Vibe coding got you to MVP and your first customers. Congrats!
Now you have to:
Onboard a dev who's never seen the code
Debug with no observability
Refactor without reading every file
Add a feature without breaking 3 more
That's the gap between "it works" and "it's engineered"
I hear a lot of people say that "the build is done".
I don't know that the build is ever really done. Supporting software is a full time job. Vulnerabilities pop up, versions are deprecated, memories fill, etc.
Stale code left alone tends to break.
I'm looking to connect with more folks starting out their business and relying heavily on AI.
If you're interested in -
- Taking AI prototypes to production
- Building with Claude, Cursor, Lovable, Replit, v0, etc.
- Vibe coding without creating a mess
- Code audits for AI-built apps
- Security, auth, and data safety
- Spec-driven development
- CI/CD, testing, and deployments
- Understanding what your AI-generated code actually does
Let's connect! And if you're working on something, drop your link below. I'll check out every one of them.
Ever wonder if your code is actually safe? Want a free audit of your code base? Comment below!
I’ll pick a few and send free private audit on 9 different areas analyzing every best-practice I know.
No public roasting. Export it straight to your coding assistant for remediation.
@djgelner Yeah I've experienced that with some of the out of the box tools. Did you get it exported to GitHub?
Check out my post on https://t.co/nL2JIyHH7P.
https://t.co/rDqw68ZkZu
I audited https://t.co/hRRsLcnhl8.
Really cool project. Also a good example of why “it runs locally” and “it is safe to operate” are very different things.
The audit found:
unauthenticated API key export
unauthenticated MCP config that can execute OS commands
integration tokens exposed to browser JS
almost no tests around the LLM pipeline
near-zero production observability
Final score: 3.5/10
This is not a roast. It is just what happens when a fast-moving AI codebase starts crossing the line from prototype to something people may actually deploy.
Full audit here:
https://t.co/qhPoPPPETE
Want me to audit your repo?
Comment below!
I’ll pick a few and send free private audit on 9 different areas analyzing every best-practice I know.
No public roasting. Just useful feedback.
I audited https://t.co/hRRsLcnhl8.
Really cool project. Also a good example of why “it runs locally” and “it is safe to operate” are very different things.
The audit found:
unauthenticated API key export
unauthenticated MCP config that can execute OS commands
integration tokens exposed to browser JS
almost no tests around the LLM pipeline
near-zero production observability
Final score: 3.5/10
This is not a roast. It is just what happens when a fast-moving AI codebase starts crossing the line from prototype to something people may actually deploy.
Full audit here:
https://t.co/qhPoPPPETE