVeil taught me something interesting:
Security tools have become incredibly good at finding problems.
Developers still struggle with understanding and fixing them quickly.
If you're building developer tools:
How do you balance power vs simplicity?
I keep finding that the simpler the experience becomes, the more people actually use the product.
Question for developers using AI coding tools:
What's the most common security mistake you've seen generated by AI?
I've seen everything from exposed secrets to weak authentication patterns while working on a developer security product.
Curious whether others are seeing similar patterns or completely different ones.